JPRS ID: 10309 USSR REPORT CYBERNETICS, COMPUTERS AND AUTOMATION TECHNOLOGY

Document Type: 
Collection: 
Document Number (FOIA) /ESDN (CREST): 
CIA-RDP82-00850R000500030018-4
Release Decision: 
RIF
Original Classification: 
U
Document Page Count: 
167
Document Creation Date: 
November 1, 2016
Sequence Number: 
18
Case Number: 
Content Type: 
REPORTS
File: 
AttachmentSize
PDF icon CIA-RDP82-00850R000500030018-4.pdf10.04 MB
Body: 
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY JPRS L/ 10309 - 8 February 1982 _ USSR Re ort p C~3ERNETICS, C4MPUTERS AND AUTOMATION TECHNOLOGY CFOUO 2/82) Fg~$ FOREIGN ~ROADCAST INFORMATION SERVICE FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 NOTE JPRS publications contain information primarily from foreign newspapers, periodicals and books, but also from news agency transmissions and broadcasts. Materials from foreign-language _ sources are translated; those from Engiish-language sources ~re transcribed or reprinted, with the original phrasing and other characteristics retained. Headlines, editorial reports, and material enclosed in brackets are supplied by JPRS. Processing indicators such as [Text) or [Excerpt] in the first line of each item, or following the last line of a brief, indicate how the original information was processed. Where no processing indicator is given, the infor- mation was summarized or extracted. , Unfamiliar names rendered phonetically or transliterated are enclosed in parentheses. Words or names preceded by a ques- tion mark and enclosed in parentheses were not clear in the original but have been supplied as appropriate in context. Other unattributed parenthetical notes with in the body of an item originate with the source. Times within items are as ' given by source. The contents of this publication in no way represent the poli- - cies, views or at.titudes of the U.S. Government. COPYRIGHT LAWS AND REGULATIONS GOVERNING OWNEk~HIP OF MATERIALS REPRODUCED HEREIN REQL'IRE THAT DISSEMINATION OF THIS PUBLICATION BE RESTRICTED FOR OFFICIAL USE ONLY. APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIA7. USE ONLY - JPRS L/10309 8 February I982 USSR REPORT Ci(BERNETICS, COMPUTERS AND AUTOMATION TEC~iN(?L4GY (FOUO 2/s2) CONTENTS GENERAL Variation of Feeding Video Information to YeS Computers........... 1 Synthesis of Parallel Microprogramming Structures 5 Verification of Model Configuratians for Pulsed Itadio - Engineering Systems 8 Efficient Servicing of Computer Equipment Reviewed 10 PXOblems of Automated Control Systems for Industrial Proceases Reviewed 18 HYBRID COMPUTERS Hybrid Computi,ng Machines and Systems: Local Automated Control Systems and Computer Devices 28 Algorithm for Analyzing Patching Scheme of Operational Modules Usec'~ To Simulate Automatic Programming Systems for Analog....... 37 Methads for Organizing Diagnosis of Special-Purpose Processors and Devices 43 Synthesis of Optimal Digital-Analog Regulator for Controlling Thermal Ob~ect 50 - OPTICAL PROCESSING Holography and Optical Processing of Information: Methods and Apparatus 57 - a- [III - USSR - 21.C S&T FOUO] Fl1R (1FF~!'i ~ T T!CF nNT.Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY s Recording and Processing of Modulated Optical Signals 60 Holographic Visualization of Underground Ob~ects 65 Processing Seismic Information With Coherent Optical System.,..... 69 Holographic Method of Checking Reflectors 74 . Beam Holography 76 SOFfiWARE Abstracts of Articles in Journal 'PROGRAMMING', September- O~tober 1981 79 Software Implementation of Multiprocessing 82 ImplE:mentation of 'FOREKS' Compiler for A.S-6 Central Processor.... 90 " System of Dialogue Preparation of Tasks for Unified Series Computers 102 Organization in Dispak Operating System of Determinate Output of Znformation Over Entire Field of Output Units of ; , Multimachine Computing Complex 110 ` PUBLICATIONS Control Algorithms for Spacecraft 116 Machine Mathematical Modeling....o 127 Autnmation of Exploratory Design (Artificial Intellioence in - Machine Design) 138 Data Processing Eqtiipment 144 Homo~eneous Computer Systems, Structures and Devices 147 J - b - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY GENERAI~ � UDC 621.391 VARIATION OF FEEDING VIDEO INFORMATION TO YES COMPUTERS Kiev AVTOMATIKA in Russian No 5, Sep--Oct 81 (manuscript received 19 Mar 81) pp 76-78 [Article by V. Ye. Reutskiy] [Text] There are many problems involving processing video informati:on in vari- cus fields of science and technology. The growing volume of this information and the heightened requirements for precision and speed of solutions led to intensive development of ineans and metTiods of automating the processing of video information. Digital p,-ocessing methods using general purpose computers and specialized processors pl;iy tfie leading role in this. This leads to the ! necessity of developing new i~iput units, on the one hand, and a desire to use ~ input units already included in the computer equipment on the other. I . ~ Feeding video information to computers requires a unit to match the information ~ and physical characteristics of the television camera signals and the signals ~ used in the input/output interface of the particular computer. In this case equipment expenditures to realize the block to match the characteristics of ~ the signals of the input/output interface of the YeS [Unified System] computer are several times as much as expenditures to realize the matching unit for the I characteristics of television camera signals. ~ i ~ For this reason there will unquestionably be interest in an experiment con- ' ducted with a simple hardware connection of an industrial television camera to i a YeS computer through a low-~~ower input unit with.minimum expenditure of time ~ and resources. , The variation of feeding video information that was develope~? consists of a television camera, a matching element, and a YeS-6022 unit (the standard punched tape data input unit for YeS computers) in which.the photoreader is switched off. A VZOR television camera was used to convert visual information into an elec- trical analog signal. This is a small semiconductor camera with a power supply of 12 volts (in principle any s:andard television camera can be used). For complete autonomy the camera also has a synchronous generator (consisting of a synchronous pulse generator, a frequency divider, and control signal shapers) which controls the work of the television camera, analog/digital convertor, and control block. 1 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 - FOR OFFICIAL USE ONLY The frequency of the master clock. (genezatox) ~as selecte~ at around 936 kilohertz ~n the fiasis o~ preliminary tests of the. multiplex cfiannel of a YeS computer to which was connected a Xe~-6022 unit uzith a;~ingle field data input speed on tfie order of 260-280 kiloFiytes, based on the real speed and necessity of receiving from one generator syncfironous pulses to control tlie television camera and strobing pulses for the analog/digital convertor and control block. The possibility of completely fill�ing the byte being trans~- mitted to the YeS computer was employed to increase the speed of data input. A four-gradation system of defining the brightness of each point in tfie video information was adopted for this purpose, and tfie bit format of the analog/ digital convertor was defined in two ranks. After the information from four points is packed in one byte, the data input speed is 234 kilobytes. The precision requirements for the analog/digital convertor are fairly low (four gradations with a television camera signal on the order of 0.5 volts), but the speed requirement is quite important. Therefore, a double-rank (simplified) variation of a high-speed analog/digital convertor was used. This will make it possible in the future to increase the speed to 8-10 mega- hertz because it is determined chierly t~y the operating speed of the comparators used in the convertor. The television camera and analog/digital convertor work asynchronously witli the computer, but information is transmitted only when th~ signal of the computer processor which authorizes transmission of information from the YeS-6022 coin- cides with the beginning of transmission of a frame by the televi~cion camera. After the frame is transmitted the YeS-6022 unit goes into a waiting mode and the next frame is fed folluwi.ng a program request by the comput~r processor. The entire television camera signal matching block, the synchronous generator, . and the signal shapers are housed in a standard YeS TEZ [c2rd] which is installed in the YeS-6022. When the television came~a is connected in, it is necessary to remove the input/output signal shapers ef the photo- reader (TEZ 0071, place 5A43 and TEZ 0072, place SA44) from the unit and in- sert the newly developed TEZ in place SA43. The necessary connection between place 5A43 and the free terminals is inatalled, which does no[ hinder work of the YeS-6022 with the photoreader. - According to the system adopted of classifying brightness by four levels, the - frequency of the master clock is four timea the frequency of data transmission to the computer, and the synchronous generator issues control signals t~~at are rigidly interrelated by multiples: strofie 1- strobing signal of the compara- tors of the analog/digital convertor 936 kilohertz; strobe 2- strobing signal of the assembly for packaging in a byte - 936 kilohertz; strobP 3- signal for transmission of a byte of information to tfie YeS-6022 - 235 kilo- hertz; synchronous signal of the line frequency for the television camera 16,625 hertz; synchronous signal of the frame frequency for the television camera - 50 hertz. The figure below shows the time scheme of the work of the synchronous gener- ator. It is built with integrated microcircuits and housed on the same board with the other elements of the matching TEZ. 2 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFIC[AL USE 4NLY , _ _ - _ _ _ _ ~m-~zoN f p .1 3 6 /780 !7C! !)OZ I7D3 (Strobe 1~'~~ � 100-,~:70y (Strobe 21mo~p 900-f00,v (Strobe 3) j ~ i 31j C!/ 3 I ~ /HC I ~a) ~yc~;~ am EC-6011 ?0 ~ I b /biv~rueNUe ~ ~ NQ ~0.�Ql�BQYS/ ?0 MC UN~MQ!(f/U i ' I Time. Diagram of tfie GIork. of the Matching Unit Key: (a) Launch Signal from YeS-6022; (b) Authorization To Transmit Information; ' (c) 3 Microseconds; ~ (d) 1 Milliseconds j"MC" = Millisecond]. I Because the computer and the television camera ~zork asynchronously, an additional ~ element was put in the control scheme to authorize transmission of data from the ~ television camera. It authorizes transmission only from the initial moment of i ~ image scanning bS the television camera. 'fhe end of the transmisston of a frame i is determined by program means (by the number of transmitted bytes of t~ie array ~ of data). I ! As noted above, the speed of information input is limited by the actual speed of the ur.it being used and the channel of the YeS c~mputer. Therefore, with the I above,indicated frequency of data transmission to the YeS computex (234 kilo- ~ bytes), the analog/digital computer strobes and converts to digital form 60 four.- gradation points or (after packaging) 15 bytea of information per 4ne televi- - sion line where the total number of lines in a frame is 313. Thus, one tele- visionframe contains 4,695 bytes or 18,784-gradation points ir. digital form. With a standard line length of 64 microseconds the distance between adjacent strobing points is about 1 microsecond in time. Therefore., all points tfiat - occupy less than 1 microsecond during line strot~ing may be lost. Thts must be taken into accountwhen feeding data from the television camera. The image fed to the computer may be printed out hy program means for convenient observation and checking, al?:hough this does involve difficulties with observing 3 FOR OFFICIAL USE ONLY , APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE aNLY the televtsion format during out~put. In thQ garticular case every fourth_line of the television frame ~ras outputted to a digital pr~ter; in this case 78 lines with 60 points to a line were printed out and separated by gaps to ap- - proximate tTie i;levision format. COPYRIGHT: Izdatel~stvo "Nauknva dumka", ~~Avtomatika", 1981 11,1?6 CSO: 1863/48 ~ 4 FOR JFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY UDC 681.322,01 SYNTHESIS OF PARA~.LEL MICROPROGTcAMMING STRUCTURES Kiev KIBERNETIKA in Russian No 5, Sep-Oct 81 (manuscript received 23 Aug 77) pp 48-54 [Article by O1'ga Leonidovna Bandman, Sergey Vladimirovich Piskunov ~nd Stanislav Nikolayevich Sergeyev] _ [Excerpts] The organization of effective computations in uniforta structures [1-3] is connected with providing structural and functi~nal correspondence between the algorittun ar.3 its performing unit. Th3s correspondence is usu~lly achieved by means of a selection of cell functions and their connections and ~ constitutes a problem in syntt~esizing uniform structures. The solution of such i a problem for a class of devices with a multiple flow of commands and a multiple i flow of data is proposed in the article. The suggested synt~esis method is ' based on a microprogrammed presentation of algorithms which allows interpreta- j tion to the network from automatic units. The microprogramming concept, expanded ~ for organization of the comnutational process in large groups of low-power i computers, is called parallel microprogramming. ; A cellular mass, which is a terminal combination of named cells into which ~ symbols are inserted from some terminal alphabet of conditions, serves as the ~ conversion ob~ective in parallel microprogramining. A substitution micro- i command havind a left and a right component is the primary conversion of the cellular mass. Execution of the microcommand is conducted aimultaneously over all cells of the mass. A mass of cells is computed for each cell in accoxdance with the left component of the microcommand. If the obtained masa is included in the processed cellular mass, some of its component is replaced by the cellular mass corres ondin to the ri ht com onent. P 8 8 P The combination of microcommands recorded in randota order is called micro- programming. All microprogramming ::~icrocommands are executed simultaneously over all cells of the mass. Microproqramming construction is based on algorithms of parallel substitutions (4, SJ which are based on the concepts , of a cellular automatic unit [6] and of a normal algorithm [7]. 5 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R040500030018-4 FOR OF FiC1AL USE ONLY Any methods of synthesis can be used for obtaining a logical structure of the network's primary automatic unit. For example, each primary automatic ur.it becomes microprogrammed, containing within itself the entire microprogram. A block diagram of the primary automatic unit is depicted.in illustration S.a. Equipment performance is possible when using special microprogrammed large integrated circuits (BIS), constant memory units (PZU), associated memory units (ZU) and programmed logic matrices. Replacing the microprogram provides the possibility of ad~usting the network to that or another algorithm. Networks of this type are called microprogrammed cellular structures. It is obvious that the parallel microprograms performed in them are limited by the memory capacity of the primary automatic units, and therefore such an approach is acceptable for performing small (according to the number of microcommands) algorithms. The second approach consisting of a microprogram memory which is repeated in each automatic unit is carried out within the limits of the network (illustira- tion S.b.). The remaining network is called the data memory. The primary automatic units in it are simple enough: they contain a memory for storing the alphabet symbols, a comparison unit and a multiplexer [4]. rhe microcommands can be fed into the data memory with a different degree of parallel in time: all simultaneously, successively by micro~ommands and even successively by sycnbols. However, a marginal parallel in space takes place in all cases. Performances of this category are called homogeneous computers (4]. It is obvious that computers with limited parallel both in time and in S~fi~(d have the greatest output. A structure is proposed in reference I2 in which during performance of the primary automatic units, a combination of frequent and spatial coding of microcommand symbols and the use of the construction principles of multi-stable elements provide a simultaneous feeding of hundreds of microcommands to each automatic unit fo the data memory. At the same time, homogeneous computers with a consistent delivery of microcommands to the data - memory show great pr~ctical interest, inasmuch as their performance is possible from existing microprocessing sets [13]. Two approaches can be used in composing algorithms for equipment interpretatir,n. In the first case, a control unit is proposed which atores the microprogram FS and delivers control signals to the microprogram structures performing F~1,,....., F'q [14]. In the second case, the memory is divided into units corresponding to the data for F'1,...,Fq, and the microprogram memory contains the microprogram comprising F. Bibliography 1. Evreinov, E. V. ~nd Prangishvili, I. V., "Digital Automatic Control Units With an Ad~usted Structure," Moscow, Energiya, 1974, 239 pages. 2. Bandman, 0. L., Evreinov, E. V., Korneyev, V. V. and Khoroshevskiy, V. G., - "Homogeneous Computer Systems cn the Basis of Microprocessing BIS [Large Integrated Circuits], in the collection "Problems of Theory and Construc- tion of Computer Systems," No 70, 1977, pp 3-28. 6 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFF[CIAL USE ONLY 3. "Homogeneous Microelectronic Associated Processors, edited by I. V. Prangishvili, Moscow, Sovetskoye radio, 1973, 280 pages. 4. Kornev, Yu. N., Piskunov, S. V. and Sergeyev, S. N., "Algorithms of Generalized Substitutians and Theix Interpretation by Networks of Automatic Devices and HomogQneous Computers," IZVESTIYA AKADEMII NAUK SSSR: TEKHNICHES?:AYA KIBERNETIRA, No 6, 197Y, pp 131-142. 5. Kornev, Yu. N., Piskunov, S. V. and Sergeyev, S. N., "Problems of Construct- ing Algorithms of Generalizeii Substitutions in a Separate Context," VYCHISLITEL'NXVE SISTEMY, No 47, 1971, pp 117-119. 6. Yamada, H. and Amoroso, S., "Tesselation Automata," INFORM. AND CONTROL., Vol 14, No 3, 1969, pp 299-317. 7. Markov, A. A., "Theory of Algorithms," TRUDY MATEM. INSTITUTA AKADEMII NAUK SSSR, No 42, 1954, p 375. - 8. Peterson, I. L., "Petri Nets," ACM COMP~JT. SURVEYS, Vo~ 9, No 3, 1977, pp 233-252. 9. Lipton, R. I., Snyder, L. and 7.alzstein, Y. A., "Co~parative Study of Models of Parallel Computation," 15th Annual Sisnp. Switching and Autom., 1974, pp 145-155. ; 10. Baranov, S. I., "Synthesis of Microprogrammed Automatic Units," i Leningrad, Energiya, 1974, 216 pages. I ; 11. Anishev, P. A., "Concerning the Determinability of Parallel Chart- ! Diagrams," in the collection "Problems of Theory and Construction of I Computer Systems," No 75, 1978, pp 40-52. j 12. Author Certificate 6fa4168 (USSR)s "Computational Hc~mogeneous Structure," j Kornev, Yu. N. and Pi3kunov, S. V., Published in B.I., No 1.9, 1979. Sergeyev, S. N., "Performance of Algorithms of Parallel Substitutions 13. ~ in Microprocessing Systems," in the collection "Problems of Theory and ; Construction of Computer Systems," No 73, 1978, pp 25-39. 14. Bandman, 0. L., "Synthesis of Asynchronous Microprogrammed ContYOl by ; Computer Processes," KIBERNETIKA, No 1, 1980, pp 42-47. COPYRIGHT: IZDATEL'STVO "NAUKOVA DUNIICt1", "KIBERNETIKA", 1981 9889 CSO: 1863/52 7 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICiAi. USE ONLY UDC 518.9 VERIFICATION OF MODEL CONFIGURATIONS FOR PULSED RADIO ENGINEER~NG SYSTEMS Kiev KIBERNETIKA in Russian No 5, Sep-Oct 81 (manuscript received 3 Jun 80) pp 40-47 [Article by Yuriy Anatol'yevich Belov, Vladimir Leonidovich Maka.rov, Viktor Gennadiyevich Shelepov and Vladimir Borisovich Shul'zhenk.o] [Exc~erpts] When examining models like operators transforming some given input data into definite output data, the question always arises whether the process of handling the flow of incoming infortnation corresponds to that function which the model muat perform. In other words, how to be convinced that the correct reaults are being received in any of the proceases which can be simulated in accordance with the model data. The question of adequacy of the mathematical models is one of the basic, . decisive and inevir.able questions of madeling. Considering the complexity of checking such compliance for models of actual or pro~ected radio engineering systems (RTS), it ia possible to formulate a _ more simple task~-to check whether the model (operator) is presenting data from the field of permissible incoming valu~s to data of the f ield of pei'mis- sible output values. A number of well-founded methods of ch~c~ing program accuracy was w~orked out for solving the immediate problem in programming: Floyd's inductive supposition method [1], the hierarchical structuralization method j2], Hoare's tnethod [3] and others. The purpose of present work is to develop an inductive supposition method of checking program accuracy for a new and important, in a practical respect, class of mathematical ob~ectives--model configurations. It should be mentioned that stricC proof of accuracy in constructing a mathe- matical model and the corresponding simulation algorithm according to an assigned physical description (structural configuration) of the IIS [Information measuring System] is an important procedure which allowa, generally speaking, 8 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY one to ascertain the inadequacy of thE physical model to the actual process and, theref ore, to localize operacions for correcting the model on the lev~l of physical presentations alone. ~ Bibliography 1. Manna, C., '�?�tathematical Theory of Computation," New York, McGraw-Hill Book Co., 1974, 226 pages. 2. Robinson, L. and Levitt, r.., "Proof Techniques for Hierarchically Structured Programs," ~LTDitENT TRENDS IN PROGRAMMING METHODOLOGY, No 11, 1979, pp 57-73. 3. Hoare, C. A. R., "An Axiomatic Basis of Computer Programming," CAC~I, Vol 12, No 10, 1969, pp 112-135. 4. Kotov, V. E., "Introduction to the Theory of Program Configurations," Moscow, Nauka, 1978, 257 pages. - 5. Leonov, A. I. and Fomichev, K. I., "Monori~alse Radar," Moscow, Sovetskoye radio, 1970, 392 pages. 6. Leonov, A. I., Vasenev, V. N. and Gaydukov, Yu. I., "Radar Simulation," Moscow, Sovetskoye radio, 1979, 264 pages. 7, Kuz'min, S. Z., "Digital Processing of Radar In�ormation," M^ecow, Sovetskoye radio, 1967, 399 pages. 8. King, Dzh., "Checking Compiler," in the book "Aids To Managing Large Systems," Moscow, 1977, pp 23-41. 9. Anisimov, A. V., Belov, Yu. A., Lyashko, I. I. and Makarov, V. L., "The Adequancy of Mathematical Modeling of a Complex Information Measuring System," DOKLADY AKADEMII NAUK SSSR, Vol 240, No 2, 1978, pp 202--206. 10. Belov, Yu. A., Makarov, V. L., Shelepov, V. G. and Shul'zhenko, V. B., "One Approach to Checking Adequacy of the Block Diagram of a Functional Al~orithm for the Structural Configuration of a Pulsed Information ~ Measuring System," DOKLADY AKADEMII NAUK SSSR, Vol 255, No 1, 1980, pp 36-40. COPYRIGHT: IZDATEL'STVO "NAUKOVA DiJMKA", "KIBERNETIKA", 1981 9889 CSO: 1863/52 9 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY EFFICIENT SERVICING OF COMPUTER EQUIPMENT REVIEWED Moscow TEKHNICHESKIYE SREDSTVA OBRABOTKI INFORMATSII in Russian 1981 (_signed r4 press 17 Feb 81) pp 307-318 [Selections from chapter 9 of book "Data Processir~g Equipment", by Vasiliy Nikolayevlch Kriushin, Ni'tolay Matveyevtcfi Surtn, Valeriy Pavlovich Chuprikov and Nina Grigor'yevna C'ner~:yal~:, Izdatel'stvo "Finansy i statistika", _ 12,000 copies, 320 pages] [Excerpts] Table 9.1. Technical Servicing Norms for Computer Keypunch, Electro- me::hanical, and Electronic Keyboard Machines by One Worker at Computing (Informa- tion-Comput;:ng) Centers, Information-Computing ~tations, and Machine Accounting Stations of the System of the USSR Central Statistica? A~~.ministration. Numb~r of Average Num- Machines ber of Hours . Sexviced by To Service One Wor ker One Machine ~ es and Models of Machines in a Mo~nth for a Month ~ Machines for Mathematical Processing of Dato � Recorded on Punchcards ~ The T-5M and "_'-�5MV Tabulators 5 34.9 The TA80-1 Tabulators 3 58.2 The VP-2 and VP-3 Electronic Computing Attachments 9 19.4 to ~abulators Machines for Ordering Arrays of Punchcards - The SE80-3 and SE80-3/1M Electronic Sorters 6 29.1 The S80(45)-5M and S80(45)-7 Sorters 10 17.5 . Collaters (RPM's) 7 24.9 Machines for Preparing Punchcards The PD 45-2 and PD 45-2/1M Keypunches 12 14.5 _ The P80-6, P80-6/1M, and Zoy~ntron-415ts Keypunches 9 19.4 10 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY Number of Average Num- Machines ber of Houses _ Serviced by To Service One Worker One Macfiine T es and Models of Machines in a Month for a Month Machines for Preparing Punchcards The PA80-2, PA80-2/1M, PA80-2/2M, PA80-2/3M, and Zoyemtron-415a Keypunches 6 29.1 Summary Punches of the PI80(45)-U Type 11 15.9 Reproducing Punches of the PR80(45)--U Type 9 19.4 The K80(45)-6, K80(45)-6/1M, and Zoyemtron--425ts Verifiers 10 17.5 The KA.80-2, KA80-2/1M, KA80-2/2M, KA80-2/3M, and Zoyemtron-425a Verifiers 7 24.9 The RMK80/45 and RMA-80 Decoding Machines 8 21�8 ~ Keyboard Machines The SDV-107, SDV-108, SDK-133, AYeS, AYeSVye, Askota-110, -112, -114, -117, and -314, and Other Adding Machines 30 5.8 The VMA-2, VK-2, SAR, KYeL, and.Other Electro- Mechanical Keyboard Computing Machines 25 6.9 The Elka-22, Zoyemtron-220, Elektronika, and Other Electronic Keyboard Computing Machines with Discrete Elements 30 5.8 The Iskra-108, -111, and -122, Elka-43 and Elka-50, and Other Electronic Keyboard Computing Machines with Integrated Circuits 50 3.5 The VA-345M, FM-346, FMR, FMYe, and FMYeL Invoice Machines 9 I9.4 The Zoyemtron-381 and 382, Iskra-23 and 522, and EFM-446 Electronic Invoice Machine 12 14.5 The Zoyemtron-381, -383, -384, and -385, Iskra-2302 and -534 Electronic Invoice Machines with Punch Attachments 9 19.4 The Askota-170 Bookkeeping Machines 9 19.4 The TM-20 Electronic Multiplier Attachments 27 6.5 The Askota-170/55 Bookkeeping Machines with Tape ' Punch Attachments 8 21'8 Tecfinical servicing and repair of computer equipment is done by enterprises of Soyuzschettekhnika (All-Union Production Technical Assoctation for Tecfinical Servicing and Repair of Computer Equipment of the USSR Central Stattsttcal Administration) on the basis of contracts concluded between the manufacturing enterprise and the client. 11 _ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC(AL USE ONLY Soyuzschettekhnika repair enierprises can do technical servicing, current (smalZ-scale) repair, ~edium repatr (1 and 2), and capital repair (1 and 2). With tfie approval of tfie USSR State Committee for Prices a price list for - wholesale prices for all types of repair work and tecfinical seroicing of tfie computer equipment of the USSR Central Statistical Administration was ratified and put into effect as of 1 January 1977. This price list envisioned separate wholesale prices for different types of repair, technical servicing, and other jobs (launcfiing machines into operation, storage and packaging, repair of electrical motors, repair and replacement of particular assemblies). Wliere tfiere is a contract for technical servicing, curren;. (small-scale) repa3:r, installation, and launching newly received equipment in operation are done by Soyuzschettekhnika enterprises free of charge. Let us review the planning for repair of computer equipment using the example. of organizations of the USSR Central Statistical Administration. The estab- lished rules for planning repair of camputer equipment within the system of the USSR Central Statist3:ca1 Administration envision tfiat rayon and city computing centers (and machine accounting stations) will suBmit information on the technical condition and number of hours worked by each machine in the past year and plans for the machine's workload in the current year to the com- puting centers (machine accounting stations) of the otilast statistical adminis- trations and the appropriate repair enterprises each year (based on condition as of 1 January). On the basis of these data the repair enterprise makes up a technical passport for each machine, sets up a card file, and plans .^..nd maintains records of iepair work. The rayon and city computing centers (machine accounting stations) submit a plan of machine repair for the coming year to the computing centers (machine accounting stations) of the oblast statistical administrations each year, be- fore 1 February of the current year, according to the number of hours worked by each machine. This plan indicatc~s repair enterprises by service zones using the following form: _ Form No 9.1 Plan for Repair of (:omputer Equipment in 198 (Name of Computing Center or Machinc� Accounting Station) For the Service Zone of (Name of Repair Enterprise) Machine Name Factory Number Number of Items Included in that Capital. Medium Current Repair Repair Repair 12 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY Each year the computing centers (macfiine accounting stations) of the oblast statistical administrations compile a sutmmary request (plan) in tfie same form showing their repair needs for tFie coming year. Tfiis is sulimitted to the Main Admin~stration of Computer Work (Glavmekhschet) of tFie republic central statistical administration l~efore April of the current year. Tfiis administra- tion then sends the summary request for the repu5lic central central statis- tical administration to Glavmekhscher of tfie USSR Central Statistical Ad- ministration before 1 June of the current year. By 1 July of the same year Glavmekhscfiet of the USSR Central Statistical Admintstration su~imits infor-- mation on the computer equipment reapir needs of the system of tfie USSR Central Statistical Administration to the Soyuzschettekfinika Association. Soyuzchettekhnika compiles a summary plan of computer equipment repair and brings it to the attention of the repair enterprises. At the large cumputing centers (machine accounting stations) where medium and current (small-scale) repair is done by in-house personnel, the work planning system envisions compilin~ annual plans of machine repair which list all machines subject to repair and indicate the type of repair and the time it will take for each machine. In addition to compiling the annual repair plan tiiey work out a repair schedule for each month which envisions planned withdrawal of macTiines from work positions. The schedule is usually drawn up on the basis of the volume of repair work in the current month and the calendar amount of working time of workers engaged in repair. The schedule is used for operational moni- ' toring and accounts related to calculating expenditures for repair. Detailed trouble lists are compiled before performing each type of repair Repair workshops which have the necessary amounts of furnishinos, bencfi equip- ment, tools, and spare parts are organized to do repair work at the computing _ centers (machine accounting stations). In the repair workshops the work of service personnel is usually organized on the brigade principle: eacfi brigade is assigned to a certain group of machines. In addition~ a definite worker . is responsible for repairing each machine. The Organization of Technical Servicing - Technical servicing of computers involves a set of organizational--technical measures which must be done to maintain the operating reliability of machines within required parameters. These activities include: selection of appropriate service personnel; acquisition of hardware and software for diagnosit~g mal- Functions; supplying spare parts, tools, and accessories for the machines; supplying service apparatus for testing external units; supplying special furnishings and auxiliary equipment for operating and repairing computers. Optimal organization of the technical servicing of computers is a major task in using them efficiently. There are various types of technical servicing of computers: individual, group, and centralized. With individual servicing each computer is provided with a complete set of service apparatus, spare parts, tools, accessories, and appropriate service personnel. Individual servi.ce facil.ities include the following: apparatus for 13 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFF[CIAL USE ONLY monitoring tfie ~asic elements of tfie computex and poaaex supply~; monitoring and adjusttng apparatus for autonomous checks of individual units of com- puter hardware; a set of tes.t programs; tools and repair accessories; auxiliarq equipment, attachments, and special furnisfiings for storing the property of the computing center. With group servicing several computers concentrated at a single computing - center are serviced by in-house personnel. The structure of group service is the same as for individual service, but it is supplemented for otlier computers. - Witfi centralized computer servicing the service apparatus, spare parts, tools, accessories, and service personnel are significantlp reduced, but wi.tli this - form of servicing the tim.e required to restore computers depends significantly on the operational features of tfie work of centralized servi.cing points, tn particular how far they are from the computing centers. The type of technical servicing is determined by the user depending on the place where the computer is installed and the range of jobs i.t does. The number of engineering-technical personnel needed to service computers depends on the type of technical servicing and the mode of opexation of tfie. computer. Thus, for individual servicing of one YeS-1020 computer wfiich is used in three shifts, ~he following engineering-technical staff is reco~nended: Machine Chief 1 Shift Chiefs(Senior Engineers) 3 Shift Engineers (Electrical Engineers. and 3 Electromechanical Engineers~ 2 Senior Electricians 2 ~ Electricians Precision Machinery Mechanic 12 Total Scheduled preventive maintenance work involves a set of ineasures atmed at keeping computer units in working condition and preventing breakdowns and failures during their operation. The period of scheduled preventive work is an essential and continuing stage in keeping computers in working condition. Reducing the time tfiat tfiis work takes increases the usable work time of the computer, that is, tTie time during wnich the machine is engaged in problem-solving or debugging programs. The volume of scheduled preventive maintenance work depends on the teclinical condition of the computer units and the qualifications of engineering- technical personnel. The length and periodicity of this work are determined by the manufacturing plants in the appropriate operating instructions. The essential feature of preventive work is the following: when a machine is being prepared for problem-solving th.e working cond3:tion of tlie machine itself as well as its blocks and individual elements must be.tested using 14 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY specially prepared problems or test-programs~ with kno~n answers. In case of an error the programs envision that tfis machine will fialt automatically and a description of the nature of the trouble will Fae printed out on the typewriter or a signal will appear on the control console. Tfie test program is used to check not only tfie conditxon of the computer, but also wiietfier the program to solve tTie particular specific problem has been written correctly. Experience operating contemporary computers shows that it is necessary to oli- serve each machine systematically, clean its blocks and assemblies, test tfie work of individual units, and replace wornout parts, In addttton to identifying and eliminating ordinary malfunctions, servicing computers involves daily, weekly (biweekly), monthly, and semiannual (annual) scheduled preventiVe main-- tenance. The following periodicity and length of scheduled preventive maintenance work is recommended for YeS [Unified Sys.tem] computers: Table 9.4 - Periodicity af Pre- Computer Name ventive Work Number of Hours YeS-1020 ' DaiJ_y 1 Weekly 4 Monthly 16 Annual 72 YeS-1030 Daily 1 Biweekly 4 - Monthly 8 Semiannual 72 YeS-1050 Dai.ly 1 Biweekly 4 Monthly 8 Semiannual 72 Daily preventive maintenance includes an ex~ernal Znspection of the condition of _ the machine units and power sources; testing the :nachine with monitoring tests; eliminating malfunctions when there are deviations from technical norms; check- ing, cleaning, and adjusting external units. Weekly (biweekly) preventive maintenance envisions a set of weekly (biweekly) activities: testing the reliability of inechanical fastenings; testing the ven-- tilation and power supply system; checking the work of the units with a pre- - ventive alteration of the supply voltage of �S percent. Monthly preventive work envisions cleaning and lubricating tlie assemtilies, mechanisms, and power blocks and test:Cng the functioning of tTie computer hard- ware using diagnostic tests with an alteration of �5 percent in supply sources. 15 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 _ FOR OFFICIAL USE ONLY Annual (semiannual) preventive work includes the eame jofis tfiat are done in monthly preventive work as well as dismantling, cleaning, and lub.ricating all mechanical assemblies of sxternal units w~tfi adjustment or replacement of parts at the same time. In addition, the cables and supply lines are inspected. Tlie instructions on operating individual units whi.cfi are attacfied to the machine by the manufacturing plant give detailed descriptions of preventive worTc. They also indicate possible malfunctions in the machines. Careful scheduled preventive maintenance work signif icantly reduces th.e appear- ance of random malfunctions in macfiines. But finding and eliminating malfunc~ tions quickly depends in large part on the experience and qualifications~of tfie engineering-tecfinical personnel who are operating the m~chines. Types of Malfunctions and Methods of Detecting Them Malfunctions may occur in computers for various reasons. Among them are con-- cealed production defects, violations of operating rules, and external influ-- ences (blows, vibrations, overheating, and the like). Malfunctions occurring - for these reasons generally cnme during the period of experimental operation. During working operations most malfunctions are related to failures of inte~ grated circuits and semiconductor instruments. Experience operating the machines shows that the cause of this is instability of the voltage in the machine cir- _ cuits. Failures in computers often occur because resistors go out, despite their high reliability. More than SO percent of all failures occur because the contacts that connect the current-conducting element with the outlets are broken or disrupted. Failures also occur as the result of rupture of the dielectric in the capacitors. Trouble-free operation of the semiconductor in~truments depends on hDw well tfiey are manufactured. These instruments are sensitive to overloads of current and voltage and therefore above-limit power supply regimes must ~e avoided if they - are to be used for a long time. Semiconductor instruments are also sub.3ect to the influence of heat. An increased temperature causes cbange in tfie parameters of the element and can result in failure of the computer. To make the search for malfuc~ctioning elements easier each element in the ma- chine has an address which gives the number of the unit, the frame, console, and place which the particular element occupies. These coordinatea are shoam in the structural and schematic diagrams by which tfie searcfi for malfunctions is conducted. The machines are supplied with service apparatus, including an oscillograpfi, an electrical measurement instrument, a stand for testing TEZ~s [cards] and a stand for testing power supply blocks, among others, to determine the operating stability of the particular elements. This apparatus plays a signifi- cant part in the search for machine malfunctions. However, tfie automatic moni- toring equipment is primary. It makes it possible to check whether the machine is~operating correctly and, in many cases, to correct errors that are detected. This includes both hardware and software for automatic monitoring. 16 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC!AL USE ONLY The software includes monitoring programs for test proUlems which are run be- fore performing the main protilem. When tt?e test prohlems are solved cor-- rectly the probability of correct work by the machine in solvtng ttie main problem is determined. During solution of the main prolilem monitoring proce-- dures are carried out in tfie form of a double check of the same program. ~ The hardware includes monitoriiig equipment tliat operates, independently of tfie program. For example, tlie YeS-1020 processor uses fiardware monttoring 5ased on "modulus two" and hardware duplication techniques. The "modulus two" monitoring method, in particular the ~~odd number check," is done byte by liyte, which is to say each byte of the data has a check bit that added to the number of units in the byte makes it uneven (odd). The hardware duplication method is used to monitor tlie work of the arithmetic-- logical unit. By this method each bit of information is processed in direct and inverse form. The results of this duplication are compared for eacfi.Fait. If the direct and inverse levels coincide this indicates an~error. COPYRIGHT: Izdatel'stvo "Finansy i statistika", 1y81 11,176 CSO: 1863/S1 17 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY PRQBLEMS OF AUTOMATED CONTROL SYSTEMS FOR INDUSTRIAL PROCESSES REVIEWED Moscow VOPROSY EKONOMIKI in Russian No 9, Sep 81 pp 67~77 - [Article by D. Palterovich and M. Gornshteyn: "Automating the Control of Technological Processes in Industry"] [Excerpt] The national economic effici.ency of automated control systems for technological (industrial) processes [ASU-TP's] depends not only on tfieir scientific-technical level, but also on the extent to ~fitch tliey cover exist- ing industrial processes and aggregates, and this figure is still low. In 1979 ASU-TP's had been introduced at slightly more than one percent of all in-- , dustrial enterprises. Each of these enterprises had, on the average, about two systems. All of the ASU-TP's introduced could be classified as follows: direct control system - 52.7 percent; information-advisor systems 16.8 percent; information recording systems 30.5 percent. We should observe here that the rate of development and extent of application of these systems does not meet national economic needs. Calculations that we ~ have made show that at the end of the llth Five Year Plan in tfie prtmary sec- tors of extracting and manufacturing industry, the proportion of industrial ' facilities equipped with ASU-TP's, although it is growing, is only about 30 percent of the total number of complex industrial facilities grepared for the ~ use of ASU-TP's and at which the use of such systems is economically expedient. These findings illustrate that the national economic need for A~U-TP's is great and cannot be fully met in one five-year plan. In view of the high efficiency of ASU-TP's, steps should be taken to meet the need for them more fully in the llth and 12th five-year plans. In our opinion, it would be wise to increase the proportion of expenditures for automation in gross capital investment. In petroleum extraction and nonferrous metallurgy, this propc:rtion is 2-2.5 percent, while in electrical power, ferrous metallurgy, and 'i:he gas industry it is 4-5 percent, in the building materials industry - 5.~ ~,ercent, in the chemical industry - 8 percent, and in petroleum refining - 7 percent. For a long time work to build and employ ASU-TP's was carriPd on in the context of rapid development of ASUP's [automated production control systems]; this - could not help being reflected in the overall distribution of capital, and 18 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02/09: CIA-RDP82-00850R000500434418-4 FOR OFF[CIAL USE ONLY apparently was not always justified. The proportion of investment for ASU-~P's in total capital investment for tiuilding automa.ted control systems o~ all types, despite significant growth in~ the lOtfi Five-Year Plan compared to the Eighth, was still insignificant and in the last 10 years fias b2en only 12 percent while investment for ASUP's was 68 percent. This distrihution of re- sources led to a situation where at the beginning of 1980 tfiere were 4,370 auto- mated control systems: 2,460 ASUP's and 1,649 ASU-TP's. At the present time there is less than 0.7 of an ASU-TP for each operating ASUP. But for efficient integrated automated control of industrial and economic-organizational processes at contemporary industrial enterprises there should be from a few to several dozen ASU-TP's for each ASUP. Only in this case can the automated control sys-~ tem for technology become�the foundation of an automated Production control sys tem (ASUP). At the same time, because building and operating ASU-TY's requires signif~cant � production expenditures,* the paramount challenge is to make optimal use of production resources for this purpose and to determine the rational order and scope of work to set up and employ ASU-TP's. Analysis of ASU-TP's shows the high economic efficiency of this form of tech- nical progress. One-time expenditures for the devel.opment and introductior~ of ASU-TP's are repaid on the average in 1-3 years, chiefly by reducing use of material and energy resources (by 2-5 percent), increasing the production of out- put (by 2-8 percent), improving the quality of output,~and other factors. The norm of efficiency of capital investment for setting up ASU-TP's was raised slightly for the llth Five-Year Plan. In the last three five-year plans several billion rubles have been,invested in development of ASU-TP's. Hundreds of scientific research, planning-~design, and technological organizations and enterprises and thousands of engineering- technical employees of various ministries and departments are engaged in des.ign- ing and introducing them. But the level achieved and the results of work to build and introduce ASU-TP's still do not fully meet national economic require- ments. In the first place, as already noted the scale of capital investment in auto- mation is inadequate to supply automated systems to all large industrial aggre- gates. In the second place, the choice of objects of automation, and conse- quently the distributior~ of capital investment among these objects, does not always meet the criterion of maximum national economic impact. In the third place, information and advisor systems still predominate in the structure of existing ASU-TP's. The proportion of control systems is rising slowly, and their functional capabilities are often limited and do not provide, on the one hand, for integration of the ASU-TP and ASUP into a single system for control of technology and production or, on the other hand, sufficient flexibility and adaptability of the industrial facility to change in work conditions. In the * The average cost of setting up one ASU-TP with a computer was about 1.3 mil- lion rubles, with an average labor intensity of 90 worker-days. 19 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAI. USE ONLY fourth place, the qual~ty and reliability of the functior.ing of ACU-TP's often declines because of disproportions in the development of their scientific- technical base and inadequate matching of the computer equipment employed, es- pecially peripheral units and software, to technical and economic requirements. In the fifth place, many ASU-TP's are built as unique developments for one-time application, so the level of unification and standardization of structural, ~ algorithmic, and technical concepts is low and they do not circulate. In the sixth place, the organization of the process of developing ASU-TP~s is not al- - ways oriented to rapid introduction and fittir~n the character of the ob~ect of automation. In the seventh place, the level of use of ASU-TP's in terms� of time and their functional capabilities is inadequate in many cases, which greatly reduces their economic efficiency. One of the most important problems that must be solved in the area of raising the efficiency of automatation of technological processes in industry is de- veloping the scientific-technical base of automation, improving planning, and straigntening out the or};anizational forms of work related to the development and application of ASU-T]''s. New scientific prin~ciples and engineering methods of building ASU-TP's must be employed to develop and introduce highly efficient automated technology. The specific ways to ac:.omplish this are determined by both the needs of industry and the capabilities of current automated control equipment. Contemporary forms of automating control, with their high requirements of in- dustrial equipment, themselves become a powerful factor in the transformation of equipment and technology on the basis of the advances of the current scientific-~ technical revolution. Aiitomation of control creates conditions for increas~ing the unit capacity of aggregates and the continuity of their operation, makes it possible to raise temperature, pressure, and other parameters of process.es, in-- sures operator safety under conditions of aggressive environments, exposure to radiation, and the like. In this way, the efficiency of ASU-TP's has a clearly marked socioeconomic character. In terms of functional capabilities tlie most modern domestic ASU-TP's today are as good as the best foreign models and can in certain cases match them for reliability, flexibility, and diversity of tech- nical resources employed. To raise the scientific-teciinical level of ASU-TP's it is essential to increase significantly the volume of diagnostic, optimiza- tion, and control functions in them and to make broader use of progressive scientific methods of control - direct ciigital control, optimal control by adaptive models, and others (at thermal power units, for example, in 10 years the number of parameters measured has increased six time~ and tfie number con- , - trolled has risen four times). Raising the scientific-technical level of ASU-TP's significantly broadens their functional and technical capabilities. For example, the ASU-TP's that were de- veloped earlier for sulfuric acid production facilities were information- dispatch~r systems and did not accomplish the tasks of operational control of production. The Kupol ASU-TP that has now been set up for the Gomel' Chemical Plant envisions direct digital control of sulfuric acid production. The basic distinction of thi_s system is its transition from automatic monitoring and analysis of a number of variables to automated regulation of a set of inter- related parameters based on the use of mathematical models and the techniques of 20 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02/09: CIA-RDP82-00850R000500434418-4 FOR OFF[CIAL USE ONLY combined control, to automatic selection of the hest technological regimes and coordination of the work of the entire line. Deviations in the regulated parameters from assigned values were cut by three-fourths to four--fiftfis~with introduction of the Kupol ASU-TP, wfiich made it possible to reduce losses of ex- pet.sive raw material by one-third to one-half. The growing complexity of ASU-TP's requires an enlargement of the memory of com- puter machines, broadening of their links with the object, development of ineans for operator "communication" with the system, and the like. One of the major scientific-technical advances in building ASU-TP's iri the lOth Five Year Plan was the use of third-generation control computer complexes in most cases. With respect to volume of.output and technical characteristics (with the exception of reliability) they meet the requirements of ASU-TP's. But the structure of production of computer equipment still does not meet the national economic need for it. Few specialized computers are available, while general-~urpose computers are too expensive and are not always adapted to the requirements of the control objects. Growth in the production of control computer complexes is far outstripping growth in the production of supplementarp units: internal memory units, units for communication with objects, and peripher~l devices. The structure of production of computer resources should be modified to sig- nificantly increase the proportion of peripheral units and units for communica- tion with the object. Industry should also produce a broader assortment of peripheral units (flexible disc stores, specialized production engineer-operator terminals), specialized sensors, actuating mechanisms, and certain other types of automation equipment that are produced in small series. . About one-third of the ASU-TP need for specialized instruments and automation equipment (they constitute 20-25 percent of the total production of technical ~ means for ASU-TP's) to monitor the main parameters of the .industrial process ; is being met. Growth in this need demanda the develqpment of essential design i subdivisions and experimental facilities oriented to devising not only individual , instruments, but entire sets of equipment for ASU-TP's. The report by N. A. Tikhonov at the 26th Congress of the CPSU notes that the production of minia-- ture electronic control machines as a constituent part of the basic industrial equipment, instruments, and various control and monitoring systems and devices is expanding significantly in the llth Five-Year Plan. This will step up tfie development of the technical base of the ASU-TP's significantly. The low reliability of the computer equipment and instruments being produced by industry is a serious obstacle to rais~ng the efficiency of ASU-TP~s. This makes it impossible to implement a number of technical concepts (for example, direct digital control) efficiently in industry and leads to parallel use of several computer devices and duplication of the system by manual control. As a result, one-time and ongoing expenditures of the system rise considerably and its economic efficiency declines. A further increase in the efficiency of ASU-TP's demands an improvement in reliability, particularly in tfie electronic part of the computer, raising the trouble-free period of work to 10,000 fiours. The efficiency of ASU-TP's depends significantly on improving the organ~zation of work to set them up. A specific feature of this work is that it cannot be 21 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 ~ FOR OFFICIAL USE ONLY done apart from the particular characteristics of the machinery and tech- nology of automated production. Technical policy in the field of automation of the control of particular industrial processes- and aggregates (production facilities) must therefore be viewed as a part of tlie technical development or re-equipping of the corresponding production su~isector. It must be coor- dianted with technical policy in the f ield of development of technology and designing of the primary production equipment. Expanding the scope of work to set up ASU-TP's demands a transi.tion to more productive, "industrial" methods of developing, implementing, and distributing them to insure an improvement in quality and a reduction in the labor-- intensiveness and time of work to set up ASU-TPfs. The most important steps to raise the efficiency of setting up ASU-TP's are tlie following: standardization and unification of design concepts; automation of their design; experimPntal field testing of pilot modeis of ASU-TP's. The rate and scale of development of automation of control over industrial processes is largely determined by the interrelations of the sectors that per- form scientific, design, and installation-adjustment work in setting up ASU- TP's, and the manufacturers of the industrial equipment and technical means of control. Comprehensive organization of work to automate the control of indus- trial processes should be based on uniform intrasectorial long-term programs of scientific research, design , production of technical means, installation, ad- justment, preparatfon of the control ob~ect, insuring efficient functioning and development of ASU-TP's, 2nd circulating them. Organizational forms of work to automate industrial processes have been taking shape for a long time under conditions of the multisectorial structure of in- dustry. Because numerous ministries hring together what are essentially several sectors that differ not only by the nature of their output but also by specific features.of industrial processes, work toward automation has been spread out at subdivisions of many different organizations. Even within a parti:cular sector numerous organizations subordinate to different departments work on solving the same technical problems of autorsating production. All this obstructs the con- tinuity of the process of development and introduction of ASU-TP*s. Certain specialized institutes develop ASU-TP's that are suitable for one-time use at a particular site, which retards the formulation of standard ASU-TP's for widespread introduction. Work sometimes goes forward in parallel at several enterprises, and the result is that different ASiJ-TP's are developed for identical enterprises. But experience shows that the total length of working time to set up an ASU-TP using standard concepts is 30-40 percent less than where there is individual development, and costs are 20-30 percent lower. The main wealmesses of the organization of work today to set up ASU-TP's result from the following factors, among others: the absence of uniform leadership and technical policy in the Eield of planning, coordinating, and monttoring ~ this work; the lack of specialized organizations responsible for technical policy and the organization of work to automate certain types of industrial processes; duplication in solving the same scientific and technical problems; and, the unsatisfactory state of experimental prod.uction faciltties. 22 _ FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY To improve the efficiency of work on ASU--TP's thei.r organizational forms must be coordinated with the content and scope of work on automation. Tlie time has come to develop a long-term program of scientific researcfi and planning-design work on automating the control of industrial processes, includ- ing tfie manufacture of pilot models of automated control systems for complex - industrial processes, development of mathematical models of control objects, development of standard algorithms and programs for ASU-TP's, and so on. The sectors should have long-term plans for setting up ASU-TP~s, and the resource requirements established by these plans must be taken into account in five' year and annual plans. Solving the problems of automating the control of industrial processes, wtiich. _ requires substantial labor, material, and financial resources, can only be done by large organizations. The most rational form of these organizations is the = science-production association. But there are still very few of these associ- ations. To avoid duplication in work to develop and introduce ASU~TP~s in various sectors, a number of steps should l~e taken to determine the specific specializations of organizations in automating simt.lar in~3ustrial processes and main designers should be appointed for groups of processes. Ttiese organi- zations must be responsible to the customer for performance of the full range of work related to ASU-TP's. In our opinion, work to set up pilot automated control systems for complex industrial processes should be concentrated at organizations of t~ie Ministry of Instrument Making, Automation Equipment, and Control Systems. After testing - the pilot systems should be turned over to sectorial organizations f~r distri- bution. This is a useful system of work because, in addition to tfie fact that this ministry has strong scientific, planning, and design organizations involved with ASU-TP's, of the common features of inethods of studying industrial processes as control objects, the possibility of standardizing the information being used, and the general industrial character of monitoring and regulation equipment and computer technology. The complexity of the equipment used in ASU-TP's necessitates a large number of highly qualified service personnel. Qualified specialists are needed for the software of the control computers and also to repair them. Tfierefore, it - is important to improve the organization of work to service the control com- puter complexes that are employed in ASU-TP's by estatilishing a procedure for centralized servicing by the manufacturing ministry. This will make it pos- sibte ta impr.ove the quality of service and reduce its cost. Among the factors that ar.e slowing down the introduction of automation at indus- trial enterprises and increasing the time of work we may also include unsatis- factory material-technical supply. In various sectors tfie planning of material- technical supply for work on automation is often done apart from tT~e time and scope of the work, and resources planned for automation are sometimes spent for other purposes. Therefore, we feel that special-purpose funds should be estab- . lished for material-technical supply of automation work and they sfiould be coordinated with the volume of capital investment in automation. 23 FOR OFFICI~?L USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 f~OR OFF(CIAL USE ONLY Poor material-technical s.upport for research and development and automation has a negative effect on work to set up ASU--TPTs. In addition, tfie organiza-- tions which are doing tfiis work must submit orders (requests) for technical means and materials one year before the beginning of the work mentioned above; ~ in some cases this leads to failure to s.upply essential materials and articles, while in other cases above-norm stocks of scarce materials are formed, working capital is frozen, and so on. We believe that it is necessary to establish an organizational solution so that orders for teclinical means and materials for conducting scientific research and experimental design work are filled ~ within three months. The ef~iciency of setting up ASiJ-TP~s is often lowered by incomplete supply of equipment. Individual assemblies, mechanisms, and instruments in unitary industrial aggregates equipped with automation means must be installed tiy primitive methods in the local areas. Both primary and auxiliary industrial equipment should be designed with b~iilt-in sensors and actuating mecfianisms or with assigned places where they :ire built in and then delivered by the man- ufacturing enterprises together witli automation instruments and equipment. At many enterprises, especially sec~ors where equipment operates under diffi- cult conditions (chemistr,~, petrochemistry, metallurgy, the tuel industry, and others), ASU-TP's are ~ised inefficiently because of a lack of spare parts. We must increase spare parts production considerably (to 25 percent of the volume of production of automation equipment) and create reserves of such parts. It would be wise to plan to meet the spare parts needs of operating ASU~TP's on an equal basis with supply of equipment to start-up projects. In a number of cases the setting up of automated technology is held back by failure to insure that projects planned for control by traditional methods and means are technically and organizationally prepared for automation. For many years automation was thought to be an independent stage of activity - toward which one should move on the basis of using already-existing industrial equipment. The typical negative aspect of the process of automation at existing enterprises was (and still is today in many cases) "building on" automation to existing equipment which is itl-suited for work under conditions of automation of control. Often even the new equipment proves unsuited for automatic regu- lation or for the volume of c~~ntrol functions included in the ASU-TP. As work on automation of control progressed it became increasingly clear, on the one hand, that it was not wise to switch certain aggregates and processes - (obsolete or unpromising ones) to automated control and, on the other hand, that industrial aggregates had to be rebuilt when transferred t~ automated control. Progress in the field of automating the control of industrial processes and a guarantee of its economic efficiency lie not in automating aggregates desig- = nated for manual control, but rather in setting up automated aggregates and industrial complexes tfiat are oriented to automated c~ntrol from tfie Fieginning of their development. ASU-TP's should be developed parallel with the creation of new tecfinology and new industrial equipment. To insure that ASU-TP~s are launched in operation together with the launching of production facilities it is necessary to care- fully plan the times at whicTi ASU-TP's are set up at construction projects 24 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 EOR OFF[CIAL USE ONLY - and sites being rehuilt. This procedure should I~e codified b.y working out norms and requirements for industrial equipment designated for work in an automated mode. Widespread and efficient development of ASU--TP~s is possible only if tfiere is allout standardization of technological concepts, diagrams, and camplexes. When technological diagrams are not standardized the result is greater complexity - of concepts during automation and difficulties in distributing the system. In addition, raising the level of technological standardization makes it possible to reduce the number of design stages and cut the realization time and cost of the system. Broadening the sphere of rati~nal application of ASU-TP~s is also linked to growth in mass production, the need to insure its smooth rhythm, and the necessity of increasing the stability of the composition of raw ma- terials and basic industrial parameters. Many steps toward automation (for example in ferrous metallurgy) are taken in sections tfiat have significant equipment downtime, violate production schedules, and suffer constant fluctu- ations in the content of raw materials. In order to carry out the measures to accelerate the launching of production capacities and facilities and raise the efficiency of capital investment as out- lined in the decree of the CPSU Central Committee and USSR Council of Ministers on improving tfie economic mechanism, it is essential to work out a sound approach to selecting the sites for which ASU-TP's are planned, The determining char- acteristics when selecting an industrial site to be equipped with a computer- based ASU-TP and when establishing the class of the particular control system should, in our opinion, be the following: the future promise of the industrial process, its degree of refinement and level of standardization from the stand- points of nature and structure; the dimen:~ions of the unit capacity of tfie installation; the value and scarcity of tlie output produced; the complexity of - the process from the technological and op~~rating standpoints; the frequency, level, and economic consequences of disturbances that affect the industrial process; the frequency of changes in the ~~roduction situation; the level of supply of monitoring and measuring instru~nents; and, the nature of internal (prodtiction conditions) and external (marl:eting conditions) constraints on pro- duction of output. Analysis of the technical-economic prereqi~isites of automation is very important when selecting a control object. Underestimation of the role of this analysis may lead to one of the significant si~ortcomings that have been experienced in planning the development and use of ASU-TP's in industry. Often the decision to set up an ASU-TP is inadequately substantiated relative to the particular site. This relates above all to determining the technological potential of the _ site and the potential oF control; identifying, evaluating, and analyzing pro- duction losses related to the level of control; and, evaluating the prepared- ness of the site for introduction of an ASU-TP. It is difficult to overcome this problem because the customer, when selecting tlie object and determining the level of its automation, is not at all responsi6le for the economic and technical wisdom of the decision being made. It appears that there should be a fi~ndameiital re-eval~iation of tfie role a.nd content of thP stage of "technical-economic s~~bstantiatio:~ of setting up an ASU-TP," in which - 25 F(IR OFFICIAL USE ONI.Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY the goals of setting up the ASU TP are formulated, i.ts structure is laid out, and interrelationsfiips raith the control system are establislied. It is in this stage, irc fact, that tiie foundations should Fie laid for the use qualities of automated control systems that make it possible to guarantee tfie cli~nt tiiat it will have at least normative economic efficiency. The number and quality of ASU-TP's are determined above all by national eco-- nomic need for them and the resources of the scientific-technical base. It is therefore necessary to analyze the national economic need for ASU-TP~s in order to improve the planning and organization of work to set tfiem up. This work is now done from one case to the next and does not have any kind of scientific-- _ methodological su6stantiation, which leads to great divergences in estimations of this need. The estimation of sectorial need for ASU-TP's made in 1977 for the appropriate period revealed that the estimate was several times less than the one established by earlier estimates done in 1973. The primary reason for ~ the discrepancies was that during the period of compiling the 1973 estimates the sectors did not have an adequately clear idea of the essential nature of ASU-TP's and their jobs and place in the overall system of production control. When indicating their need for ASU-TP's, the sectors in effect estimated the need for other forms of automation, including local automation. needs are overstated or understated as a result of an incorrect estimation of the technical feasibility and economic wisdom of introducing ASU-TP's for spe- cific industrial facilities. The ASU-TP's now being developed and introduced are very diverse botfi with - respect to scale and nature of problems solved and the functional-algorithmic structure corresponding to them and ways and meane of solving them. Therefore, in order to estimate the national economic need for ASU-TP's, to plan for setting them up and provide production resources for this work, and to monitor per- formance of the assignments of national economic plans tfie full set of ASU TP~s should be broken down into subsets (groups) characterized by definite features. The most significant of these features are the following: (a) nature of tfie occurrence of the controlled industrial processing time degree of continuity of arrival of raw material, relative length of particular operations; (b) de- gree of functional development - complexity of information and control func~ tions realized; (c) information output - the number of technological vari- ables measured or monitored by the ASU-TP. In our opinion, to improve the planning and organization of work to set up ASU- TP's and to distribute production resources for automation of control it would be advisable to work out a system of norms. This system should include norms for specific capital investment to set up ASU-TPts of different classes, labor expenditures for different stages of setting up and using them, serviCe lives, and expenditure of operating materials. On a national scale it is necessary to conduct factory certification (by "pass-- ports") of the most important types of automatable industrial processes according to their basic technical-economic characteristics as control objects. This will make it possible to identify the primary kinds of control objects, group olijects by types, establish the characteristic features of automatable processes, aggre- gates, and facilities as control objects, and more closely coordinate tfie - 26 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY creation of new technological processES and aggregates with automation of their control, thus insuring maximum economic efficiency. In the current pfiase of development of ASU-TP~s, during the period of their growing distribution, it is necessary to take a large number of economic- organizational measures related not only to the development ef systems (tfieir organizati.onal structure and preparation of the ob~ects), but also to production control under e~nditions of existing ASU-TP's, furtiier development of function- ing systems, and raising the efficiency of automated control. Many of these economic-organizational questions are either being decided in an unsatisfactory manner or simply not considered at all. Thus, there is no comprehensive develop-- ment underway on questions of improving the control system and methods of planning, monitoring, recording, and analyzing the production-economic activity of the entervrise under ASU-TP conditions. Proper attention is not being given to improving control processes under ASU-TP conditions, or to economic stimu- lation of ASU-TP developers and ope~ations personnel (giving them incentive to improve the efficiency of control systems~. We do not today have a clearcut system of economic stimuli for the development, ogeration, and continuous refinement of ASU-TP~s. In most industrial sectors the wages of automation equipment service personnel are lower tfian those of pr~' mary production workers. We must eliminate this ine.quality and increase the icnentive for production and management personnel to use ASU-TP's efficiently. Precise organizational forms Eor managing the development, introduction, and elaboration of ASU-TP's have not been established at the enterprises. There often is no specialized subdivision in charge of setting up and elaliorating the ASU-TP. The number of ASU-TP service personnel at most industrial control oii- jects is inadequate. The monitoring-measuring instrument and automation shops are inappropriate for the jobs of operating ASU--TP's. There is no statute on ASU-TP services at enterprises which would estahlish ap- propriate legal norms and rigidly define the accountability of the client and developer during the setting up and use of an ASU-TP. The client is not ac- countable for formulation of the specific technical-economic goals of setting up the ASU-TP, for the economic and technical wisdom of the decision he makes, for timely preparation of service personnel, for general preparation of the object for work under ASU-TP conditions, and for insuring that the system is used efficiently. The functions and responsibility of tfie client during develop- ment and use of the ASU-TP should, in our opinion, be clearly established in a special statiite ratified by the USSR State Couunittee for Science and Technology. Along with other pattis to improving ASU~TP's that have been considered, solving the problems of economic-organizational support of these systems will make it possible to improve the efficiency of their setting up and use in industry. - COPYRIGHT: Izdatel'stvo "Pravda". "Voprosy ekonomiki", 1981 11,176 CSO: 1863/46 27 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFF[CIAL USF. ONLY HYBRID COMPUTERS HYBRID COMPUTING MACHINES AND SYSTEMS: LOCAL AUTOMATED CONTROL SYSTEMS AND COMPUTER DEVICES Kiev GIBRIDNYYE VYCHISLITEL'NYYE MASHINY I KOMPLEKSY: LOKAL'NYYE ASU I USTROYSTVA VYCHISLITEL'NOY TEKHNIKI in Russian Na 4, 1981 (signed to press 6 May 81) PP 117-121 [Annotation and abstracts of articles in collection "Hybrid Computing Machines and Systems: Local Automated Control Systems and Computer Devices", edited by G.Ye. Pukhov (editor-in-chief), et al., Izdatel'stvo "Naukova dumka", 1000 copies, 121 pages~ - [Text] Annotation This collection presents the results of scientific research on the theory, methods and algorithms for hybrid computation, on the development of computer and peripheral devices, and on analyzing the accuracy, reliability and diagnostic methods of hybrid systems. Some problems in developing hybrid local automated control systems are considered. The collection is intended for local automated control system ar.d compuCei hardware developers, as well as students, graduate students and scientific workers specializing in the area of hybrid computer technology. UDC 536.629+681.34 APPROXIMATION OF DYNAMIC CHARACTERISTICS OF OBJECT WITH DISTRIBUT~D PARAMETERS USING R~-NETWORK MODEL [Abstract of article by G.V. Biryukova and G.V. Yevstratov] [Text] It is shown possible to obtain a~homomorphic mathematical model of an object with distributed parameters (DP-object) using a hybrid computer system (HCS) containing a dynamic analog model of the thermal object in quesCion in the RC-network. An algorithm is given for approximating the dynamic characteristics of the DP-object using an ordinary second-order differential equation with a delayed argument. Four bibliographic references. - 28 . FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFF[('IAL USE ONLY UDC 681.33 DIGITAL MODELING OF DISCRETE-ANALOG NETWORK PROCESSOR [Abstract of article by G.N. Azarov and V.Ye. Prokof'yev] [Text] Problems of investigating the accuracy of RC-models on pulse-time adjustable conductances using digital modeling methods are examined. It is shown that there is no propagation of inethodological error in the discrete- analog network model from node to node, and the methodical ~-rror is obtained as a function of the model parameters. 'Itao illustrations, one table, one bibliographic reference. UDC 681.3 HYBRID DEVICE FOR DEFINING CORRELATION FUNCTIONS OF NORMALLY DISTRIBUTED STATIONARY RANDOM PROCESSES [Ahstract of article by V.F. Kornilovskiy] [Text] A device is described for defining the correlation functions of normally distributed stationary random ~rocesses. The device consists of a digital (sign) correlator and a digital-analog converter. The device uses series 155 integrated circuits. Four illustrations, three bibliographic references. UDC 518:517.944/947 - MODELING OF PROCESSES OF INTRA-RESONATOR GENERATION OF SECOND OPTICAL HARMONIC CONSIDERING TRANSVERSE HETEROGENEITY OF RADIATION [Abstract of articl.e by A.A. Glushchenko, B.P. Dovgiy, V.V. Obukhovskiy and V.L. Strizhevskiy] [Text) Equations modeling the subject generation are investigated numerically ~with the combined influence of nonlinear, diffraction and dissipative losses for various gains. The optimal linearity which provides the greatest intensity of the second harmonic is found. Four illustrations, 11 bibliographic references. UDC 681.3.06:681.34 ALGORIT[~M FOR ANALY7ING PATCHINC SCHEME OF OPERATIONAL MODULES USED TO SIMULATE AUTOMATIC PROGRAMMING SYSTEMS FOR ANALOG COMPUTERS [Abstract of article by A.N. Klimenko] [Text) An algorithm is described for analyzing the patching schemes of operational modules in the analog sectiun of a hybrid computer section c~~hich can be used to select the optimal version from the viewpoint of reducing the number of modules 29 FOR OFFiCIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY used in the analyzed circuit. The structure of the algorithm and an example of its use are examined, and the results are discussed. Two illustrations, t~o bibliographic refarences. UDC 62-503.3 SOME AI,GORITHMS FOR IMPLEMENTING CONVOLUTION OPERATOR AND THEIR APPLICATION [Abstract of article by A.F. Verlan' and B.B. Abdusatarov] [Text] Various questions involv:d in the numerical realization of linear integral operators and Voltaire equations of the second sort with arbitrary and - partionable kernel are considered. A series of functional diagrams of special- purpose computers used to implem~~nt integral operators and solve integral equations are also presented. S~~ven illustrations, three tables, four bibliographic references. UDC 681.33 DEVICE FOR AUTOMATED INPUT OF PARAMETERS OF DISCRETE-ANALOG NETWORK MODEL [Abstract of article by V.V. Garmash] [TextJ A device for inputting network model parameters and interfacing with the memory circuit of the node element is examined which can be used to fully automate the process of inputting RC-network parameters. The device can be used as the basis for creating a"RC-network digital computer" hybrid system. UDC 681.3.056 WIDEBAND CODE-TIME INTERVAL CONVERTER [Abstract of article by A.N. Bazhenov and Yu.I. Gerashchenko] [Text] A converter is examined which uses information from a digital computer or code assignment device to form a time interval in eight ranges (from 10-~ to 103 sec) in increments of 0.1% of the maximum limit of the range. The con- verter is implemented using series 155 integrated circuits with minimum hardware. One table, one illustration, four bibliographic references. UDC 621.314 ANALYSIS OF INFLUENCE OF AVERAGERS ON CODE-TIME INTERVAL CONVERSION ERROR [Abstract of article by M.G. Rokhman~ [Text] This article analyzes the influence of buffer averagers connected to the input of an adjustable frequency divider on variation of the error of a code-time 30 FOR OFFICIAL l1SE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY interval converter. The most characteristic combinations of binary sub- harmonic components of the output frequency of the adjustable frequency divider are examined. One illustration, four bibliographic references. UDC 621.317.08 INVESTIGATION OF ROTATION SPEED-TO-PULSE FREQUENCY FEEDBACK CONVERTER [Abstract of article by V.I. Dotsenko and B.A. Furman] [Text] This article demonstrates the advantage of modulating the U-factor of the sensing element of a feedback converter prior to modulating the feedback coefficient. A method for improving the converter resolution is examined, and recommendations are given for the choice of frequency and operating clearance. Three illustrations, two bibliographic referencea. UDC 681.34 PROCEDURE FOR CONFIGURING PATCH PANEL WITH MATRIX SWITCHES FOR HYBRID COMPUTER SYSTEMS [Abstract of article by V.N. Gugnin] [Text~ This article considers configuration of the patch panel and logical description of the modules in a device for controlling an automatic patch panel using structural matrices of acceptable connections between decision modules in the analog section of a hybrid computer system. The application of the configuration of the patch panel of two-stage switching circuit is demonstrated using an example. Eight illustrations, two bibliographic references. UDC 681.33 NODE ELEMENT OF DISCRETE-ANALOG NETWORK PROCESSOR [Abstract of article by V.M. Andriyevskiyj [Text] Questions involved in constructing a node element of a discrete-analog network processor based on pulse-time ~~djustable conductances are considered. The advantages and disadvantages of these node elements are pointed out, and ways of improving their accuracy and expanding their functional capabilities are examined. The functional diagram of a node element with memory devices which assumes the use of a digital computer to automate parameter setting is proposed. Three illustrations, six bibliographic references. 31 FOR OFFICIAL USE ONLY . APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 ~OR OFFIC'lAl. USF. ON1.Y UDC 681.333(088.8) SIMULATION OF DOMAIN-BOUNDARY FORMATION MODULES USING ALGEBRAIC LOGIC [Abstract of article by K.A. Babordin] [Text] The application of the apparatus of algebraic logic is demonstrated for formalizing the development stage of the circuits used in the domain-boundary formation modules which make up probabilistic hybrid computer systems which are oriented towards solving boundary problems. Five illustrations, three bibliographic references. UDC 681.326.74 METHODS FOR ORGANIZING DIAGNOSIS OF SPECIAL-PURPOSE PROCESSORS AND DEVICES [Abstract of article by R.S. Khalatyan) [Text] This article examines questions involved in organizing microdiagnosis of special-purpose processors (SP) and devices using diagnostic instructions represented by ~,ecial microprograms. A generalized SP structure is presented, along with the sequence used by the special processor to execute the diagnostics. Besides a two-level (program and microprogram) organization ~ of SP diagnostics, another version is proposed which combines analysis of the response of the SP with a procedure for diagnosing it at the microprogram level. The possibility of diagnosing peripheral devices without operator intervention is indicated. ~ao illustrations, 16 bibliographic references. UDC 681.326.7 FULLY SELF-VERIFYING TWO-LEVEL CONTROL CIRCUITS FOR m OF 2m+1 AND m+l OF 2m+1 CODES [Abstract of article by V.V. Neshveyev] [Text] A simple method is proposed for partitionin~ a get of code words of an m of 2m+1 and m+l of Zm+l codes into two non-intersecting subsets which allows for the structure oE these codes. It is fairly easy to use the partitioning obtained to construct fully self.- verifying control cirr.uits for m of 2m+1 and m+l of 2m+1 codes. One illustration, five bibliographic references. UDC 621.382 AUTOMATED MONITORING SYSTEM FOR DISCRETE DEVICES [Abstract of article by V.V. Antosik, P.M. Demochko and R.I. Krutykh] 32 FOR O~'FIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY [Text] This article examines an automated control system controlled by an "Elektronika 100/16I" computer which can be used for functional control of discrete devices with up to 512 outputs. The functioning accuracy of the control system with single-output and two-output control circuits is estimated. One illustration, four bibliographic references. UDC 658.5;681.3;62--501.72 METHOD FOR DESIGNING DIGITAL SYSTEMS FOR REGULATING TECHNOLOGICAL PARAMETERS WITH SELF-TESTING [Abstract of article by T.G. Mashchenko, O.I. Potepukh and T.A. Yskovenko] [Text) This article examines the possibility of self-testing during functioning of digital systems which regulate technological parameters. Duplication testing is used to organize system self-testing. A digital system for regulating technological parameters is used as an example for examining the methodology of designing self-testing systems. Three illustrations, four bibliographic references. UDC 681.324 ESTIMATING EFFICIENCY OF SPECIAL-PURPOSE COMPUTERS [Abstract of article by Sh.Sh. AgzamovJ [Text] Criteria are proposed for estimating the efficiency of special-purpose computers which allow for the specific characteristics of the computers, the problems to be solved, and the specific requirements of pract~cal utilization of the devices. Allowance is made for limitations on the parameters and the influence exerted by each parameter on the operation of the device. The efficiency of special-purpose computer:: is estimated with respect to a single abstractly simulated standard structurE� which has optimal parameters and solves one specific problem. One illustration, seven bibliographic references. UDC 681.142.65 ~COMPARATIVE ESTIMATION OF ACCURACY AND SPEED OF COMPUTERS IN SOLVING STANDARD PRO}3LI?:dS A RF.AL-7'IAif? RING TI:ST ; _ [Abstract of article by Sh.Sh. Agzamov] [Text] This article considers problems involved in comparative estimation of the accuracy and speed of the EMU-10 analog computer and of digital computers - in solving the ring test problem (stanclard problem) in real time. A comparative estimate of the accuracy and speed of an ~i-7000 digital controller and a digit-analog computer system is made for solving a standard problem using 33 FOR OFFI('[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFIC[AL USE ONLY the first-, second- and fourth-order Runge-Kutt method. One illustration, five tables, six bibliographic references. UDC 621.391.175 ANALYSIS OF CONTROL SYSTEM DATA PROCESSING DEVICES [Abstract of article by G.F. Krivulqa and A.A. Ushakov] (Text] This article examines the organizational principles of data processing devices in control systems with a"pass-fail" classification of the object. The estimate and results of computer modeling of different methods of organizing control systems are presented. Two illustrations. UDC 621.9.06--529: 681.323 EFFICIENCY OF UTILIZING INTERPOLATION ALGORITHMS IN NUMERICAL PROGRAM CONTROL SYSTEMS [Abstract of article by V.D. Baykov and S.N. Vashkevich] [Text] The use of additional microcircuits to increase the capacity of micro- processor devices is proposed for numerical program machine tool control systems. Quantitative estimates are given for interpolation algorithms with and without additional microcircuits, and the effect of their use is evaluated. A criterion is cited which demonstrates the possiblity of increasing microprocessor capacity by using additional microcircuits. Ztao tables, six bibli.ographic references. UDC 681.325.65 AUTOMATION OF ADJUSTMENT OF ELECTROTECHNICAL LSI TOPOLOGY TESTING MODULE IN SAPR USING SYMBOLIC METHOD [Abstract of article by V.P. Rubtsov and A.M. Abbsov] [Text] A method is examined for automating electrotechnical testing of the symbolic plan and then the topological drawing of large-scale integrated circuits. The task is reduced to the solution of a system of logical equations. The method has sufficient technological stability and guarantees complete testing. Three illustrations. ' UDC 681.34 COMPUTER CAPACITY IN EXECUTING FUNCTIONAL DA7'A TRANSFORMATIONS [Abstract of article by N.I. Korsunov] [Textj The capacity of digital, analog and hybrid computers in computing functions is examined on the basis of the input data tran~formation rate. - � 34 FOR OFFICI,~L USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY Relationships are obtained which can be used to determine the capacity of computers and the necessary memory load. The effi.ciency of using various types of - computers for function computation is analyzed, and an example is presented and used as a basis for obtaining numerical results. Five bihliographic references. UDC 621.3.019.3 SPECIAL STAGE IN OPERATION OF DIAGNOSTIC AUTOMATIC PROCESS CONTROL SYSTEM ArCOMPANIED BY MANIFESTATION OF DEFECT [Abstract of article by V.P. Shargovskiy and S.A. Strel'tsov] [Text] A technical diagnosis stage is isolated from the instant the occurrence of a~3efect is registered to the instant the tested object and the technical devices associai.ed with the diagnostic automatic process control system are disconnected. A justification is provided for the importance of accelerating preliminary locations of a defect of the tested object as a whole, or the automatic process rontrol system as a whole. Two illustrations, three bibliographic references. UDC 681.33 SYNTHESIS OF OPTIMAL DIGITAL-ANALOG REt;ULATOR FOR CONTROLLING THERMAL OBJECT [Abstract of article by G.V. Yevstratov] [Text] This article provides a brief review of different approaches to synthesizing optimal control of thermal objects, noting their shortcomings. Based on the example of synthesizing a digital-analog regulator which is optimal in terms of speed it is shown that this problem can be solved efficiently on a hybrid computer system containing a dynamic model-analog of the thermal bbject in the RC-network circuit. Four illustrations, eight bibliographic references. UDC 62-83 TECHNICAL LINEARIZATION OF CONTOURS OF DIRECT DIGITAL REGULATION [Abstract of article by A.I. Ovcharenko] 'jText] An original device for technical linearization of direct digital regulation systems is investigated; speed and error characteristics are obtained. Three illustrations, two bibliographic references. 35 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL tISF. ONI.Y = UDC 62-50-503 PARAMETRIC IDE~ITIFICATION OF HIGH SPEED AUTOMATIC REGULATION SYSTEM CIRCUITS WITHIN CLASS OF LINEAR IMPULSE MODELS [Abstract of article by A.I. Ovcharenko] [Text] An approach is proposed for synthesizing digital automatic regulation systems based on a priori knowledge of the parameters of the linear analog system-model. Criteria are proposed for the correspondence of the impulse and continuous models and parametric identification of the impulse linear model. is done. The condition for suppressing periodic modes caused by level quantization in the analog-digital converter is obtained. Two illustrations, six bibliographic references. UDC 681.3.02 ARRANGEMENT OF COMPONENTS ON PRINTED CIRCUIT BOARD SUBJECT TO VIBRATION [Abstract of article by E.N. Rybnikov and A.I. Khyannikyaynen] [Text] This article exami.nes the problem of optimal arrangement of electronic radio components on a cir~:uit board subject to vibration. A foundatior~ is provided for the optimization criterion probability of failure-free operation. The problem is stated in general form, and limitations are formulated. Five bibliographic references. COPYRIGHT: Izdatel'stvo "Naukova dumka", 1981 6900 CSO: 1863/64 - 36 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY UDC 681.3.06:681.34 ALGORITHM FOR ANALYZING PATCHING SCHEME OF OPERATIONAL MOBULES USED TO SIMULATE AUTOMATIC PROGRAMMING SYSTEMS FOR ANALOG Kiev GIBRIDNYYE VYCHISLITEL'NYYE MASHINY I KOMPLEKSY: LOKAL'NYYE ASU I USTROYSTVA VYCHISLITEL'NOY TEKHNIKI in Russian No 4, 1981 (signed to press 6 May 81, manuscript received 10 Jan 80) pp 20-23 [Article by A.N. Klimenko from book "Hybrid Computers and Systems: Local _ Automatic Control Systems and Computer Devices", edited by G.Ye. Pukhov (editur-in-chief) et al., CzdaLel'stvo "Naukova dumka", 1000 copies, 121 pages] [Text] In designing hybrid computer systems (HCS) with automated analog computer programming systems (AACPS), the problem of arranging the operational units of the analog section in the HCS design units, or modules, must be resolved [1], The quantitative and qualitative makeup of the HCS modules combined by a common hierarchical switching system, and of the problem set [2], must be established a priori in each specific HCS during the design process. In order to minimize - developmental hardware costs without limiting the functional capabilities of the future HCS with AACPS ir is necessary to make a preliminary analysis of the operational unit patching schemes synt}tesized by the AACPS for various mathematical descriptions of the systems of differential equations to be solved. As a result, this analysis should provide material for optimal arrangement of the operational units by modules and to determi~te the nature, type and number of connections between operational units within and wit:hout a module. ~ This article describes an algorithm for analyzing the patching circuits of operational units in the analog section of an HCS. The structure of the algorithm and an example of its utilization are examined, and the results are discussed. Ttie criterion for optimal arrangement oE operational units in modules is an indicator wt~ich makes it possible to co~icentrate operational units in each module - which have the maximum number of interconnections and minimal interconnections with units which are part of other modules. The unit combination efficiency indicator v = r/(r + s), (1) 37 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY where r is twice the number of interconnections between the operational units which are proposed to be contained in a single module; s is the number of external connections of these units. Figure 1 shows the flowchart of this algorithm. The initial data for the program consists of the following files: string array SMT[1:T] the module composition by types of operational units; integer array SMK[1:T] module composition in - terms of number of like operational units; boolean array P[1:N,1:F] an ordered file of connections between operational units in the circuit being analyzed; string array MS[1:N] an ordered file of tl~e operational unit types which are part of the circuit being analyzed. Here N is the number of operational units in the circuit; F= N+ Q, where Q is the number of nonstructural inputs and outputs in the operational unit patching circuit. - Block PM1 executes the following operations: converts file SMK to an analogous file MKT [1:KRT] by removing those operational unit types from SMK which are not used in the circuit being analyzed; forms file MN [1:KRT], which contains information about the maximum number of different types of operational units in the circ~x~.t being analyzed; computes M, which is the maximum attainable number of operational units in the module for a particular circuit; _ pads files RP and SP [l:N] with values or ri and si computed by file P for each operational unit. Block EF1 computes the efficiency of combination for two operational units connected directly to one another. Block EF2 computes the efficiency of combining MTEC = 3, 4, M operational _ units in a module. The combination of operational units in the circuit wh ich are acceptable from the viewpoint of qualitative and quantitative module composition are examined for each value of M'CEC, and file KORD [1:N(M-2)] is used to store the combination with maximum ei~ficiency which contains that operational unit for each of the N operationr~l units. File EF 2[1:N(M-2)] stores the current maximum efficiency values achieved during the sorting, while file KORD stores the coordinates of the combined operational units. If this section of the algorithm were to be implemented on the basis of a combination generator with complete trial and error, the time expended would be estimated by the number M (2) KC = ~ ~TEC. MTEC=3 38 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY ( 1 ) avaea M7�C~ =2,3ada- / ~aKaHUUUa.eu- ( 5 ~ ycmoyaBKO' ~ 9 ~ ~eNavo~onnzn ` 2 ~ ~udONNeAtCa4 C~~~N020 3N04PHUA /lOI y~ (1 Noeo 9 a e~~b - PMf ( 6 ) emo ,4a~3) MTEG�a ada uenovo~e- EfA (11) ~ioa~~~~"�dm� ~y Ec~i (16 ) MTEC Qo ~ 3) ( 3) HE. ~l aeu ~ M Q~ o ou eu ~ 7) b~l MTEC l ao Hem~4) Hem~4) ~4) ' ~M (3) Mrec: ~ ~ I 2~ no2 Hem MTEC+r t ~ ~ 4 ) ove�~o�v ~ 8 ) ~yy~0 ~ 3 ) .6Qt~ raw~mcnup Q Q~ 3 ) MTfC-2 Hem brP2 Ol Ndn ( L? ) QQ ~ 3 ~ /ly4lUPH ~ 4)Hem aueu Qa ( 3) 15 ) , EF~ eoprroN~ 13 ) BadaNUe oyeped- vM2 (14 ) ,y~ BaAuo~rmn Figure 1. Key: 1. start 9. MTEC: =2; assign initial value 2. initialization and data PO1 input block 10. assign initial value of P02 3. yes 11. assign initial version for 4. no file RE S. set iriitial combinatio?ZS 12. P02 improved? 6. combination acceptable? 13. end of alternate? 7. end combination 14. assign new version 8. set next combination 15. PO1 improved? 16. end The number of trials is reduced by supplementary analysis of the conditions for placing an operational unit in a module allowing for limitations with respect to number and type contained in files MN and MKT. After eliminating identicat combinations from file KORD, block PM2 forms binary file NOM [1:MIKON[M-2]~, wiiose boundaries are stored in file MIKON [1:(M-2)] for each value of MTEC, for each value of MTEC. These are then used as the values - of the variable IKON. _ 39 _ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY The next segment of the flowchart of the algorithm processes file NOM, which contains the set of combinaCions which can be used as the basis for organizing modules. Some of these combinations form nonintersecting sets, i.e., they contain different operational-unit numbers which are assigned to the operational units during synthesis of the functional diagram. These combinations correspond to maximal utilization of the capabilities of the hardware content of the modules when they contain the operational unit of a specific circuit. The choice of combinations which are nonintersecting sets in the aggregate is not unique. The algorithm selects as optimal the version which corresponds to one of the maximal numbers of nonintersecting sets. This version provides the minimum number of residual combinations at the same time. These occur as the result of eliminating the operational-unit numbers which belong to the non- intersecting sets from the combinations which comprise intersecting sets. The residues, which are subsets, are eliminated. Finally, the number of design modules fitted in a particular circuit will be minimal. i ' ~ S ~ . ~ i i ~ I i s r------i~ ~ I ~I I i ~ ----u, j ~l ~ I ~r I I ~ I ~ f ~ p g 9 /0 II I ~ I - ~ ~ ----J 1~----~ j c, ~ ~ . i ~ i t ~ I v, ~ � ~ I /l 4 ~ ~ ~ I I ~I Figure 2. File RE [1:IKON] is used to store the intermediate results. The final results for c~ach value of MTEC are stored in file REO [1:IKON]. The optimal final result of all MTEC values obtained is selected. This is entered in file REOPT [1:IKON]. These files are formed by blocks BF1, BF2 anci BF3. Optimality indicators PO1 and P02 are used in searching for the optimal version. The program which implements this algorithm is written in PL/I and includes only 300 operators. The debugging was done on a YeS-1020 computer. Program translation and editing requires 9 minutes. The example presented below was solved in 2 minutes. 40 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 NOR OFFICIAL USE ONI.Y The functional diagram of the operational unit patching scheme shown in Figure 2 is used as the example. Example. Initial data: SMT [1:T] = C, H, Y, U, SMK [1:T] = 2, 2,3,2; N= 11; B = 23; P[1:N,1:F] = 0,0,0,0,1,1,1,0,0,0,0,1,0;0,0,0,0,0,1,1,1,0,0,0,0,0;0,0,0, 1,0,0,0,1,1,0,1,0,0,;0,0,1,0,0,0,0,0,0,0,1,0,0,;1,0,0,0,0,0,0,0,0,1,0,0,0;1,1, 0,0,0,0,0,0,0,0,0,0,0;1,1,0,0,0,0,1,0,0,0,0,0,0;0,1,1,0,0,0,0,1,1,0,0,0,0;0,0, 1,0,0,0,0,1,1,1,0,0,0;0,0,0,0,1,0,0,0,1,0,0,0,1;0,0,1,1;0,0,0,0,0,0,1,0,0. Here C, H, Y, U are the names of the operational units. Results: r M= 4; KRT = 2; I~IICT [1:KRT~ = 2,2; MN [1:KRT] = 6,11; RP [1:N] = 0,0,0,0,0,0,2,2,2,0,2; SP [1:N] = 4,3,4,2,2,2,2,3,3,3,2; MIKON [1:(M-2)] = 5,11; MTEC = 3; NOM [1:5] = 1,5,7;2,6,7;3,4,11;3,8,9;5,9,10; EF2 [1:5] = 0.600; 0.666; 0.800; 0.714; 0.600; REO [1:5] = 2,6,7;3,4,11;5,9,10;1;8; - MTEC = 4; NOM [1:6] = 1,2,7,8;2,6,7,8;3,4,8,11;3,5,8,9;3,4,4,9,11;1,5,7,10; EF2 [1:6] = 0.625; 0.714; 0.800; 0.625; 0.800; 0.615; REO [1:6] = 3,4,9,11;1,5,7,10; 0; 2,6,8;0;0; REOPT [1:6] = 3,4,9,11;1,5,7,10;0;2,6,8;0;0. The dotted lines in Figure 2 set apart the sections of the circuit which the program placed in different design modules in accordance with the data held in file REOPT. In terms of the number of combinations tried, the following results were obtained far this example. With a complete exhaustive search of (2) KC = ~311 + ~411 - 165 + 330 = 495. ~ Consi.dcring the limitations included in the algorithm, this number of combinations was KCtrunc - l51 + 199 = 350. ~ 2he number of combinations for which the efficiency is computed is KCaccept = 135 + 150 = 285. 41 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY It follows from this data that in this example when exhaustive search is used, 210 combinations are unacceptable, while 65 are unacceptable for the truncated - search. We note in conclusion that repeated application of this program to different functional diagrams of operational unit patching makes it possible to work out an optimal composition of modules for a given class of problems solved by an HCS, and to obtain statistical data for equipment loading and configuring automatic HCS patching systems. BIBLIOGRAPHY 1. Hannauer, G. Automatic patching for analog and hybrid computers. Simulation, 1969, 12, N 5, p. 219-232. 2. Kalashnikov, V.I., Klimenko, A.N. "Approach to creating automatic patch panel for analog computer", in "Teoriya, matematicheskoye obespecheniye i primeneniye neodnorodnykh vychislitel'nykh sistem: Materialy seminara" [Theory, Software and Application of Heterogeneous Computer Systems: Seminar Materials]. Moscow, December 1973. Moscow, Izdatel'stvo MDNTP, 1973, pages 153-158. . COPYRIGHT: Izdatel'stvo "Naukova dutnka", 1981 ~ 6900 CSO: 1863/64 42 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFIC7AL t1SE ONLY ~ - UDC 681.326.74 i METHODS FOR ORGANIZING DIAGNOSIS OF SPECIAL-PURPOSE PROCESSORS AND DEVICES Kiev GIBRIDNYYE VYCHISLITEL'NYYE MASHINY I KOMPLEKSY: LOKAL'NYYE ASU I USTROYSTVA VYCHISLITEL'NOY TEKHNIKI.in Russian No 4, 1981 (signed to press 6 May 81, manuscript received 11 Jan 80) pp 56-59 [Article by R.S. Khalatyan from book "Hybrid Computers and Systems: Local Automatic Control Systems and Computer Devices", edited by G.Ye. Pukhov (editor-in-chief) et al., Izdatel'stvo "Naukova dumka", 1000 couies, - 121 pages] [Text) The extensive use of microprogrammin~; in computing practice has led to the development of diagnostic methods with improved localization capacity. Based on analysis of computer microdiagnostic procE�dures (microprogrammed diagnostics), the item [1] isolates methods of implementation which use programs of instructions which comprise the instruction system of the computer being - diagnosed, as well as special microprograms. It should be noted that certain assumptions limit application of the methods considered in ~1]. For example, the remark in [2] with respect to the implicit - assumption that micro-operations can only "disappear" is valid. We might add that the method proposed in [1~ for obtaining the set of micro-instructions ~which makeup a special diagnostic microprogram facilitates obtaining micro- instructions with compatible micro-operations; however, the question of selecting those micro-instructions which would satisfy the "observability" condition defined in [3] still remains unresolved. The results of [4,5], which examine ~"unspooling" procedure which can be used to define a special microprogram, agree with the transportability and observability cnnditions in [3]. However, these articles examine discovery for an autonomous device. The case of asynchronous devices, for all of which Eransportability condition must be met in diagnosis, is omitted. Articles [6-8J are practical examples of microdiagnostics executed by special microprograms for devices interfaced with computers. However, these articles r�esolve the problems of diagnosing only individual sections, rather than the entire device. For example, [7] develops a flag for loss of data occurring due to loss of synchronization signals for a disc controller. 43 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY The development of computer peripherals and increased requirements for repairability ; make it necessary to work out methods of organizing diagnostics with application ~ to broad classes of interfaced devices. One such class consists of special- ( purpose processors (SP) for processing data files. Analysis of SP development ~ carried out in [9-16] allows us to isolate the following common structural properties: ~ the presence of communications devices which facilitate two-way communications between the SP and central processor, as well as the arrangement of controlling the exchange of control words and operands with the CP; the presence of two asynchronous microprogrammed controllers, one of which operates with the controller and the other with the arithmetic device; indexed addressing of data read from or written into random-access memory (RAM); data verification; the availability of words for controlling conversion of data read from or written into RAM. We note that the presence of a buffer makes it possible to combine data process and input, while the bucket-brigade principle allows m data to be processed simultaneously (where m is the number of stages in the brigade), thus providing rapid input. SP data input begins after servicing the interrupt request sent by the controller to the central processor through the communications device. In addition to the interrupt request, the central processor recognizes the operating status of the SP and its availability. These features make it possible for the SP diagnostic system to use its MUU [microprogrammed controller] with special microprogxams designed for micro- diagnostics. When this is done, the microprograms will be used as diagnostic instructions, allowing the central processor access to these microprograms and writing in RAM the SP responses to the input sets from RAM, producing an interrupt request. The constant addresses intended for predefined control or check information are often used to write the SP responses in RAM. These may be used continuously - for the nondiagnostic operating mode of the SP, e.g., writing the SP responses ' according to the ASK [expansion not given] address. Then the constant addresses are used in order to avoid using unchecked address formation schemes. In organizing diagnostics for arithmetic unit schemes using the bucket brigade data processing principle, data is taken from all stages of the brigade, maki.ng 44 FOR OFFICIAL USE ONLY ~ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY _ , ' r-f5-3----~ 17/6 __5 ~ �9 ~ - - u~ yc yy /dy~-f 1' Myy1 ~ I ~2 s ~ ~0 ~ ~a D/1 ~riA ~C Pl1MSl ~yK J 1 I 8 1f f3 ~ 12 q~i P!1 6n K /4 PP ~ Figure 1. [Key given in subsequent paragraph Tr.] ' large combination circuits making up the brigade checkable. ~ The SP systems testing mode can be augmented with microprograms which are used to test the SP autonomously when the central processor is busy. ~ When servicing an interrupt request, the central processor sends the SP a i ~ diagnostic instruction which enables a microtest microprogram which checks the unchecked part of the SP. Microprogram level, either pre-prepared data ~ needed for the test is read from RAM, or internal test micro-operations are i executed, after which there is an interrupt request which flags completion of i the microtest. After the interrupt requesr_ is serviced, the central processor ~ analyzes the response in order to localize trouble in the section checked by ' the microtest. Figure 1 shows the component parts of an SP and the diagnostic hardware, where ' the dotted line indicates the following devices which operate in the diagnostic mode: 1-- central processor; 2-- RAM; 3-- communications device; 4-- controller; S-- microprogrammed operand and control word controller; 6-- RAM address formation circuit; 7-- data verification formation circuit; 8-- central processor--SP data exchange format converter; 9-- microprogrammed operand processing controller; 10 flag register to coordinate work between the two microprogrammed controllers; ll register for transmitting operands to processor; 12 bucket-brigade processor; 13 operand and intermediate result buffer; 45 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY ~ 14 result register; 15 section of communications device which receives diagnostic instructions; 16 section of controller which allows SP to operate in diagnostic mode in case of malfunction; 17 section of cantroller which allows unformed addres~es to,be used for writing into RAM; 18 switching control for transmission of data from bucket brigade stages; 19 sectxon of microprogrammed operand processor used to store microprograms which control autonomous diagnostics for processor. Figure 2 shows the seqnence in which the SP diagnostics are executed: 1-- preparation of operands for diagnostic instructions and area of RAM where SP responses will be written; 1' "avail.able" SP state; 2-- execution of microtest microprogram; 2' CP awaits completion of microtest; 3-- analysis of SP response in order to localize trouble; 3' SP "interrupt" status. (1) (S) P~~ P~~oma ' 1 I~ Bano,uimwe Baoua ~ 3anycx,uuKponpcepa,vwe+~ 6 ~ ~ 2) mava~rou Koway a~~ dtrou,ncmuvsr~ro~l Ko,waHaa ( 3) An~aitu~ peaKUUUC/T ~ ~anp~c npepadaH~ot~ 7)' i O&.~yaru~cNUO t~1 cd�aadeH ( H) ( ) aanpa4 ! PI Figure 2. Key: 5, operation of SP 1. central processor 2, execution of diagnostic 6. enable diagnostic instruction instruction microprogram 3, analysis of SP response 7. interrupt request - 4. request servicing 8. SP free This is a two-level organization of the SP diagnostics. The actual diagnostic procedure and analysis of results are activated at the program level, while the procedure itself occurs at the microprogram level. It is also possible to obtain a two-level organization for the case in which - only the diagnostic procedure is activated at the program level, and the procedure and its analysis are activated at the microprogram level. Then the microprogrammed controller must be able to fetch both micro-instructions and constants (as is done in the "Nairi" machinea) and to compare particular bits of the constant used for the standard with the response being tested. With this microprogrammed controller design, trouble can be localized using the indication of the microinstruction address, after which the comparison is made and the 46 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 NUR OFFI('IAi. USE: ONI.Y operation of the microprogrammed controller stops if the standard and the response do not match. This sort of SP microdiagnostic organization is possible when the microprogrammed controller has enough memory to store the standards. In general, when special microprograms are used for SP microdiagnostics both with and without combination, the problem arises of limiting the amount of microprogrammed controller memory allocated for them. When special microprograms are used as diagnostic instructions, the diagnostics have the advantage that the SP is tested in real time without changing the machine cycle in either the central processor or the SP. Real time testing makes it possible to determine such time parameter deviations in the SP operation as the SP--central processor connect time after an interrupt request initiated by the SP is serviced. One example of detecting a violation of synchronization in the operation of two devices in the SP (in other words, the occurrence of a critical state) is the construction of a microprogram which tests the conditions which occur after these devices have operated. In general, the problem of constructing a diagnostic SP microprogram reduces to the problem of accommodating, in the limited microprogrammed controller memory, the diagnostic microprograms of individual SP sections which are diagnosed in a defined sequence (for example, using the "unspooling" method). The diagnostic microprogram designed to control the diagnosis of a particular , class of malfunctions in an individual section of the SP together with the hardware associated with this section must have, a self-diagnosing capability. Representing diagnostic microprograms with diagnostic instructions makes it possible to use them in sections which operate under the control of the diagnostic monitor, which makes it easier to control the section. This organization was used in developing the diagnostics for the SP which works in the YeS-1045 computer system. The format of the diagnostic instructions corresponds to the format of the channel instructions used in the Unified Computer System. In conclusion, we note the possibility of organizing a"closed" procedure (without operator intervention) for diagnosing a peripheral processor or peripheral devices. In doing thi.s, the computer must have an error bit which corresponds to the malfunctioning peripheral which, when the operating system sets a fault flag, fetches the program for diagnosing the malfunctioning device from external storage. z _ BIBLIOGRAPHY 1. Ramamoorthy, C.V., Chang, L.C. Modeling system and test procedures for microdiagnostic. IEEE Trans. Comput. C, 1972, 21, N 11, p. 1169-1183. 47 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFiCIAL USE ONLY 2. Bedeshenkov, V.A., Volkov, A.F. "Automation of designing digital computers with self-diagnosis of malfunctions." In "Diskretnye sistemy: Mezhdunar. simpoz." [Discrete Systems: International Symposium]. Riga, Izdatel'stvo "Zinatne", Vol 2, p 45-54. 3. Dulyaev, V.A., Sivachenko, P.M. "Principles of using microprogrammed testing and diagnosing in small digital computers." UPRAVLYAYUSHCHIYE SISTEMY I MASHINY, No 11, 1975, p 45-50. 4. Kagan, B.M., Mkrtumyan, I.B. "One approach to designing autodiagnostic systems for digital computers." VOPR. RADIOELEKTRONIKI. SER. EVT, No 8, 1971,,p 42-51. 5. Inagaki, M. Test an3 diagnosis program generation using microinstruction. NEC Res. and Develop., 1972, N 26, p 35-52. 6. Pat. 3798613 (USA). Controlling peripheral subsystem/G.H. Edstrom, E.P. Lutter. Publ. 19.03.74. ' ~ 7. Pat. 3882459 (USA). Deadracking sy:~tem/G.J. Burlon, D.R. Taylor. Pub. 06.05.75. 8. Pat. 3911402 (USA). Diaonostic block in a computing system/P. Lea. Pub. 07.10.75. 9. Lebedev, A.V., Makarov, V.A., Sorokin, G.K. "Special-purpose processor " for processing data files." In "Mnogoprotsessornye vychislitel'nye sistemy" [Multiprocessor Computing Systems]. Moscow, Izdatel'stvo "Nauka", 1975, p 100-109. 10. Gasparayan, L.Kh., Nalbandyan, Zh.S., Tadevosyan, V.A., et. al. "Structural features and functional characteristics of matrix processor in YeS 1045 computer." VOPR. RADIOELEKTRONIKI. SER. EVT, No 10, 1978, p 105-109. 11. Nowlin, R.W., Gustafson, D. A microprogrammed machine architecture for efficient matrix multiplication. In: Ninth Annul. Workshop on Micro- programming. Sept., 1976. New York, 1976, vol 7, no 3, p 56-61. 12. Lynch, W.C. How to stuff an array processor. In: Proc. Third Texas Conf. Comput. Systems, 1974. New York, 1974. - 13. Pat. 340611 (USA). Parzllel operations in a vector arithmetic computing system/A.D. F:ilkoff. Publ. 22.05.69. 48 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPR~VED F~R RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL iJSE ONLY 14. Pat. 3537074 (USA). Parallel operating array computer/R.A. Stokes, G.H. Barnes. Publ. 27.10.70. 15. Pat. 3541516 (USA). Vector arithmetic multiprocessor computing system/ D.N. Senzing. Publ. 27.10.70. 16. Pat. 3775753 (USA). Vector order computing system/W.D. Rastner. Publ. 27.11.73. COPYRIGHT: Izdatel'stvo "Naukova dumka", 1981 6900 _ CSO: 1863/64 -i ~ i ~ ~ ~ ~ i i . t 49 FOR OFFI('IAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USF. ONLY UDC 681.33 SYNTHESIS OF OPTIMAL DIGITAL-ANALOG REGULATOR FOR CONTROLLING THERMAL OBJECT Kiev GIBRIDNYYE VYCHISLITEL'NYYE MASHINY I KOMPLEKSY: LOKAL'NYYE ASU I USTROYSTVA VYCHISLITEL'NOY TEKHNIKI in Russian No 4, 1981 (signed to press 6 May 81, manuscript received 10 Jan 80) pp 99-102 [Article by G.V. Yevstratov from book "Hybrid Computers and Systems: Local Automatic Cont:ol Systems and Computer Devices", edited by G.Ye. Pukhov (editor-in-chief) et al., Tzdatel'stvo "Plaukova dumka", 1000 copies, _ 121 pages] [Text] The investigation of heat- and mass-transfer systems requzres solving analysis and synthesis probler~.~. In solving analysis problems, the stability and performance of the system in question must be determined. Since the main tasks of design are to select opt~mal object parameters (parametric optimization) and to synthesize a system for contiolling the object, the synthesis problem consists essentially of selecting the structural diagram of the system, its parameters and technical implementation such that the required control performance is provided. The methods used to control obje~ts der~~nd upon the available data. Since it is impossible ta obtain complete information on the status of heat- and mass- transfer Frocesses under actual conditions, the problem of synthesizing optimal control of these processes with incomplete measurement is also urgent when all types of limitations are manifested in the optimal process, in terms of both the output coordinates of the object and the control action. The. latter includes probleras involved in investigation and synthesis of thermostatting systems. ~ The Department of Automation and Telemechanics at Khar'kov Polytechnical Institute imeni V.I. Lenin lias been working for several years on synthesizing optimal control in`heat- and mass-transfer systems with incomplete measurement, and on solving direct and inverse thermal conductivity problems which are very close to optimization problems in three directions. The first direction ir~ solving these problems using RC-network models which implement a numerical method is universal. It can be used to model boundary problems of field theory in any statement for bodies with complex boundaries - with sufficient accuracy, and is in terms of many indicators competitive with - 50 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAI. 1JSE ONLY digital computers. Since the electrical time in RC-network integrators is many times smaller than the actual thermal time, comprising a fraction of a second, RC-models are automated. This favorable feature reduces the labor involved in reaching a solution. One shortcoming of this method is the increased complexity of the hybrid c omputer system and its reduced reliability as the requirements for solution accuracy increase. '~he sF~cond direction of solving such problems encompasses a narrower group of objects for which, allowing for thei r singularities and using several insignificant simpliEications and th e choice of a coordinate system, it is possible to find a mathematical mode 1 which describes that object with the required accuracy. Furthermore the mathematical model is sought in the form o,f a system of algebraic equations which is suitable for solution by computer using numerical methods, or which can be solved analytically. This approach was used in [1] for the optimal arrangement of heaters along the surface of a chamber, and for using them as the b asis for synthesizing a hybrid computer system for solving optimizal-ion prob lems and direct and inverse field theory problems; of the thermostat with ass igned limitations on the temperature drops - across the thermostat chamber [sic Tr.~. The third direction in solving optimal control problems is examined in [2,3]. A number of qualitative and quantitative criteria are used to select a solution method and means for its numerical implementation. As was shown in [4], the conclusion of the advant age of numerical methods should not impede the development of analytical methods, which have the undisputable advantage that they can be used to obtain a numeric al solution at a given point at a given instant. If an analytical solution can be implemented by computer, and if the accuracy, cost and time of the solut ion are comparable to the analogous values for a solution obtained numerically, the analytical solution is preferable.. Let us consider the synthesis of a d igital-analog system for solving optimal control problems in more detail. Figure 1 shows the functional diagra m of a hybrid computer system for investigating RP-obj~cts [expansion not given]. This setup includes the following basic elements: digital-an alog network model RC-network; boundary _ and initial condition assignment secr.ion; time interval formation section; digital control and readout device; data mea~;urement and display device and regulator model which, in turn, consists of th~~ regulator itself, a reference-input element, actuator, repeater and sensor. The structure and operation of sections (5), (7), (8), (9), (10) and (11) is described sufficiently completc>ly in (5,6,8] . - '1'he main element in the hybrid computer system is the passive RC-network, which - ~erves as the dynamic model-analog of the RP-object of control. This network - 51 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPR~VED F~R RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL IJSF. ONt,Y ( 1 ) 3adamvuK ~ 2 ~ P?zyaAmop ( j ~ 3odomvu,r y0y ( 6 ~ (9) , (3) Q (6) "0 6~BN (5) ( 2) Pezynamap NO cem o ~ 7) (7) (4) n a,:cM uuvy (10) (3) (4) yy Ky Z 8) . (8) (9) ~ 5~ 3~y yNON 3Hy ( 11) Q n Figure 1. Figure 2. Key: 1. reference-input element 1. reference-input element 2. regulator 2. regulator 3. sensor 3. IU [expansion not given] 4. repeater 4. correction device 5. boundary condition 5. actuator assignment section 6. data display device 6. actuator 7. RC-network 7� discrete-�analog network 8. sensor model 9. repeater 8. data measurement and display device 9. time interval formation unit 10. digital control and readout device - 11. initial condition assignment section integrates the equations which describe the dynamics of the thermal field in the area in question. As a result of integrating the original equation, the RC-network forms the current values of certain components of the state vector of the controlled object. For example, the RC-network defines the nonstationary thermal state of elements of the construction when entering the thermostat mode. The boundary conditior~ assignment section or actuator of the regulator produce the inputs currents and voltages applied to the RC-network; these serve as the analog of the controlling thermal action. 52 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY Let us consider a special case which is fairly,important an~? often encountered in applications: this is the case of optimal control, in t erms of speed, of an RP-ob.ject when the end state of the object is reached at the maximum value of the control action, and arbitrary limitations are placed on its output coordinates and on the control actions. This condition is characteristic for transient operating modes of most thermal objects. This c auses stabilization sectors to occur in the control which correspond to moveme nt of the controlled _ object along the boundary of the region of acceptable parameters. The control - has maximum value everywhere (except for special sectors), i.e., it is on the boundary of the region of acceptable states. This feature of this class of optimal control problems makes it possible to obtain solut ions using a hybrid ~ computer system with the proposed structure. In order to obtain a solution to the optimal problem, the operation of the system is organized as follows. If the state vector is 1 o cated within the S2y-domain of acceptable parameters, the regulator allows the maximum tolerable qontrol actions to be output to the RC-network. As soon as the boundary of the S~y-domain is reached, the regulator corrects the GU [e xpansion not given] such that the maximum values of the monitored parameters are not exceeded. Thus, the problem of the synthesized optimal regulator is to organize movement of the model-analog of the controlled object along the boundary of the region of acceptable values. In solving this problem, the regul ator and the DASP [expansion not given~ represent a closed system which kee ps the output coordinates of the controlled object at acceptable levels. Figure 2 shows the functional diagram of an optimal regul ator with manual optimal solution sear.ch. Manual optimal solution search in a closed control system assumes the presence of an operator. The person u sing the data display device to observe the transient process on the object kee ps the output coordinates of the object at acceptable levels. The operator does th is by means of the correction device, which acts upon the regulator for Tzap starting with output _ of pulse T1 (decision time) fram the digital control and display device (cf. Figure 1). For time TZap the regulator outputs maximum control action to the actuator and then to the object. After that, ordinary (e.g., propo rtional) regul ation begins, which optimizes the error. The operator's task is now to use the model with the help of the correction Sevice to find a duration TZap experimentally such that th e curve of the transient process is optimal. Tt?e actuator in a therm~statting system is a controllable reversible the rmopile in which, depending upon the magnitude and sign of the inp ut current, the thermopile isolates heat or cold with the appropriate magnitude [sic Tr.]. The electrical analog of the ~hermopile is a controllable reversible stable- current source operating into a grounded load (Figure 3). The stable current source uses a single JUT402A integrated circuit, which is an operational 53 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY amplifier with simultaneous negative and positive feedback. Its autput current depends upon the input voltage . ~ I = Uin n R2 . The regulator implements the following control principle:~ P- Pmax - cons*_, 0< t< Tzap~ p = ekpkt = var, t > '[zap~ . where t is time; TZap is delay time in connecting~negative feedback, i.e., the amount of time the thermopile operates at maximum power Pmax~ E is the regulation error; kp is the regulator gain; kt is the gain of the thermopile. Time TZap is measured by the IU (cf. Figure 2). Figure 4 shows the functional diagram of an optimal regulator which automatically finds an optimal solution. Here information about the state of the object is continuously i.nput to the limitation section (8). If even one of the state parameters of the object of control is too high, the limitation section acts upon the correction device with memory (7) which reduces TZap by one step. - During th~e next solution period T1 the process is repeated, continuing until the state parameters of the object exceed their bounds. Since an RC-network hybrid computer system is a r.epetitive-solution model, it can easily be used to implement this iterative method of solving these problems. The use of iterative methods eliminates dynamic error and increases the accuracy of the sought results. Research has indicated that the solution to a problem on a search-type device is obtained as the result of n iterations. However, the total solution time due to search does not increase significantly. This is because of the high speed of an RC-network hybrid computer system. In conclusion, we note that the structure of this hybrid computer system allows it to be used to solve more complicated control problems, including those which are characterized by the presence of several output coordinates of the object which have different simultaneous limitations. 54 FOR OFFICiAL USE ONLY ' APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR UNFICIAL USE ONLY (2) (3) (4) Nq, PetysA- y0 QC-cemxa mop 8 ~1) p` dOnW.f!!r ~yr~ ( 6 ) ~ ' (S) 2 Q _ N~r � I I (7) (8) n (6) fM 1 1 RN r,.. ~yn 60 Figure 3. Figure 4. Key: 1. reference-input element 2. regulator 3. actuator 4. RC-network ' S. sensor 6. repeater 7. correction device with memory 8. limitation section BIBLIOGRAPHY 1. Yevstratov, G.V., Prokof'yev, V.Ye. "Mathematical model of temperature field and optimization of placement of heaters along surface of thermostat chamber." In "Teplomassoobmen i modelirovaniye v energeticheskikh us~ranovkakh" [Heat- and Mass-Exchange and Modeling in Power Installations]. Tula, Tula Polytechnical Institute, 1979, Part 1, p 142-143. 2. Yevstratov, G.V., Prokof'yev, V.Ye. "Use of mathematical modeling methods to identify and optimize objects with distributed parameters." In "Teoriya i metody matematicheskogo modelirovaniya" ['~'heory and MPthods of Mathematical Modeling". Moscow, Izdatel'stvo "Nauka", 1978, p 110-111. 3. Biryukova, G.V., Yevstratov, G.V. "Approximation of Dynamic Characteristic of Object with Distributed Parameters on RC-Network Model." Present col.lection, page 3-6. 4. Kozdoba, L.A. "Resheniya nelineynykh zadach teploprovodnosti" [Solutions to nonlinear thermal conductivity problems]. Kiev, Izdatel'stvo "Naukova dumka", 1976. 55 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USF. ONLY 5. Azarov, G.N., Andriyevskiy, V.M., Garmash, V.V., et.al. "Special- ~ purpose device for automating investigation of thermostatic systems." In "Lokal'nye avtomatizirovannye systemy avtomatiki" [Local automated automation systems]. Kiev, Izdatel'stvo "Naukova dumka", 1978, p 10-18. 6. Positive decision on application No. 2504609/18-24 (085553) on patent (USSR) of 30 December 1977. "Device for modeling closed distributed control systems." G.N. Azarov, V.M. Andriyevskiy, V.V. Garmash et.al. 7. Aleksandrovskiy, N.M., Yegorov, S.B., Kuzin, R.Ye. "Adaptivnye sistemy avtomaticheskogo upravleniya slozhnymi tekhnologicheskimi protsessami" [Adaptive Automatic Control Systems for Complex Technological Processes]. Moscow, Izdatel'stvo "Energiya", 1973. 8. Yevstratov, G.V. "Time-interval formation section for digital-analog network processor." In "matematicheskoye modelirovaniye i gibridnaya vychislitel'naya tekhnika" [Mathematical Modeling and Hybrid Computer Techniques]. Kuybyshev, Kuybyshev 1'olytechnical Institute, 19'77. COPYRIGHT: Izdatel'stvo "Naukova dumka", 1981 6900 CSO: 1863~64 56 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 - FOR OFFICIAL USE ONLY OPTICAL PROCESSING HOLOGRAPHY AND OPTICAL PROCESSING OF INFORMATION: METHODS AND APPARATUS Leningrad GOLOGRAFIYA I OPTICHESKAYA OBRABOTKA INFOR1~fATSII: METODY I APPARATURA in Russian 1980 (signed to press 31 Dec BO) pp 2, 235 - 236 [Annotation and table of contents from the collection Holography and Optical Processing of Information: Methods and Apparatus", edited by V. G. Skrotskiy, B. G. Turukhano and N. Turukhano, Izdatel'stvo Leningradskogo institu~a yadernoy fiziki, 500 copies, 237 pages] [TextJ The 12th Al1-Union School on Holography was held according to the plan of the Council on Holography attached to the Presidium of the USSR Academy of Sciences and was organized by the Moscow Order of the Red Banner of Labor Physicotechnical Institute. The following institutions participated in preparation and publication of the proceedings of the 12th All-Union School on Holography: the Leninqrad Institute of Nuclear Physics imeni B. P. K,onstantinov, USSR Academy of Sciences, the Nbscow Order of the Red Banner of Labor Physicotechnical Institute and the Order of Lenin Physicotechnical Institute imeni A. F. Ioffe, USSR Academy of Sciences. The 12th All-Union School on Holography was held at Pasanauri. Leading Soviet scientists and engineers assembled at the All-Union School to dis- cuss the latest advances in holography, quantum electronics and recording media for recording holograms. The proceedings of the 12th All-Union School on Holography are of interest for a wide range of scientists and engineers specializing in the field of holography. The organizing committee of the school is grateful to the participants who read such interesting reports and presented them for publication in this collection. Contents P~ge 1. N. B. Baranova, B. Ya. Zel'dovich, V. V. Shkunov and T. V. Yakov'leva, Theory of Restoring Fields by Three-Dimensional Holograms and Spectral-Angular Distortions 3-38 57 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 2. A. I. Sokolovskaya, Recording, Restoration (Conversion) of the Li~ht Wave Front and Self-Focusing--New Effects in Stimulated Raman Scattering of Light 39- 55 3. K. G. Predko and V. G. Senchenko, Information Characteristics of Fresnel Holograms W'hen Producing Images Through Scattering Media 56- 64 4. V. B. Nemtinov, The Structure and Quality of the Holographic Process 65- 85 5. V. A. Zubov and A. V. Krayskiy, Recording and Processing of Modulated Optical Signals 86- 92 6. G. R Lokshin, The Principles of Correlation Filtration in ~ Holography 93-104 � 7. V. A. Soyfer, M. A. Golub and A. G. IQiramov, The Possibe and Impossible in Digital Itolography 105-123 8. Yu. I. Ostrovskiy and N. V. Morozov, Holographic Interferometry . of Moving Objects 124-142 9. N. G. Vlasov, S. G. Gal'kin, Yu. P. Presnyakov and B. M. Stepanov, Elimination of Reflective Noise in Interferometry of Diffuse- Reflecting Objects 143-149 10. A. V. Zuyevich, V. V. Alekseyenko and V. B. Govrushin, Holo- graphic Visualization of Underground Objects 150-155 11. K. B. Gendovich and K. S. Stoyanova-Pushkarova, Processing Seismic Information with a Coherent Optical System 156-163 12. ~~A. A. Ras~sokha and V. Ya. Antsibor, Holoqraphic and Reflective Interferometry in Some Problems of Rock Mechanics 164-167 13. Yu. A. Zurukhin, Visualization of Acoustic Objects Based on Colinear and Bragg Light Scattering on Elastic Waves in C�rystals 168-175 14. A. V. Zuyevich, Using Long-Wave Holographic Systems to Visualize Seismic Images 176-187 15. Ye. I. Shterkov, Dynamic Echo-Holography 188-203 16. A. A. Rassokha, Holographic Diagnosis of Macroinhomogeneous Solids 204-211 17. V. V. Alekseyenko, A. A. Bovin and A. V. Zuyevhich, Some Results of Optical Processing of Geological-Geophysical Data 212-218 18. B. V. Feduleyev, V. P. Ryabukho and V. B. Rabkin, The Possibilities of Measuring the Temperature Coefficient of Linear Expansion of . Anisotropic and Isotropic Materials by the Holographic Interferom- etry Method 219-228 58 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONI.Y 19. A. P. Golikov, M. L. Gurari and S. I. Pritkov, Holographic Method of Monitoring Reflectors 229-230 20. N. A. Valyus, Beam Holography 231-234 COPYRIGHT: LIYaF, 1980 6521 cso: 1863/26 59 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFIC[AL USE ONLY UDC 778.38 RECORDING AND QROCESSING QF MODULATED OPTICAL SIGNALS Leningrad GOLOGRAFIYA I OPTICHESKAYA OBRABUTKA INFORMATSII: METODY I APPARATURA in Russian 1980 (signed to press 31 Dec 80) pp 86-92 [Article by V. A. Zubov and A. V. Krayskiy from the collection "Holography and Op- tical Processing of Information: Methods and Apparatus", edited by V. G. Skrod- _ skiy, B. G. Turukhano and N. Turukhano, Izdatel'stvo Leningradskogo instituta yadernoy fiziki, 500 copies, 237 pages) [Text] The use of a transient-rsference wave with linear variation - of frequency through the cross-section permits one to achieve to- tal recording and processing of a modulated optical signal. These problems were considered theoretically for monochromatic radiation in [1, 2]. F~cperiments on recording this signal are considered in this paper and the operation of schemes that permit one to find the structure of the signal and its spectrum with resolution con- siderably exceeding the width of the laser spectrum and which reaches fractions of a hertz are considered in this paper. A diagram of a two-channel experimentzl installation is presented in Figure 1. Laser emissior~ after diaphragm D1, which limits the homogeneous section of the wave, is divided into two parts by means of a prism PR. A transient wave having the following form in the recording plane is shaped in one channel formed by dia- phragms D2 and D3 and the telescopic system L1 and L2 with quasi-linear motion of the components of L1 [3~: ~~,cK~~~-~. ~Kexp[ ~wKt+~aK~1,,.*.~~..~~j . ~i~ . , y where K is the number of the specific radiation mode, the total number of modes i N, sak is amplitude, wk is the mean frequency and ak(t) is the random phase of this mode. The second channel formed by the telescopic system L3, L4 and mirror ~ is the working channel. The signal is modulated in the mean �ocal plane of this telescopic system. The effect of the object is modulated by an interrupter disk M in the experimental circuit. Part of the emission of the working channel is de- flected by a semitransparent plate to photodetector FEU and is further recorded on an oscillograph OSTs so as to have information about the ampl=.tude modulation of 60 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAL USE ONLY the signal being recorded. The radiation from the working channel impinges on the recording plane xy and the direction of propagation is described by a cosine cos6XS. The field from the object has the form: N E,IX~t) 4 ~dsFo~t)Qxp[-i.w,t+i,a8(t)+I,~xcosAx9, , (2) where s is the number of the individual laser emission mode and ep{t) is the com- plex amplitude that describes signal modulation. If ordinary conditions are ful- filled, a hologram with amplitcde transparency consisting of two components is produced: a component related to a constant glow from tw~o channels and a component with information about the signal. This hologram can be exposed within a limited time T since optical system L1 and L2 should operate in paraxial approximation when the reference wave is formed. As a result of averaging the information, only terms with s= k make a nonzero contribution during recording. As a result we have '?/2 ttxl~f Fot+l~ Eo~E~~xPL~c'"xeose,~-~~xt,d{ + c3~ -TIZ K:r . + complex integ~atior~ OCu 3 - ~13 M m3y xy Q4 A~ 1 nP e~ lV 4 ( T T A2 . n~ n2 1 A3 . ~ T ~ ~ '~1 � ~ � L Figure 1. Diagram of Experimental Installation for Recording Modulated Optical Signals Upon restoration the hologram G is illuminated by a plane laser emission wave (Fig- ure 2) consisting of M modes and the mean direction of propagation is given by a .cosine--cos~xm Ep~X~~,~ RL'EP~exP[-l.~JrnE+~dm~t~-~ a 1cCo60ym, r (4) where m is the number of the mode. All calculations were ma.de in Fresnel approximation. 61 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY xy ~ uV ~ A4 e~P ~ tN~ ~~x~t~ JI8 ~ ~ - u, ~ ~ ,M '~s . � . Figure 2. Diagram of Restoring the Structure of a Modula~ed Signal Optical system LS performs the operation of Fourier transformation in the restora- tion phase. Let us take into account that wm - ~ K Wm~ wk ar?d the direction cosines cari be selected so that Icos9~~ = IcosO~~; then for a hologram with di- mensions equal to 2xp along the x axis, the spread function has a simple form: ~~x~c \~'t Tt ~ (5) . With regard to the final resolution of the field component that produces the pat- tern of the recorded signal in the rear focal plane W of the optical system, there will be ~ ~'u~{,~ ~r~~~~tP[-~w~�~ttl.dm~t)~ Eo #eV) ' (6) i.e., the recorded modulated field is restored in three-dimensional representation. Let us determine resolution by the distance between the zeros of apparatus func- tion (5), which yields ~21 ~ ~ - ~t-WK K.~ N~ e ' ~u - w~~t N xo . c7 ) The position of the restored image is ~~iven by the maximum spread function for mo- ments of time t=+ T/2 and its dimensions and number of resolved components are equal to 4ud'~B~T~~3 ~ n=Tf~t=2u.~b'u � (8) z The experiment was performed with the following values: 71 = n.63�10'3 mm, T= 6.8 seconds, v= 2 mm/s, f2 = 300 mm, f5 = 300 mm and 2xp = 8 mm. Calculation yields: ~cos6kr~ < 0.023, 2 up = 14 mm, dt = 25 ms, u= 0.05 mm and n= 280. The exper- imental results for different modulation frequencies and different on-off time ratios are presented in Figure 3. ~ ~ 62 FOR OFFICIAI. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 _ FOR OFF[CIAL USE ONLY The scheme for finding themodulated signal spectrum F(SZ) using steady restoring wave (4) is presented in Figure 4. The field in observation plane xy, optically conjugate with the hologram G, based on (3) within the conditions used in deriva- tion of (5) , has the form r~~ . _ ~'~~t~"'~,tRPt-~u1r�~.~tta�(t~~JF,(~')pXp~l Cx~-Xt10~~~^~~. tM . JAt J . ~ axpC IWM~ +1,6;~,(t~j P .V 1 � (9) e 1' . � where the tilda sign corresponds to convolution of the signal spectrum with the spread function. For three-dimensional resolution in the recording plane, based on (9) and (8), we find - ~ " 2~]~ (10 ) $ X - u, ~SyT ' To estimate the spectral resolution, let us make use of the concept ep(t') in th~ form of a Fourier transform F(St), which yields the spread function s u, ~C~ ( Q+ P-- i--~ x), . c An estimate of the width of the spread function is made similar to (7) and yield� SS2 = 4 n/T. It should be noted that from the viewpoint of spectral resolution it is more correct to proceed fr~m the Rayleigh criterion, i.e., b~ = d~/2 = 2~r/T or, converting from circular to orginary frequencies ~Yp a I/~ 3T1Ci VXP ~J2i1 /P,vT~ ~12~ Thus, spectral resolution is determined by the total recording time. The disper- - sion range is determined by the frequency range recorded on the hologram, which is equal to A W a- 2X0 ~r A~' ~Tl ~ ~ ~ ~ ~13~ f T Dispersion in the recording plane (see (12)) and the number of resolved components (see (13) and (12)) are equal to ~,PI~Xp'~P~I~P,~~ ~ R=aJ/~JP=P~x,T . (i4) The linear dimensions of the spectrum �re determined by the maximum spread func- tion for three-dimensional resolution at x=+ x0, which yields aXe p 2 Xe~Ip, . (~s) 63 - FOR OFFiCIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/42/09: CIA-RDP82-40850R000500430018-4 FOR OFFICIAL USE ONLY with regard to the values used in the experiment (Pl = 125 m�n. P2 = 260 mm and f5 = 85 mm), we find the following estimates: SSZP = 0.9 s-1, SvP = 0.15 Hz, dXP = 0.03 mm, pv = 85 Hz~ Svp/aXp = 5 Hz/mm, 2Xp = 16.5 mm and n= 550. . ~l y . uv . ! xy ' Ati 118 . t~(#. _ EP(x.~ . . ~ ~ . . ~ ~ ~x� . . . * Figure 4. Diagram for Finding the Spectrum of a Modulated Optical Signal The experimental restuls are presented in Figure 3. As was noted, the effect of an object on the signal was modulated by modulation at frequency of 1.03 and 2.06 Hz with different on-off time ratios. Calculation of the modulation frequency from the spectrograms yields 1.0 and 2.0 Hz. Illustrations show that the real resolu- tion considerably exceeds 1 Hz. BIBLIOGRAPHY 1. Zubov, V. A., A. V. Krayskiy ar~d T. I. Kuznetsova,;PIS'MA Z:~ETF, Vol 13, 1971. 2. Zubov, V. A., KRATKIYE SOOBSHCF~NIYA PO FIZII~, No 7, 1972. * 3. Borkova, V. N. and V. A. Zubov, Proceedings of a Seminar "New Developments in the Field of Optical Holoqraphy and Their Industrial Use," 21 pages, Leningrad, Izd-vo LDNTP, 1979. COPYRIGHT: LIYaF, 1980 6521 - CSO: 1863/26 64 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY UDC 550.834 HOLOGRAPHIC VISUALIZATION OF UNDERGROUND OBJECTS Leningrad GOLOGRAFIYA I OPTICHESKAYA OBRABOTKA INFORMATSII: METODY I APPARATURA in Russian 1980 (signed to press 31 Dec 80) pp 150-155 [Article by A. V. Zuyevich, V, V. Alekseyenko and V. S. Gaveyushin from the collec- tion "Holography and Opti.cal Processing of Information: Me~hods and Apparatus", edited by V. G. Skrotskiy, B. G. Turukhano and N. Turukhano, Izdatel'stvo Lenin- gradskogo instituta yadernoy fiziki, 500 copies, 237 pages] _ [Text] The results of experiments on holographic visualization of underground test objects are reported. The experimental installa- tion is described. Holograms and restored images of test objects are presented. The design features of a marine holographic sys- tem are discussed. The past decade has been characterize3 by active development of the continental , shelf of the seas, with which the main hopes for solving the problem of expanding ~ the country's raw materials base are now linked. Development of the shelf assumes solution of a number of problems of both an exploratory and commercial nature. One of these problems is detailed study of the upper part of the sedimentary mantle. The need for this is related to problems of constructing stationary drilling plat- forms, laying oil and gas pipelines and construction of underwater storage tanks. Moreover, solving the problem of search for and identification of objects at the bottom require development of ineans of visualizing seismoacoustic images. - ~ne of the most promising directions in solving the indicated problems are holo- graphic methods. ?1n experimental check of long-wave holography systems has been made quite r~cently, but mainly under laboratory conditions [1-3] . It was feas- ible in this regard to check the capabiliti~s of holographic systems under condi- tions approaching real conditions. This check was made in a test area on an artificial reservoir. The test objects were located on the bot~om or under the ~ bottom of the reservoir. The emittP r was attached to a raft and was made immo- bile during the experiment. The em ~~:ter itself was assembled on the basis of a ~ piezoelectric cylinder (working fxequency f- 10 kHz) and was equipped with a shaper to create a main lobe r~.diation pattern of approximately 30�. Radio pulse emission with pulse length of 1.0-2.0 ms was used. The framework together with - ~he receiving antenna and part of the processing apparatus was moved along rails laid along the basin. The receiving antenna contained 12S chann~ls and the syn- thesized (linear) dimension was approximately 42a. The channels were interrogated 65 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFF'[CIAL USE ONLY - during movement of the antennas at a frequency of 250 kHz along the antenna and 50 Hz in the direction of motion. The interrogation signal was fed by cable to the processing system located on shore. The reference signal was simulated in an electronic channel and the radiation pulse parameters and readout mode were con- trolled from the control console of the processing apparatus. The information to be recorded was stored on the target of a ZELT [Shielded cathode-ray tubel� In this case movement of the line along the frame was synchronized with the movement of the antenna within the aperture to be synthesized. A hologram which was subse- quently visualized on the ELT [Cathode-ray tube] screen was formed on the target - upon completion of the synthesis cycle. The hologram was then photographed from a distance of 1.5-2 meters for the appropriate reduction, processed and the reduced copy measuring from 0.6 X 0.6 to 1.0 mm2 ~as restored in the light of a He-Ne laser. The restored image was observed in a microscopic system. The symbols "S" and "H", made of sheet steel 1 cm thick, and also a pyramid of marble cubes were used as the test objects. The dimensions of the "S" and "H" symk~ols were 2.0 X 1.5 m2 (13 X 10 a2) (the width of the letters was 0.45 m= 3~), while the side of the marble cube was 0.4 m or approximately 3 a. The "S" and "H" objects were located under the bottom one over the other so that the symbol "H" was at a depth of 3 meters whii.e symbol "S" was 2 meters above it and 1 meter from the bottom. The vertex of the marble pyramid was located 1 meter from the surface. The shape of the objects, the recorded holograms and the restored images are pre- sented in Figure 1. The images are quite acceptable with regard to the complexity of the medium and the smallness of the parts of the object, although not of very high quality, which makes it possible to identify the objects with adequate confidence. The experiments showed tnat these systems permit one tn solve the problem of ~ search and identification of objects in soils under marine conditions. This made it possible to turn to developing a marine version of the holographic complex. The complex is designed to solve problems of investigating the bottom and the bot- ' tom sedimentary mass or rather its first few tens of ineters under conditions of the continental shelf at water depth of 100-150 meters. In this case the given complex permits one to visualize (i.e., to obtain visual images of) objects of different nature located in the bottom sedimentary mass. - The marine holographic complex now developed includes an aggregate of apparatus assemblies and technical means of registration, recording, processing and visual- ization of the holographic information on board a ship under autonomous navigation. The complex consists of a ship on which the information recording, processing and visualization apparatus is located and a towed device. The towed device is de- , ~ signed to study the sounding signals, to record the signals reflected by the ob- - jects by the two-dimensional aperture synthesizing method and preliminary proces- sing of the received signals. - Electronic apparatus that records and processes the information is located on the - ship, while the final result of proceseing is an acoustic hologram visualized on , 66 FUR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FUR UFFICIAL USE ONLY an ELT screen. A visualized hologram photorecording unit and device for optical _ restoration of holograms with a He-Ne laser are located on the same ship. More- over, a sounding signal power amplifier is located on the ship, while the ship itself is equipped with a velocity measuring device required for proper recording of holograms. The latter condition is related to the need for strict geometric conformity between the position oL the antenna within the aperture to be synthe- sized and the position of the information to be imaged on the EL"T screen within the frame. The towed device of the complex consists of a receiving antenna of the preliminary processing unit of the damper-sinking emitter of the equipment system designed for towing and loading-unloading by the towed system. The receiving antenna is a hydroacoustic device having the shape of a wing in cross-section, which provides the required hydrodynamics during towing. The an- - tenna consists of preassembled sections, each of which contains rudder-stabilizers, wh~.le the outer sections are also equipped with rudders-sinking units. A total of 128 hydrophones connected to amplifiers are located on the bottom of the antenna in its middle part. The leads of the amplifiers are connected by cable to the preliminary processing unit. ~ The preliminary processing unit is made in the form of a cigar-shaped gondola and contains the electronic assemblies (phase detectors and decoder) for processing the signals coming from the receiving components by cable. - The emitter is designed to irradiate the objects of investigation. The emitter itself is made on the basis of cylindrical piezoceramics and is equipped with a - radiation pattern shaper. The skin of the emitter is made from an acoustically . strong material and mainly performs the function of providing the hydrodynamics required for towing. Both gondolas are firmly connected to each other and are at- tached to tne antenna by two metal rods. The antenna together with the gondolas has low positive buo;:~ncy. The outboard part of the system is towed by a tow cable to which the ciamper-sinking unit is attach~_d, that ensures towing safety with sharp jerks of the rope. The cable is used to supply power to the oetboard electronic units, to control the operation of these units, to gather information and also to excite the emitter. The equipment system includes a swivel, lowering- raising system and a spring hook. The ~arameters of submerging the antenna and stabilization of its course are varied manually in the given version, while the parameters themselves are selected empirically. Foam plastic floats designed to stabilize the depth of submersion are ~ttached to halyards on the edges of the antenna. Tests of the running qualities of the outboard part of the complex, conducted in 1979 in a basin of the Sea of Azov, showed its complete suitability for operating under maritime conditions by the long-wave holography method. The complete com- plex will be tested in the near future. These experiments make it possible to de- - termine the capabilities of the long-wave holography method in solving problems of marine geology and geophysics and consequently to determine the place of holo- graphic systems in solving problems of exploiting the mineral resources of the world ocean. 67 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY BIBLIOGRAPHY 1. Meterel', L. F., "The Comparative Importance of Phase and Amplitude in Acoustic Holography," "Akusticheskaya golografiya" [Acoustic Holography], translated from EnglisY~, Leningrad, Sudostroyeniye, 1975. 2. Farr, J., Experiments in Rcoustic Holography Using Computers," "Akusticheskaya golografiya", Translated from English, Leningrad, Sudostroyeniye, 1975. 3. Zuyevich, L. V., V. V. Alekseyenko and V. M. Sugak, "Recording and Restora;ion of Long-Wave Holograms in Small Apertures," PIS'MA V ZHTF, Vo~. 4, No 6, 1~~78. COPYRIGHT: LIYaF, 1980 6521 - CSO: 1863/26 68 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY _ UDC 535 - PROCESSING SEISMIC INFORMATION WITH COHERENT OPTICAL SYSTEM Leningrad GOLOGRAFIYA I OPT7CHESKAYA OBRABOTKA INFORMATSII: METODY I APPARATURA in 13issian 1980 (signed to ~~ress 31 Dec 80) pp 156-163 [Article by K. B. Gendovich and K. S. Stoyanova-Pushkarova, Peoples Republic of Bulgaria, from the collection "Holography and Optical Processing of Information: Methods and Apparatus", edited by V. G. Skrotskiy, B. G. Turukhano and N. Turu- ]cl~ano, Izdatel'stvo Leningradskogo instit~ta yadernoy fiziki, 500 copies, 237 paqes] ~ [TextJ Some experimental applications of coherent optics and holog- raphy methods are considered for processing seismic information. A coherent opt.ical-electronic system w~s used for spectral-correla- tion proce~sing of seismic time profiles containing a large volume of information (approximately 10~ bits). An autocorrelogram and autocorrelation functions were found for the root or qroup of roots of the seismic pzofile selected by the interpreter. The effect of attenuation of short waves was estimated by the "24-fold addition" method and the unified statistical characteristics of the factors that determine the capability of separating single waves to investi- gate the seismic profile was found.* Problems of search and study of mineral depostts, solution of which is impossible without the participation of computers, have continuously become complicated during the past few years. Most computers that are used, distinguished by low internal storage and relatively low speed, are incapable of operatianally processing the continuously increasing volume of incoming information. OiTie of the methods of progress in the field of processing seismic information is to coherent-optical computers characterized by high speed, economy and flexibility in variation of the prngram-algorithmi.c complex [1]. The purpose of the experiments is to obtain autocorrelograms of the seismic pro- file and also the autocorrelation functions of its individual sections. The investigation was conducted by the tntal deep point method (MOGT) from a seis- mic profile of the Black Sea hasin. The seismic profile was processed by the 1 * ~ The report was made at the llth All-Union School on Holography. . 69 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFIC[AL USE~ ONLY "24-fold addition" algorithm to reduce the interfering effect of reverberation caused by the water layer and the effect of multiple reflections produced from - some reflecting boundaries in the geological structure of the region. As a resul�t a time (seismic) profile consisting of 384 tracks (channels) with four seconds of recording of each channel represented graphically by the variable width method was obtained. The length of the seismic profile in the X direction on the profile was 9.6 km. Interfering multiple waves create imaginary reflecting boundari~s on the seismic profile, i.e., on the time image of the geological profile, which make difficult both correct running of the following algorithm for digital machine processing and clear interpretation of the results. The effectiveness of_ the mentioned algorithm can be determined directly from the autocorrelogram and autocorrelation functions of individual sections but the type and number of the reverberation systems, their attenuation factors, tectonic dis- turbances and structural camplications in the investigated region are determined fron the ratio of amplitudes of the secondary maximums and the main maximum of the autocorrelation function. ~'he optical-electronic system used is shown in Figure 1, where 1 is a laser, 2 is a filter-collimating unit, 3 is microscopy of the seismic profile, 4 is an astig- matic lens, 5 is a semitransparent mirror, 6 is a television camera, 7 is a monitor, 8 is a storage digital oscilloscope, 9 is a microcomputer and 10 is a photorecorder. 1 i ~ i 5 ~ ~ ~ ~ . 6 ' g 8 7 O 0 ~ ~ Figure 1. Aptical-Electronic System for Producing Autocorrelogram and Auto- correlation Functions: 1--laser; 2--filter-collimating unit; 3-- microscopy of seismic profile; 4--astigmatic lens; 5-semitrans- ~ parent mirror; 6--television camera; 7--monitor; 8--storage oscil- loscope; 9--microcompu~er; 10--photorecorder The original seismic profile is reduced by a factor of 18.7. The correct position of the components in the optical system are monitored by means of a special scale built into the pattern of the seismic profile prior to microfilming. The square of the modulus of the microscopic ~pectrum is recorded in the spectral plane of the optical system (2l. The positive microfilm produced in the next phase _ is installed at the input of the system at the microscopic position. The desired autocorrelogram of the seismic profile is found in the spectral plane of the opti- cal-electronic system. The spectrum and autocorr~iogram are shown in Figures 2 and 3, respectively. e~n image of the autocorrelogram can be observed on the monitor by means of the television camera. 5ome changes were made in the camera to obtain _ 70 FOR OFFI('[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY autocorrelation functions so that the light �ield impinging on the camera tube had the capability of scanning in sequential lines. The digital delay unit of the storage oscilloscope was used to select the camera tube scanning line. The video signals obtained from line scanning are converted to digital form and are recorded in the memory of the microcomputer and in this case the scanning aperture can be formed with sequential addition of adjacent television scanning lines. The auto- correlation functions of the track or group of tracks of the seismic profile se- lected by the interpreter are observed on the oscilloscope screen (Figure 4). The experimental method of obtaining the autocorrelogram excludes the necessity of precise adjustment of the filter-hologram in the Van der Luchta method, as does precise setting of the i.nput object and filter in the front focal planes of the first and second lenses, respectively. At the same time, a one-dimensional spec- trum of the seismic profile is obtained as an intermediate result, which creates specific conveniences for interpretation of the seismic wave field. The inconven- ience of this method, like the Van der Luchta, is the two-step process of producing _ the autocorrelogram. Initial experiments devoted to investig.iting the possibilities of using the optical medium to record a layer of Ag-As2S3 semiconducting metal produced by evaporation without physicochemical processing were initially conducted to eliminate the in- conveniences of the two-step process. The experiment was conducted by the scheme shown in Figure 5, where 1 is a laser, 2 is a semitransparent mirror, 3 and 4 are a filter-collimating unit, 5 is micxoscopy of the seismic profile, 6 and 7 are an astigmatic lens, 8 is the medium and 9 is a mirror. A wavelength of = 514.5 nm was used for holographic recording while the carrier frequency of recording was equal to 1,000 lines/mm. The spectra are multiplied after the reference wave has - been eliminated and an autocorrelogram of the input object is observed in the ou~- -put plane [3J . - : 2.3~ 5 6 b ~,I ~S(x,t) ~ S;"~`; : ~ ~ ~ , ~ ~ 4 13 12 11 ~ . _ ~ ~ 0 ~ ~ 9 . , � ~ Figure 5. Optical-Electronic System of Multichannel Holographic Correlator: 1--laser; 2--semitransparent mirror; 3 and 4--filter-collimat~ng ~ unit; 5--microscopy of seismic profile; 6 and 7--astigmatic lens; 8--medium; 9--mirror; 10--television camera; 11--monitor; 12-- storage oscilloscope; 1.3--microcomputer Overcoming the two-step process of finding the autocorrelogram and autocorrelation functions of the seismic profile creates real prerequisites for developing a more 71 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY co~nplex multichannel optical-electronic device which can be included in the total cycle of machine processing of seismic information: in an optical spectral analyz- er (OSA~--filtration by frequency, by apparent velocities (indirection) and by wave number, and in a correlator--finding the autocorrelogram and autocorrelation func- tions. This system permits one to realize different combinations of filtration - parameters with relatively low expenditures of time and funds and estimation of the effectiveness of these parameters. One has i.n mind in this case that digital fil- tration of such a large volume of information occupies much machine time, while optimization of the operator for filtration makes processing more expensive as a whole. The purpose of making a spectral-correlation analysis of the results of "24-fold addition" is to estimate the effectiveness of suppressing multiple seismic waves wh i ch m ake difficult clear geophysical interpretation of the investigated pro- file. The derived autocorrelogram shows the presence of intensive multiple waves with a delay with respect to the multiple-generatrix in the range T= 0.02-0.2 second. Autocorrelation functions (Figure 4) which characterize the presence of different types of reverberation systems were found on some segments of the auto- correlogram. Accurate and complete interpretation of the results requires a knowl- edge of additional qeophysical and geological information and is the object of other investigations. The effectiveness of attenuation of multiple waves by "24-fold addition" (coefficient K) was estimated and the unified statistical char- acteristic of factars was found that determine the possibility of separating single waves from interfering multiple waves (coefficient P): - . AO where Ai is the amplitude of auxiliary maximuir,s and Ap is the amplitude of the main ma ximum; p.~ ~~~L , ~ 3 where T1 is the distance along the time axis between the first lateral maximum and the main maximum, TZ is the distance between the main maximum and the lateral max- imum of the greatest amplitude, T3 is the width of the main maximum, m is the num- ber of auxiliary maximums and A is the total amplitude. Data for coefficients K and P are presented in Table 1. The results show that "24-fold addition" is sufficiently effective for sections II and III (Figure 4), but it is not sufficiently effective for some multiple waves (K > 0.1). It is ob- vious from the given values of P that favorable sections for separation of single waves are III and V(0.9 < P< 2), while unfavorable sections are all the remaining sections (P < 0.9). The results of the experiments, like their geophysical interpretation, show the possibility of including the descrihed analysis in the entire cycle of processing _ and interpretation of seismic infornlation. Development of an optical-electronic spectral analyzer-correlator, which will simultaneously accomplish several process- ing algorithms, is especially promising. This device may also find application in 72 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/42/09: CIA-RDP82-40850R000500430018-4 FOR OFFICIAL USE ONLY Table l. _ a~ A , ItI 1V V VI 1.� Q,]~, 0.075 0,086 Q,~Q~ 016 1Q._;~_2 2. 0.048 0,083 0.021 0.206 01~05 Q,086 3. 0.06G ~ 0,049 0 032 0,057 4. 0,02L� � 0,039 0,067 5. O.OG9 0.200 6. . 0.020 4038 7. ~ . 0,029 ~ 0,029 g. O.OZO 0,019 g. . 0.019 P 0,490 0,490 2,01 0,070 1,08 0,230 in other multichannel geophysical investigations (for example, aeromagnetic inves- tigations and aerogravitational investigations) which are related to processing and interpretation of large files of geophysical information. BIBLIOGRAPHY l. Potapov, O. A., "Opticheskaya obrabotka geofizicheskoy i geologicheskoy infor- matsii" [Optical Processing of Geopnysical and Geological Informati~n], Moscow, Nedra, 1977. 2. Fontanell, A. and G. Grau, "Correlation Optique en Lumiere Coherente," GEOPHYS- ICAL PROSPECTING, No 1, 1971 . ~ 3. Goodmano J., "Vvedeniye v fur'ye-optiku" [Introduction to Fourier Optics), Moscow, Mir, 1970. COPYRIGHT: LIYaF, 1980 6521 CSO: 1863/26 73 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02/09: CIA-RDP82-44850R444544434418-4 FOR OFFiCIAL USE ONLY UDC 535 HOLOGRAPHIC METHOD OF CHECKING REFLECT~~RS Leningrad GOLOGRAFIYA I OPTICHESKAYA OI~RABOTKA INFORMATSII: METODY I APPARATURA in Russian 1980 (signed to press 31 Dec~ 80) pp 229-230 _ [Article by A. P. Golikov, M. L. Gurar~. and S. I. Prytkov from the collection "Holography and Optical Processing of :nformation: Methods and Apparatus", edited by v. G. Skrotskiy, B. G. Turukhano an~i N. Turukhano, Izdatel'stvo Leningradskogo instituta yadernoy fiziki, 500 copies, :'.37 pages] [TextJ The development of a r~ew holographic method to measure the radii of curvature and local ::urface distortions of large mirrors are reported [1]. Serial production of ~large-diameter corcave mirrors and especially of convex mir- rors is now weakly supported with checking equipment. The complexity of developing nonaberration wave fronts of large aperture leads to an increase in the errors of interference methods [2]. Methods that do not require special reference components --shift interferometry and interferometry with a scatterer--are applicable to anal- ysis of only convergent radiation beams. It is significant for these methods that the base of the measuring installation increases in proportion to the radius of curvature of the mirror being checked, which in the case of concave mirrors of small curvature makes the installation cumbersome and leads to a reduction of measurement accuracy due to an increase of interference and in the case of canvex mirrors requires the use of large-size compensators. A schematic diagram of the experimental i.nstallation is presented in Figure 1.J The mirror to be investigated is illuminated by a diverging flux of LG-38 laser emis- sion through a scatterer arrar:ged tightly against the mirror. The reflected flux through this scatterer impinges on a photographic plate and its hologram through which the surface of the scatterer is subsequently observed is recorded on it. A shift of the mirror leads to formation of an interferogram localized on the scat- terer. The geometric parameters of the surface of the mirror being checked can be determined by the shape of the interferogram. An increase of ineasurement accuracy is achieved by introducing compensatory rota- tion of the mirror in addition to shifting of. it. The interferogram is reduced to standard form corresponding to rotation of the mirror around its center of curva- ture by the combination of shift and rotation. The radius of curvature ~s found 74 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFiCIAL USE ONLY by the measured values of shift and rotation, while the distortion of the mirror surface is found by the shape of ths bands of the interferogram. ~ / ~ ~ - Figure l. Schematic Di~igram of Experimental Installation The holographic measuring installation with scatterer installed in front of the mirror can measure the radii of curvature of mirrors from 200 mm and above in di- ameter with an accuracy on the order of 0.1 percent (for radii of curvature in the range from 10 to 50 meters). The overall dimensions of the installation over a wide range are not dependent on the curvature of the mirrors being checked and precision optics is not used during measurement. The relatively small base of the measuring device and its independence of the curvature of the surface being checked permits one to use the suggested method for production checking of inetal- optics under plant conditions. BIBLIOGRAPHY 1. "USSR Inventor's Certificate No 593070," BYULLETEN IZOBRETENIYE, No 6, 1978. 2. Dukhopel, I. I. and L. G. Fedin a, OPTIKO-MEKHANICHESKAYA PROMYSHI.ENNOST', No 7, ' 1973 COPYRIGHT: LIYaF, 1980 6521 CSO: 1863/26 75 FOR OFF'1('.[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02/09: CIA-RDP82-00850R000500434418-4 FOR OFFICIAL USE ONLY - UDC 535 BEAM HOLOGRAPHY Leningrad GOLOGRAFIYA I OPTICHESKAYA OBRABOTKA INFORMATSII: METODY I APPARATURA in Russian 1980 (signed to press 31 Dec 80) pp 231-234 ~[Article by N. A. Valyus from the collection "Holography and Optical Processing of Information: Methods and Apparatus", edited by V. G. Skrotskiy, B. G. Turukhario and N. Turukhano, Izdatel'stvo Leningradskogo instituta Yadernoy fiziki, 500 copies, 237 pagesJ [Text] ~ analog of the holographic image obtained in coherent light is an inte- gral photograph recorded in noncoherent light. An i.ntegral photograph, like holography, is characterized by "complete recording" about the object, its wlume and its relative position in space. The same pattern can be recreated from the part of an integral photograph and from a fragment of it as from the entire photograph. ~ The image of an integral photograph may be called beam holography since only the amplitudes rather than the phase of the wave are recorded when using noncoherent light and a description of the process is found within the framework of geometric (beam) optics. The principle of integral photography was suggested at the beginning of this cen- tury by G. Lippman, more well known for his idea of interference photography, which was a prototype of the modern reflective hologram substantiated and developed by Yu. N. Denisyuk. G. Lippman [1] called his method integral photoqraphy because the three-dimension- al image produced is synthesized from images of a large number of small micro- scopic images of the object photographed with a scanning lens photographic plate and one can say the restored wave front is integrated from the number of elemen- tary beams represented by these. images. The process of photographing a three-dimensional image on a scanning lens plate is shown schematically in Figure 1. One surface of the scanning lens plate R has fluting in the form of smal.l lenses contacting each other, while the other surface is coated with a light-sensitive layer on which the small lenses are focused. The object AB, located in front of the lens side of the platP, is exposed on the light- sensitive layer by the small lenses in the form of a set of microscopic images - albl~ a2b2r a3b3 76 - FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000504030018-4 FOR OFFICIAL USE ONLY a. P . Q' qi a, y 1 ' b~ j a~ a ~1 ~ a 'I, A ~ ! ti ?i x e ~ ~ ~ n; ~ . bl a, . a Figure 1. Figure 2. A photograph can be made on~this plate without the use of a lens. The process of achieving an integral photograph is a two-step one. Therefore, after the plate R has been exposed, developed and fixed, it is placed opposite the other, exactly the same plate P, turninq the plates with the lens elements toward each other (Figure 2) and they are illuminated from behind by scattered light. - The inverted positive imag~ is copied, i.e., elementary images albl, a2b2, are inverted to images albl, a2b2, The second plate is a positive image which restores the inverted path of the beams from the object (the quantified wave front from the points of the object is re- stored). By illuminating this plate P on the emulsion side and looking at it from the lens side (Figure 3), one can see a three-dimensional image of the photographed subject AB'. The three-dimensional image of the subject is recreated by the wave fronts in the directions of beams from each elementary image. Therefore, the visible image is imaginary and is located at a quite specific distance on the plate. P � a' A~ ~ ~ ~r~:' ~ F~ ~ R B' ~ ~ ~ - ,w 1~. ~ Figure 3. Figure 4. one can see a side view of the object when viewinq the integral photograph from the side. When one approaches the panorama of the visible picture is expanded and the perspective and angular size of the subjects change: 77 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY An integral photograph made in noncoherent light permits filming of full-scale � three-dimen~ional objects of unlimited dimensions at arbitrary distances and per- mits color images to be made. Different modifications of integral photography are known [2l. Thus, for example, it is possible to increase the three-dimensional resolution of the produced image by constriction of the aperture when taking an integral ghotograp~x through a large- diameter objective (or lens}. A synthesized integral phctograph can be made when the full-scale object is photographed on a set of frames from different angles of approach and then the print of these frames is projected from these same angles of approach onto a scanning lens plate, as is shown in Fic;ure 4. Conversion of a focused integral photograph by coherent radiation to a reflective hologram w~..t~ altered three-dimensional image scale yields interesting results. Quasi-integral photographic systems, so-called autostereoscopic images, which find apglication in the form of three-dimensional advertising images and as mass ster- eoscopic isoproduction, are of practical interest. The principles of beam holography accomplished by using scanning optical systems permit one to solve many problems not accessible by tradition~l classical devices. Scanning integral stereoscreens, scanning objectives an~i eyepieces are knowm. - Scanning optical systems provide recording not only of the three-dim~nsior~al angles of approach of an object but of its time angles of approach as well. The highest movie filming speeds are achieved by using scanning frame cameras. Scanning opti- cal systems are promising in co.aputer equipment as storage devices, as processors, controllable slides and so on [3]� Finally, one should also point out the analogy of scanning systems to holographic - systems in the field of interferometry. Essentially the Moire phenomenon is inter- ference of three-climensional frequenries of raster gratings that is used in mea- suring equipment. BIBLIOGRAPHY 1. Lippman, G., JOURNAL DE PHYSIQUE, 1908. 2. Valyus, N. A., "Rastrovaya optika" [Scanning Optics], Moscow-Leninqrad, GITTL, 1949. 3. Valyus, N. A., ~Matrix Type of Optical Computers," "Radioelektronika opti- cheskogo diapazona" [The Electronics of the Optical Band], Moscow, VZMI, 1970. COPYRIGHT: LIYaF, 1980 6521 CSO: 1863/26 78 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY _ , SOFTWARE ABSTRACTS OF ARTICLES IN JOURNAL 'rROGRAMNIING', SEPTEMBER-OCTOBER 1981 Moscow PROGRAMMIROVANIYE in Russian No 5, Sep-Oct 81 pp 95-96 UDC 51 : 621.391 PROBLEM OF EQUIVALENCE FOR GENERALIZATION OF LL (k) GRAMMARS [Abstract of article by Nepomnyashchaya, A. Sh. ] [Text] A proo~ is given of the solvability of the proble:~i of equivalence for grammars of P classes, wh~ch are specified by uniting mLL(k) grammars with limitations on outputs. P classes, as well as deterministic a~tomata with ~ a magazine memory corresponding to them, represent new classes with a solvable equivalence p roblem. UDC 519.767.2 DESCRIPTION OF SEM~~NTICS OF PROGRAMS BY MEANS OF SUBSTITUTIONS IN GRAPHS [Abstract of article by Gostev, Yu.G.] - [Text] It is suggeste~ that programs be interpreted as combinations of substitu- tions tu be applied to graphs of a certain kind. This makes it possible to draw the theory of algorithms closer ~:_o methods of describing computations used at the present time and in addition makes it possible to describe computations on random data s tructures. UDC 681.3.06 - SI:MANTIC t10DELS OF PROCEDURES [Abstract of article by Pastor, F.] [Text] Seman tic models of procedures and other objects encountered in programs written in a high-l.evel programming language are discussed. A method is suggested ,for constructing complicated models of statements ~rom simpler ones, based on a composite app roach. ~ 79 FUk OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02/09: CIA-RDP82-44850R444544434418-4 FOR OFFICIAL USE ONLY UAC 687,.3.06 PROGR~r'INIING SUB.S~I~U~IONS OF T,ArTGU~,GES [Abstract o~ artic7,e l~y Chexemisi.nov, A.I.] - [Text] The problem of programming substitutions of 7.anguages is discussed--the operation of converting string data forming the liasi:s ~or constructing different kinds of algorithms for processing data represented by~ chains oP characters. A system o~ notation ~or defining substitutions of languages is suggested. UDC 681.3.06 CONSTRUCTION OF A DATA BASE BASED ON 'L'HE CONCEPT OF ABSTRACT TYPES [Abstract of article by Zamulin, A.V. and Skopin, T.N.] [Text] The concept of astract types of data as applied to designing systems for controlling data bases is discus~ed. The procedure discussed makes it possible on the basis oP a universal programming language with abstract types to develop data bases in accordance with a specific application model. IJl)C 681.142 _ SOFTWARE IMPLEMII~TTATION OF MtJLTIPROCESSING [Abstract of article by Godunov, t1.N., Yemel'yanov, N.Ye. and Sverdlo~~, S.S.] [Text) A description is given of a multiprocessing executive routine (MP executive) created within the framework of the INES-2M [economic information system] SUBD [system for control of data bases], which can also be used independently. The MP executive makes possible parallel execution of the computing process and working with a disk-type virtual memory utilizing the distinctive features of the mult~-- processing mode of executing a routine. UDC 681.3.06 IMPLEMENTATION OF A'FOREKS' COMPILER FOR THE AS-6 CENTRAL PROCESSOR [Abstract of article by Galatenko, V.A, and Khodulev, A.B.] [Text] A description is given of the instruction generator in the compiler from FOR7'RAN Eor the central processor o~ ttie AS-6 complex. Tnformation is presented on organization of the AS-6 central processor. The characteristics of the AS-6 central processox and BESM-6 are compared. An eva luation of the AS-6 central pro- cessor as a FORTRAN machine is given. - ~ 80 FOR OFFIC[AL USE GNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAL USE ONLY UDC 681.3.06 INFORMATZON S~k1RCH AS' ~~RQ~k~S z~ I~~VGUI;~~IC R~CQGNi~~ON 0~ CH~R~1C'~ERS [~bstract o~ axtic7.e ~y~ Samedova~ M,A.~ [Text] The linguistic suppart vf a data search of objects o~ a medical nature represented in a normal.ized natura7. language is di;scussed. . , UDC 681.3 SYSTEM OF DTALOGUE PREPARATI~ON OF TASKS FOR UNIFTED-SERTES COMPUTERS [Abstract of article by Khanykov, V.V., Rybakov, A.V. and Anan'ina, N.V.] [TextJ A description is given of an instrument system for working with a design library for FeS computers within the framework of the operating system. UDC 681.3.06 : 51 EVALUATION OF INDICATORS OF THE CORRECTNESS OF THE FUNCTIONING OF PROGRAMS WITH A MULTILEVEL STRUCTURE [Abstract of article by Khaletskiy, A.K.] [Text] A mathematical formulation is suggested for the problem of evaluating the vector of indicators of the correct functioning of a complicated program with a tree-type hierarchical structure fpr the case when failures of its components are independent. ~ UDC 681.3.06 - RECURSIVE ESTIMATION OF ARITHMETIC FUNCTIONS IN LISP SYSTEMS [Abstract of article by Stefanyuk, V.L.] [Text] In ~his note reasons are presented in favor of *_he "explicit" computation of elementary functions in LISP systems. Examples are given of simple recursive . codes for computing sin , exp and ln , illustrating the general approach. ' UDC 681.3.001 COMBIN~D FORMAT FOR OUTPUT OF NU~RICAI, TN~'ORMATION ~'OR DESIGN DOCUMENTS (Abstract of article by Moty7.', D.N., Sokolinskiy, Yu.A. and ~arber, K.S.J ii'extJ descxiption is gj.ven o~ a combined Pormat ~or the representation o.f c:umer~cal in~ormarion making it possi,ble to output numbers with a specified relatjve ~ rror. The descrit~e.d Pormat is imp~.emen~ed ~;n ~he procedux~ ~unctiQn of the ~I,/J. I.anguage in r.he X~S [Uni~i.ed Sexies] opexating sys.tem. ~OPYRIGH~: Tzdatel.'stvo "Naukzt", "~xogxammixo~~aniye", 7.987, ~ i�~~~_ti~,5 81 ~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY UDC 681.142 SOFTWARE IMPLEMENTATION OF MULTIPROCESSING Moscow PROGRAMMIROVANTYE in Russian No 5, Sep-0ct 81 (manuscripfi received 12 Ma.y 80) pp 44-49 [Article by A.N. Godunov, N.Ye. Yemel'yanov and S.S. Sverdlov] [Text] A description is given of a multiprocessing executive routiiie (MP executive) created within the framework of the INES-2M [economic information system] SUBD [system for control of data bases], which can also be used independently. The MP executive makes possible parallel execution of the computing process and working with a disk virtual memory utilizing the distinctive features of the multiprocess- _ ing mode of executing a routine. As mentioned in [1, 2], the idea of multiprocessing, i.e., breaking down a user's task into independent quasi-parallel ~rocesses, is very attractive both from the viewpoint of ttie convenience of programming and on the plane of possible optimiza- tion of working with an external storage. Furthermore, it was indicated that the most advantageous means of parallel execution would be the hardware implementation , of a multiprocesaing system. Nevertheless, a considerable gain can be made also - ~with software implementation of such a system. _ In this article a description is given of a real multiprocessor executive system which can be used both for creating individual user routines P.nd as a system for programming various components of the software of an ASU [automated controi system], data base control systems and the like. Although the MP executive has been imple- mented within the framework of the INES SUBD [3], it repreaents a totally indepen- dent system which can be u~ed independently or as part of any software complex under the control of a YeS [Unified Series] r~perating system. The following were the key requirements in creation of the MP executive: Offe:-ing the user a developed system of macroinstructions for parallel axecution of the computing process. The creation of convenient facilities for working with a virtual memory employing disks. - Ensuring the effective control of swapping in the multiprocessing mode. 82 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPR~VED F~R RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY Making possible the auxiliary functions required for practical work with the exe- cutive routine (debugging facilities, message output facilities, management of routine execution protocols and, in particular, swapping procedures, and statistics gathering facilities). Ensuring compatibility of the MP executive with the YeS ~~erating system. The functional capabilities of the MP executive are illustrated in fig 1, where 1, represents facilities for working with the virtual memory, 2 facilities for parallel ex~~stion and the synchron3zation of processes, 3 debugging facili.ties and 4 facil'i- ties for organizing parallel message flows. Multiprocessing Instruction Set The parallel exeuction of a routine which is to be executed in the multiprocessing mode takes place through special macroinstructions. The following macroinstructions exist for the origin, termination and synchronization of processes: INIT--initiation of multiprocessing operation, PROCS--origin of a process, PREND--termination of ~.i9~ the process, HALS--cancellation of all processes begun, PRTY--changing the priority of a running process, MAYLNT--enabling suspension of a running process, WALS-- awaiting the termination of all processes started, WANS--awaiting the termination of any process started, SM--awaiting the assigned value in the semaphore and WAITM-- awaiting an operating event (in the sense of the operating system). 2 . llpatpann ~ +oneTnOa - ~urn~ ~ i i/ . ~ t 2) oC . - c~nsu Figure 1. Key: 1. User's program 2. Operating system facilities In the initiation of a multiprocessing system its initial configuration and start- ing parameters must be specified, such as the maximum number of processes e:cistin~ simultaneously in the system, the Tnaximum depth of the process origin rrze, fn- �structions for use of the virtual memory, a program identifier for tre initial pro- cess, some additional instructions for the accident`1 termination of processes, debugging printout and the like. ~ 83 F~R OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE 4NLY During initiation the system's control tables are ~ortned, special system processes are created and the initial user's process is starfied. Any process can issue a PROCS instruction for the origin of an offspring process. If there are insufficient re~ources for the creation of a new process, the parent process is halted until the necessary memory is freed. After the creation of the process it, as a rule, also starts, since, if nothing special is stipulated, the offspring process receives higher priority than the parent process. In such a manner the multiprocessing system stimulates growth of the process tree "in depth," and not "in breadth," in order to reach more rapidly the ma.ximum depth at which processes begin to end, issuing the instruction PREND. The memory freed at the termination of processes is used for producing new processes. There is also the possibility of arbitrarily assigning priority to a hegun proc~ss and, in addition, each process in the course of execution can change its priority by means of the PRTY instruction. This can result in dead-end situations when all processes begun wait for resources for the creation of offspring processes. If the assignment of priorities is left to the discretion of the MP executive itself, then it provides certain measures for the prevention of dead-end situations, on the basis of the maximum depth of the process origin tree indicated during initiation. The parent process can cancel all processes originated by it by means of the HALS instruction, whereby each offspring process is cancelled together with a11 processes subordinate to it. The WALS instruction suspends the running process until the end of the operation of all offspring pr~ocesses produced by it, and the WANS in- struction until the end of any offsprin~; process (if there are any). The most common means of synchronizing processes is the semaphore device, whereby the sema.- phore can be any memory cell. By means of the SM instruction an indication can be given regarding the entry of a certain value into the semaphore, regarding waiting for a specific value or condition in the semaphore, and regarding waiting followed by an entry into the semaphore. Virtual Memory of a MultiprocessinF, System J The multiprocessing instruction set: includes the following facilities for wo~king with a virtual memory employing disks: OPENM--open the file of the virtual memory, CLOSEM--close the file of the virtual mc~mory, EXFIL--designate the working file of - the virtual memory, LV--load into the rc~gister the word with the indicated virtual address, STV--store contents of the reg~ister according to indicated virtual address, MV~l--move field from virtual memory to virtual memory, MVS--move field from ordi- nary memory to virtual, MSV--move field from virtua~ memory to ordinary, CVV-- compare field of virtual memory, CVS--compare field of ordinary memory and virtual memory, INPG--input page with indicated virtual address, GETPG--issue clean page with indicated virtual addre5s, UPDPG--set page correction tag, FREPG--free page with virtual address indicated, 'FIX--make request for fixing of the required number of virtual memory pages. The MP executive makes it possiblP to work simultaneously with several virtual files. Each proces: :nust open the file needed by it by means of the OPENM in- structi_on and upon the end of working with it must close it by means of the CLOSEM instruction. Actual closing of the file in the sense of the YeS operating system is produced by the MP executi:�e when th~~ last process openir.g it finishes working 84 FOR OFFI('IAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC[AL USE ONLY with this file. In addition, each process is assigned, if this is necessary, an individual working file of an indicated size, it is opened and closed automatically by the MP executive, and accessing it does not require identification of the file. The LV and STV instructions are intended for working with words of the virtual memory and are similar to the assembler L and ST instructions. The MW, MVS, MSV and CW and CVS instructions are similar to the assembler MVC and CLC instructions, ~ respectively, but the length of operands can reach 32,768 bytes. The INPG instruc- tion makes it possible to gain access to a page with the indicated virtual address and the requested page will be fixed in the working storage right up to a clear indication by the programmer of completion of work with it 1~y means of the FREPG instruction. The UPDPG instruction is used in order to indicate which page en- tered by means of the INPG instruction will be corrected and, consequently, later must be automatically output according to the corresponding virtual address. By means of the GETPG instruction it is possible to obtain any "clean" page, i.e., a working storage buffer. The correction tag for this page is set automatically, in order for it later to be read out into the appropriate virtual file. Let us note that in execution of the MW, MVS and STV instructions the page correction tag is also set without an explicit instruction from the programmer. As in indicated in [l, 2), multiproc~.ssing opens up unique opportunities for con- siderably increasing the effectiveness of a virtual memory. The simultaneous ex- istence of a great number of processe;, each of which possibly issues requests for exchange with virtual files makes it possible to implement an effectivz exchange algorithm which stimulates the partial review in advance of a sequence of accesses to the virtual memory [2]. There also is discussed an algorithm for protection from dead-end situations in a multiprocessing system which can originate when working with the pages of a virtual memory under fixing conditions. ~ Additional Possibilities Offered by the.~i~ Executive The existence in tre system of a great number of independent processes creates serious difficulties in organizing the printout of the output messages of each of them. In quasi-para11e1 running of a user's program the sequence of the exe- cution of individual processes is determined dynamically by the multi.processing -system and, consequently, the procedure for the output of inessages by various pro- cesses is unpredictable. It is obvious, however, that from the viewpoint of opera- . ting convenience the separation of flows of inessages of various processes in print- ,out is tatally necessary. The use of standard YeS operating system facilities for - r_hese pur.poses is exceedingly difficult for an applied programmer because of their cumbersome nature. The use oE the printout f~cilities offered by the MP executive makes possible the automatic separation of flows of inessages of various processes and in addition frees the pr~gr.ammer from organizing control tables of data (DCB's) and from th~ need to perform the operations of opening and closing data sets. Here the re-enter3bility of the routine executing the printout is not disturbed. With the accidental termination of somP process the MP executive makes possible the output of a standard diganostic message regarding the type of error, a list output of the arPa of interruption and cf the state of registers at the moment of interrt~ tion, and also a list autput of key multiprocessing control tables at the moment of interruption. 85 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR UFF[CIAL USE ONLY . In the debuggi.ng tnode the MP executive tnakes it possib~e at poin~s indicated by the programmer to print out the current state o~ control tables o~ the multipro- cessing system and of process waiting lines. When working with files of the virtual memory there is the possibility of periodically (at a given interval) obtaining in- formation on exchanges carried out and on the currenfi state of virtual memory buffers. Information on exchange dynamics in the multiprocessing system can be very helpful for a qualitative analysis in selecting optimum multiprocessing para- meters for a specific problem (number of processes, number of buffers, virtual memory, their size and the like). Functioning of a Multiprocessing System The execution of multiprocessing instructions is performed by program modules in- cluded in the MP executive's structure. The completion of an instruction by the MP executive can involve either continuation of a running process or switching to another process. For example, if in a reference to the virtual memory it is found that the page indicated is already ~.n the working storage, the running process again receives control. Otherwise it is suspended with the appropriate waiting~ ~ code, its request for a page of the virtual memory remains in the waiting line for execution and control is transferred to one of the processes ready to run. The MP executive maintains waiting lines of processes waiting for a specific event and the waiting line of so-called ready-to-run processes, from which the process with the highest priority is selected in switching. The state of these waiting lines is constantly corrected with the completion of each multiprocessing instruc- tion by the executive routine. For example, execution of the P~',END instruction for a certain process can result in placement of the parent procPSS in the ready-to-run waitino line (if it was awaiting termination of the running of all offspring pro- cesses and the running process was the only one or the last of them), as well as one of the processes awaiting memory resources for the creation of an offspring pro- cess. Special system process W performs the function of waiting for any event (in terms of the operating system) which is specified by the process. Among them, process W fixes the end of any exchange with external units. Let us note that a11 process- es, in addition to facilities for working with the virtual memory, can order ex- change with external units through ordinary mac~~~instructions of the operating sys- tem. Process W has minimum priority and thus receives control when it remains the - only one of the processes ready to run. Furthermore, process W reviews the waiting line of events and transfers the individual processes into the ready-to-run state. If not one of the awaited events takes place, process W issues the operating in- struction WAIT. System process E implements the algorithm for exchange with the virtual memory de- scribed in [2]. Its functions include review of the waiting line of requests for exchange with the virtual memory and initiation of the exchange selected. Process E has maximum priority in the system, which makes it possible to insure most com- plete matching of the operation of the processor and input/output channel. 86 ' FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 EOR OFFIC'IA1, t)~F: ONI.ti' System processes W and E are created by the MP executive upon the init~ation of multiprocessing activity, and process E is generated only when an instructi.on re- garding working with the virtual memory is contained in the INIT macroinstruction. As is obvious from a description of multiprocessing instructions (cf. fig 2), some of them are similar to operating macroinstructions of the YeS operating system associated with a multitask situation. Let us note that each process of the MP executive is at the same time a subtask in terms of the operating system. However, the issuing of ATTACH operating macroinstructions is necessary only upon the initi- ation of a multiprocessing operation and in cases of the accidental termination of certain processes. With the normal L~rmination of a process and the creation of a new one, the issuance of DETACH and A7'TACH macroinstructions is not required, since one and the same operating system subtask is used many times in succession for supporting a great number of MP executive processes. Execution of the HALS macroinstruction takes place similary without accessing the operating system. In- structions for the synchronization of processes on the one hand give a programmer more convenient facilities than operating macroinstructions. For e~mple, often the common function of waiting for all subordinate processes is implemented in the MP executive as a single simple WALS instruction. On tr~e other hand, instructions for working with semaphores, although they make it possible to simulate the operat- ing facilities for working with ENQ and DEQ resources, are completed by the MP exe- cutive without using these system macroinstrucCions, which makes it possible to speed iip execution of the program. llpci apumcm npnc~ecca 2) Mfl-r~vHUmop 1 ~ ,QrpeBa npoueccnB nona~oBamenA 3~ i ~ / ~ / \ ~ ~ I ~ ~ ~ ~ / ~ ~ 'w Figure 2. Key: 1. Priority of process 3. User process tree 2. MP executivc~ [,et us stress that although each process of the MP executive is also a subtask of the operating system the function of the process scheduler is performed exclusively by the MP executive. With each accessing of the executive routine it selects for execution a single process (subtask of the operating system) and the remaining a.re - 87 , FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/42/09: CIA-RDP82-40850R000500430018-4 FOR OFFICIAL 11SE ONLY inhibited, even if they are ready to continue running. At the same time in the interval between two accesses to the MP executive each process is uninterruptable in a given multiprocessing mixture. This opens up the possibility of extensively using qua,si-re-enterable routines for processes, the creation of which is less dif- ficult for a programmer than ensuring their abso~ute re-enterability. (The main- tenance of re-enterability of some sort is necessary in the widespread situation when a great number of processes are designed to run according to a single routine.) The concentration of the functions of controlling processes in the MP executive makes it possible to achieve good agreement between ~he work of the central pro- cessor and the execution of input/output operations. ror this it is sufficient to use for the purpose of waiting for the end of exchanges the multiprocessing macro- instruction WAITM, which results in suspension of the running process and the start of another process ready to run. The possibility of organizing exchanges of just operat-ing system facilitie:; is also not excluded (without accessing the MP execu- tive), which can entail only a slight reduction in the effectiveness of the system. Moreover, the job of controlling the multiprocessing system does not contradict the use in programs of standard broad-application operating system macroinstruc- tions, such as GETMAIN, LINK, LOAD and the like, the use of which can be dictated by the programmer's support considerations. This compatibili.ty of multiprocessing with operating system facilities ma.kes it possible also Lo adapt fairly easily to _ a multiprocessing mode a program written by the traditional sequential method. For this is required mainly a"superstr~icture" of the program's upper logic level making possible its parallel execution. . The methods of multiprocessing discussed can prove to be a convenient tool for both system and applied programming. The instruction set described above provides the programmer with added conveniences as compared with the multitask facilities offered by the YeS operating system and furthermore is implemented in order to reduce as much as possible the number of expensive accesses to the operating sys- tem's supervisory routine. The convenience of programming in terms of multipro- cessing was regarded as an important factor in creating the MP executive. There- fore the executive routine has a structure .~naking it possible easily to add addi- tional instructions xor speci~ic applicatioris and to adapt the multiprocessing system to the user's needs. I The functional capabilities of the MP executive make it possible to use it in a broad class of data processing task.s [1, 2] and in software design system tasks. On the basis of the MP executive a multiterminal SUBD IN~;S s}~stem has been de~igned which is described easily and naturally in the language of parallel processes. Bibliograpily 1. Arlazarov, V.L., Volkov, A.F., Godt?nov, A.N., Yemel'yanov, N.Ye., Zenkin, V.D., Konstantinov, G. and Lysikov, V.T. "H~rdware Implementation of Multiprocess- - ing," AVTOMATIKA I TELEMEKHANIKA, ro 8,~1977. 88 FOR OFF(CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 2. Godunov, A.N. and Svexd7.ov, S.S. "~rganizatiQn o~ a Virtual Memory in a Multiprocessing System," AVTQM~TTKA T TELEMEKHANIKA, No 12, 1978. 3. Arlazarov, V.L., Dyukalov, A.N., Yemel'yanov, N.Ye., Tvanov, Yu.N., Kochin, Yu.Ya., Tokarev, V.V. and Faradzhev, T.A. "The TNES In~ormation System," AVTOPiATIKA I TELEMEKHANIKA, No 6, 1979. � COPYRIGHT: Izdatel'stvo "Nauka", "Programmirovaniye", 1981 8831 CSO: 1863/45 89 FOR OFF[~IAL USE ONLX APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY UDC 681.3.06. IMPLEMENTATION OF 'FOR~Y.S' ~:OMPILER FOR AS-6 CENTRAL PROCESSOR Moscow PROGRAP'II~IIROVANIYE in Russian No 5, Sep-0ct 81 (manuscript received 24 Dec 80) pp 50-58 [Article by V.A. Galatenko and A.B. Khodulev] [Text] A description is given of the implementation vf a FOREKS FORTRAN compiler for the central processor of the AS-6 complex. A brief description is given of the AS-6 central processor. The results of statistical studies of ob~ect pro- grams are giveri. The AS-6 is a multimachine complex which can include centr~al pr.ocessors (AS-6 TsP's), a BESM-6 computer and peripheral machines (PM-6's). The architecture~of the AS-6 central processor is rather complicated. Let us mention, for example, the presence of three types of registers, the page segment organization of the memory, and the extensive and quite individual instruction set. In this article a description is given of a compiler from FORTRAN for the AS-6 TsP obtained by modifying the FOREKS compiler for the BESM-6 [1, 2]. The FOREKS is a two-operation compiler. In the first operation the program is translated from FORTRAN to the internal language and in the second from the internal ianguage into machine instructions. The internal representation used by the compiler for the BES:~i-6 proved to be totally suitable for the AS-6 central pxocessor; therefore, only the instruction generator is a part of the compiler specific to the AS-6 central processor. It is precisely to the instruction generator for the AS-6 . central processor that this article is mainly devoted. In addition, in this article data are presented which characterize the AS-6 central processor as a FORTRAN machine. The instruction generator for the AS-6 central processor, as are the other parts of th~ FOREK~ compiler, is written in the ASTRA language [3] and operates in the BESM-6, producing a text in AS-6 central processor autocode [4]. The capacity of the generator is about 3500 ASTRA instructions (about 6000 BESM-6 words). The input lanf;uage of the FOREKS compiler conforms basically to the FORTRADI-77 standard [5] and also contains facilities absent in FORTRAN-IV, such as "structural" conditional instructions, I'F-THEN-EZSE-ENDIF, cycle instructions, DO-REPEAT (in ~ co~bination Gri.th the EXTT statement), as well as the CHARACTER (character change) and BIT (bit change) data types. - 90 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 ' FOR OFFICIAL USE ONLY 1. Brief Descxiption of AS-6 Central Processox In this section are descritied only those features o~ the AS-6 central processor which influence the instruction generator. More~detailed informa.tion on the AS-6 central processor, including a description of the instruc~ion set, can be found in [4]. 1~. Addressing Mechanism Ti~e memory in the AS-6 central processor is virtual and segmented. The maximum s.~ze of a segment equals 218 48-bit words. Qne task can work with no more than 4096 segments. The virtual memory is accessed by means of 4$-bit descriptor re- gisters. There are 16 of these registers. They contain the following information: 1. The number of the segment which is accessed (12 bits). 2~. The address within the segment. This address is called the base and occupies 2~[+ bits. 3. The type of base (2 bits). In the AS-6 central processor it is possible to address not only whole words, but also half-words, bytes and bits. The type of base shows in which kind of units the address within a segment is given. 4. The number of the index register or half-word adder through which indexing will be performed. . The descriptors contain also other information which is not important for our pur- poses. There are eight 24-bit index registers in the AS-6 central processor in a,ddition to the descriptor registers. The operations which can be performed in them are addition and subtractic~n but their main function is the modification of addresses. In an AS-6 central ~~rocessor instruction the address of an operand in the memory is given by means of the number of the descriptor register, the shift l~16 bits) and the indexing tag. The operand is accessed from the segment indicated i.n the descriptor register. Its address within a segment is computed as the sum of the shift indicated in the ir.struction, of the base of the descriptor and, if an indexing tag has been set, the contents of the index reg�~ster ox half word adder whose number is given in the descriptor. The type of descriptor base must agree with the type of operand or be a word base. In the latter instance before being added with the shift and contents of the index register or half-word adder the descriptor's base is multiplied by 2, 6 or 48. Descriptor registers can be used not only for addressing, but also for organizing cycles: In the descriptor register are placed half-word steps and the final value of the control variable; the instantaneous value of the control variabJ.e must be stored in the index register. There is a 3-operand cycle end instruction which with a positive value of the step is similar to the BXLE instruction in YeS [Unified Series] computers (cf. [6]). Tf the step is negative, a transfer is made, when the new value of the control variable is not lower than the final value. ~1 . FOR OEFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 2. Adder Registexs - The eight 48 bit adder reg:~.sint8 asewelliasQl gic'operat ons.a~~~~etic operations with a f loating and fixed po , , Numbers with a fl.oating point have a 40-bit mantissa, a 7 bit floating point part and a sign bit. Negative numbers are represented in direct code. The floating point pqrtion ia he aceCOf a he~decimalZdigit.ofTheerangedof~representablesnumbers formed with an accur y is from 16-65 to 1663. The set of operations on numbers with a floating point is the traditional one. The mode for the ex~cutNormalizationeofctheeresultSCan bedinterlocked,sascwe111aseover- locking registe . flow and loss of significaace interrupts, etc. In the AS-6 central processor there are aaddedngsubtractedfandWmultipliedh dThere precision numbers. These numbers can be , . is no instruction for dividing numbers with double precision. Whole numbPrs can be represented by two methods. Word whole numbers contain 47 bits and a sign. Ordinary arithmetic and logic operations, in addition to the di- vision operation, can be performed on these numbers. Half-word whole numbers contain 23 bits and a sign. 'The right halves of adders, which are called half-word adders, are used when working with them. Negative numbers are represented in complement. The operation set includes arithemetic (including d3.vision) as ~ool canlbeiusedeforimodifyingaaddresses~lf-word adders, as the index registers, , The AS-6 central processor permits working with individual bytes and bits. Here the far right bits~edtin theeo8erationesets bOperations with bytestareaperformed functians are incl P in the right byte of adders. The adder registers can be used for ope~~ationa with sequences of bytes and bits. Here the length of a sequence must not exceed 12 bytes (96 bits). 3. Memory-Memory Format Instructions Sequences of bytes and bits can be processed not only in registers but also in the memory. The set of corre~ponding AS-6 central processor instructions is similar to the instruction set of the SS f~rmat in YeS computers with the only difference that the length of operands is n4t obligatorily specified directly in an instruc- tion: It can be placed in an index register and its number cari be indicated in the instruction. Here the length of operands can reach 218 '.~ords. 4. Dynamic Loading The dynamic loading strategy (cf. [7], sec 4.9.2) is used in the AS-6 central processor: Programs are loaded with respect to first access to them. As a result 92 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 - FOR OFFICIAL !1SE ONLY there will be no super~luous programs i~n the memory~. Programs which have been completed and have become unneeded wi11 sooner or l.ater be ~orced out into the - secondary storage. 5~. State Stack - A unique feature of the AS-6 central processor is the state stack (hidden stack) mechairism. Tt makes possible the automatic saving o~ return addresses and the saving and resetting ot registers. The stage stack operates in the following manner. Accessing of a subroutine is - written in the form of two instructions. In the first of them a so-called saving mask is established, indicating which registers are to be saved_ The secand exe- cutes the jump per se. Here in the state stack is registered the return address and certain otner characteri$tics of the module from which accessing takes place. - While the subroutine is run the registers indicated in the saving mask are entered - into the state stack before the first execution of any instruction altering their conte-:.ts. Thus, unu~ed registers in the state stack are not stored. A single return instruction is sufficient in order to exit �rom a subroutine. It causes the resetting of registers hidden in the state stack and transfers control to the return point. 2. Machine Represen tation of Language Obj ects Numbers with a floating point are represented naturally. It was a more complicated matter to select the representation of whol.e numbers. We settled our choice on half-word whole numbers in order to render effective access to elements of arrays - and the implementation of cycles. Of course, with this the range of representable whole numbers turned out to be rather short: from 8388608 to 8 388607. The structure of the AS-6 central processor makes it possible to address the on-line memory ~*ith an accura~y of a bit. Therefore, for logic values the AS-6 FOREKS compiler assigns in terms of a single bit. This makes it possible to work effect- ively with very large (up to 223 elements) logic arrays. - Thus, regions of the memory of diff.erent sizes are assigned to values of different types. This contradicts the FORTRAN-77 standard [5]. However, the memory savings which originates when using whole-number and especially logic arrays in our opinion outweighs this disadvantage. Values of the CHARACTER and BIT type are represented as sequences of bytes (bits). They are worked with only hy means of inemory-memory format instructiens. Registers are used only in convertj.ng from one type to another, e.g., from CHARACTER to INTECER. Accessing an element oC an a~ray (e.g., A(T)) is a rather complicated operation (setting the descriptor, setting the inciex and accessing per se). In a3d:~.tion, operations with descriptors are performed rather slowly in the AS-6 central pr.o- cessor. Therefore it is advisahle to place arrays as much as possible in a single segment, since then a single descriptor can be used without resetting for working 93 FOR OFFiCIAL USE ONi.Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY with several arrays� The ~.ength ok an axray cannot exceed the size o~ a segment (218 words). An individual segment is assigned to each common (COMMON) block. Tt follows from this that dif~erent common blocks are protected ~rom one another and improper working with one block cannot influence another. 3. Algorithms Used . Methods of translating from FORT.RAN are described in detail, e.g., in [8]. We will dwell briefly on some algorithms used in the instruction generator for the ' AS-6 central processor. 1. Generation of Instructions f~r Line Sections A sequence of FORTRAIJ program instructions which does not contain control jump labels and instructions (a line section), after syntactical analySEachntriadicona tion, is represented in the form of a graph of triads (cf. [8]). tains an operation indicator and references to triads representing operation oper- - ands. Triads of a special form represent variables. ror a specified graph it is possible to generatP various sequenc~es of instructions for its execution differing in zhe computation pro^edure for triads and in the use of registers. The follow- ing recursive algorithm is used in the realization described for the purpose of generating triad computation instructions. 1. Instructions are generated Eor computing the second operand of the triad (if they have not already been generated). 2. The same for the ~irst operand. 3. Instructions are generated for the performance of an operation of a triad on its operands. _ This algorithm is applied seuqentia7.ly to all triads representing instructions of the source program. The use of registers is described below. Although this generation procedure is not always optimum (cf. [9]), practically speaking it make~ it possible to produce a sequence of instructions close to optimum. 2. Distribution ot Registers In the AS-6 central processor there are eight adders, eight index registers and 16 descriptor registers. The adders and index registers are distributed uniformly. A table of registers is constructed. Each element of this table contains a refe- rence to the triad which this register represents and the field describing the state of the register (0--free, 1--Q::cupied by a constant which must be indicated directly in the instruction, 2--occupied and there is a copy in the memory, 3-- occupied and no copy in the memory, 4--register occupied and its contents cannot be altered). In turn, the triad contains a counter which indicates for how many triads it is an operand. Each time the triad is used as an operand in the genera- tion of instructions its counter is reduced by a unit. When it becomes equal to 0(i.e., the triad is no longer needed), the registers associated with it are - 94 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OF'FIC[AL USE ONLY ~marked ati Pree. I~ an addex, ,~or exampl.e, is needed ~or the next computation an3 rionc are free, the addex with the minimum value of the state �ie],d is occupied. Here attention is not pz~id to the counters in triads. Although theoretically this algorithm is not aptimum, in practice it is totally satisfactory. Let us present the following data. Sixteen ~ORTRAN programs with a total size o~ 5614 strings were translated. With this 28383 machine instructions were generated, 5235 of them being instructions for setting adders. Tn only 98 instar~ces did a setting instruction cancel values whose presence in .the register would be use�u1 later on. In 27 of these Por the purpose of freeing an adder it was necessary to copy it into the memory. Tn seven programs there was always a free adder and in 12 programs it was not necessary to copy adders in~o time variables. Data on the - most "inconvenient" pxngram: 579 strings, 2578 machine instructions, 530 adder settings, 17 of them (3.2 percent) cancPled out needed values, and in all 17 cases it was necessary to copy the adder into the memory. As an experiment the table of adders was reduced from eight to four elements. The regisrer distribution algorithm remained totally workable: Of 5=?41 setting in- structions 564 canceled out needed values. In 117 cases (2.2 percent of the total numbe~ of settings) it was necessary to copy the adder into the memory. The con- clusion can be drawn that for a locally optimizing compiler the presence of eight adders renders unneces~ary complicated algorithms for evading the triad graph and for the distribution of registers. In our opinion the optimization examples described in sec 6 can produce a great savings. The distribution of descriptor r~gisters is accomplished somewhat differently. The difference is that one descriptor can be used for access to several var.iables = and arrays; therefore it is difficult to determine from counters in triads when a certain descriptor becomes unnecessary. In the generator described descriptor registers are marked as free only at the end of a line section. An analysis of 10 programs (3370 strings in ~ORTRAN, 18,459 machine instructions) demonstrated that of 3320 descriptor register loading instructions 170 (5.1 percent) spoiled informa- tion which later had to be put back into registers. (The instruction generator uses 11 descriptor registers.) On average the result is not bad, but the behavior of the most "inconvenienC" program is consi.derably worse than the average: 103 - out of 577 (17.9 percent) of instances ~f loading into des~ripto~ registers de- ~ stroyed needed values. This program contairis accesses to subroutines with a great number of recurrent parameters. (In one line section are contained six accesses to a subroutine orI.th 34 parameters, and 29 parameters of the last of the six accesses tia3 been transferred Previausly.) With the available number of descriptor regis- ter.s it is impossible to avoid completely the expulsion of required info:mation, but knowl.edge oE the moment when inform:~tion in a register becomes unnecessary in this case woiild have reduced the number of "superfluous" loading events twofold (the tota]. number of loading events wot~ld have been reduced by approximately 10 percent). ror the purpose of optimizing cycles, i.nvariant and increment expressions (i.e., expressions which vary by a constant ariount after each completion of a cycle) are computed before entering a cycle (.n the cycle's prologue) and are placed in registers. 95 FUR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 N(?R OFFICIAI. USE ONI.Y 3. Distribution of Memorg The me*nory ~or local variables and arrays of the subroutine to be transJ.ated is assigned in a single segment insotar as this is possible. Then the following segment is used, etc. ~ Let us describe the algoxithm in greater detail. Loca1 variables are distributed in a segment in the ~ollowing order: - 1. Simple vz~riables o~ the real type and other types requiring one or two words. 2. Simple whole-number variabl~s. 3. Time variables of the CHARACTER type. 4. Simple logical variables ancl time variables of the BIT type. 5. Logic arrays, variables and arrays of the BIT type. 6. Variables and arrays of the CHARACTER type. 7. Whole-number arrays. 8. Arrays whose elements occupy not less than one word. At the start of a program is placed the instruction for setting the base descriptor - which indicates the point between variables and arrays; it has a word type of base. _ Simple variables and arrays are addressed with reference to this descriptor if it makes possible a 16=bit shi,ft in the instruction. ~'his distribution of variables is caused by the fact that in the AS-6 central processor the size of an instruction depends on the size of the shift in it. The size of the field for shifti.ng in an instruction varies from 0 to 2 bytes depending ~n the size of the shift. This memory distribution arrangement increases the number of variables with slight shifts relative to the base descriptor. As already mentioned, an individual segment is assigned for each common blork. The subroutine's instructions and its co~istants are distributed in an individual seg- ment. Writing into this segment is :inhibited by hardware; therefore, the poss~- bility of the erasure of. instructions is eliminated, as we11 as the possibility of chan~ing constants by means of.assignment to forn?al parameters. 4. Accessing of Elements of Arrays The AS-6 central processor instruction set ma4ces it possible to program accessing of an element of an array by several methods. The instruction generator described ~ uses only index registers ~or indexing and not half-word adders, indexing with which is performed more slowly. Tn addition, the instruction generator in the line section obviates the installation of new descriptor register~ since operations with descriptors are performed rather slowly. However, in the cycle prologue 96 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAL USE ONLY descriptors are established, inasmuch as is possible, ~or a11 arrays which are worked with in the cycle. 5. Creation of Machine Tnstructions . AS-6 central processor instructions have various modifications. For example, in the majority of 2-operand instructions the first operand must be in a register and the second can be in the on-line memory, in a register or part of a register (e.g., in the left ha1P oP an adder for operations with half-words), and also can be a constant speci~ied directly in the instruction (constants can be of two lengths-- ~ 1 and 3 bytes). Tnstructions for ~umps, the formation of descriptor registers, - etc., have a special format. Accordingly, the writing of an instruction in auto- code also has several formats. In view of this, the method of creating machine instructions used by us, making it possible to generate uniformly various sequences - of instructions, can be of interest. The GYeNKOM subroutine for the creation of machine instructions receives the follow- ing pa.r.ameters from the triad processing subroutinQ: reference to the so-called macroinstruction, and several macroinstruction operands (their numrer depends on ~ ti~e macroinstruction). An operand can be the number of a registsr or a reference to a triad or to an element of a table of characters. The macroinstruction consists of picturc~s of machine instructions. In a picture are described rhe operation (it is repr~~sented by a reference ~o a description of - the machine instruction) and operands oi~ the instruction in question. A certain operand can be the same for all uses of a given macroinstrt~=tion and then its value is presented in th~ picture. Otherwise the ciescription of the operand in the in- ~ struction's picture indicates the number of the parameter of the GYeNKOM subroutine which must he put in place of this operand. The description of a machine instruc- tion contai~:s an operation code (the symbolic name of the instruction in a~itocode) and several more bits demonstrating how the features of the result are cha~ged y when the ii.struction ia question is executed. The GYeNKOM suoroutine analyzes the macroinstruction an1 generates macnine instruc- tions, substituting in place of operands not specified in the macroinstruction - ~values conveyed as parameters. FurtheLmore, if the parameter is a reference to a triad it is ascertained whether the value of the triad is found in a Yegister and ~whether it is a constant which can be placed in the instruction. Various modifica- *ions of. the instruction are generated depending on the fulfillment of these con- i;:.Linns. 16. Optimization in the Instruction Generator The ~ol].owing actions are taken for the purpose of optimization in the instruction generator: 1. Jump instructions CSl1SlI1$ the rransfer o� control to the next instruction are eliminated. 2. Unneeded check instructions be~ore conditional ~umps are eliminated on account ~of. ti~e following of. instr.t~ction:~ whicli produce result control characters. 97 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407102/09: CIA-RDP82-00850R000500430018-4 : FOR OFFICIAL USE ONLY 3. The commutativity o~ opexations is employed, which makes it possible to avoid - unnecessary instructions ~or loading Tn~o adder regisCers. 4. The available instruction for th~ reverse subtraction af numbers with a floating point is used fiogether with tfie subtraction instructian. 5. Multiplication or division of a whole number by a power of two is replaced by an arithmetic shiPt. � 6. Ever}~where it is possible an oFerand which is a constant is placed directly in an instruction, which reduces the number of accesses to the m~mory. 7, When assigning a zero value, the zero write instruction is used, and when assigning a logical variable the value TRUE, an instruction for writing a 1 into the p~sition. 8. Parameters for ~unction-formulas are transferred ti~.xough value and not through reference. 9. If assignments are not made for the parameter o� a subroutine, the value itself is written into the place of the reference to the value of the parameter upon enter- ing the subroutine. 4. Comparison with the FOREKS Compiler for the BESM-6 The main advantage of the AS-6 central processor is the large virtua.l memory, which makes it possible to describe in a natural manner problems whose programming on a BESM-6 causes serious difficulties. Still another plus is the much iarger, as compared with the BESM-6, range of real numbers which can be represented and tlie hardware for working with double-precision numbers. _ As far as working speed is concerned, it is not much bettTr~ ~a~ whichedoEnot6� Data ~~hich we have accumulated demonstrate that FORTRAN p S employ operations with double precision are run on the AS-6 central processor an average of 1.2- to 1.5-Fold faster than on the BESM-6. Tn problems which inten- sively employ whole-number arithmetic the gain in speed is higher (close to two- fold). A threefold gain in speed was observed in logic tasks. Since the optimi- zation procedures carried out are the same for both computers and the design prin- ciples of the insbru~he�different~speeds of~the BESMa6fandeASe6icentralnprocessors caused basically y - However, in certain cases a program is run more slowly on the AS-6 central processor than on the BESM-6. The reason for this is the small number of postscript regis- ters. If in a cycle eight or more pages of the on-line memory are worked with a postscript will be retrieved constantly, which drastical?y reduces the speed. Let us cite the following figures. One program is run on the BESM-6 in 3 min 30 s. On the AS-6 central processor ir Cook 4 min 5 s to run. The �act was that in a short cycle nine arrays were worked with and they were a11 on di~ferent pages. When these 10 [as published] two-dtmensional arrays We~'P x'epl.aced by a single 98 FOR OFFICIAI. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFF[CIAL USE ONLY ~three-dimensiona]., as a resu~.t o~ which the el.ements which are woxked with at close moments o~ time were side by side, the running fi~me, in spite of the compli- cation of access to elements of the arrays, was reduced to 2 min 25 s. The sav- ings is quite considerable. i Thus, in writing a program it is necessary to pay attention to the fact that wor at close moments of time he done with close addresses, i.e., Chat the wox'king set be sufficiently sma1.1. ~his requirement relates to pro~ramming far any computer with a virtual memor,y. With frequent ~umping ~rom page to page, besides the retrieval of a postscript the transPer of data from the external storage into the on-line memory will often take place, which also slows down execution of the task. As far as rhF~ size of the programs generated is concerned, the instruction portion of program~ is approximately 20 percent shorter for the AS-6 central processor. S. TIiE AS-6 Central Processor as a FORTRAN I~'fachine The overwhelming majority of programs ru?i on a computer are written in a high-level language, i.e., machine instructions themselves-are generated not by a human being but by a compiler. Therefore, on the one hand in making a compiler every effort should be exerted to utilize the advantages of the machit~e and, on the other hand, in designing a computer it is necessary to pay attention to existing programming languages and methods of compilation, taking into account what the translacor can do and what it cannot do, what is easy for it and what is very difficult. FORTRAN is one of the most common languages and it is important to us to evaluate the AS-6 central processor from the viewpoint of FORTRAN, for a programmer writing in this language is interested precisely in the characteristics of a virtual FORTRAN machine and not of a real physical machine. On the whale the AS-6 central processor is quite successful as a FORTRAN machine. The virtual memory, the presence of a great number of directly addressable regis- ters, the advanced instruction set, the tinique state stack mechanism, the memory protection apparatus--a11 these represen* advantages of no small importance. It is necessary to mention thaC the AS-6 complex is designed for a broad range of applications. Therefore, if the AS-6 ceiltral processor is used only as a FORTRAN machine many c:apabil.ities of the processor will be unutilized. Of the 186 non- privileged instructions of the AS-6 central processor a FOREKS compiler can use bnly 80 (43 percent). For the s~ke of comparison, Let us say that on the BESM-6 the FOREKS uses 35 instr~~ctions out of 4~i (76 percent). - ~he program magazine (similar to the mag;izine in the BESM-6) available in addition - to the state stack is utilized poorly. 'I'he powerful mechanism for the virtual performance of operations, when an operaeion is performed in the usual manner but the result is written nowhere bi~t serves the purpose of generating control charac- ~ters, is utilized only to a sma~.l extent. The FOREKS compiler can gen~rate only two of these instructions: virtual setting and virtual subtraction. It seems to ;us that the other check instructions are also not necessary for a FORTRAN machi.ne, -and instead of virtual subtraction it is preferable to utilize a comparison in- struction whicin does not result in overflow. Of course, in the AS-6 central 99 FOR OFFICIAL l1SE ON1,Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFI('IA1. USE UNLY . processor there axe instructions i'or comparing whole numbers, bytes and bits, - but there is no instruction for comparing numliers with a~loating po~.nt. Meanwhile it seems to us that ~or the convenient implementation of a compiler from FORTRAN in the AS-6 central processor there are not enough setting and write in- structions and instructions for dividing numhers with double precision. Also in- convenient is the fact that the addition, subtraction and multiplication of numbers with dout,ie precision can be per~ormed only by having placed both operands in adder - registers. In [10] it is stated that a machine is good if a compiler can utilize its advantages , without high costs. The AS-6 central processor does not totally meet this require- ment. For example, it is difficult to select an optimum method for accessing ele- ments of an array. - Of course, the AS-6 central processor offers many opporL�unities for optimization, the utilization of which is fairly simple (they are listed above), and this is an _ important advantage of the machine. The results of a measurement of the frequPncy of th~ use of various instructions of the AS-6 central processor in the text of programs generated by a FOREKS compiler are given in table 1. The gathering and processing of statistical daCa were per- formed by means of the POPLAN system [11]. Several FORTRAN programs num~=ring a total of 2146 instructions were ~translated. With this 12,754 machine instructions wera generated. Fifty-four instructions of the AS-6 central processor were used. Table 1. Name of instruction Percentage o� use Format~on of descriptor 14.5 Writein of descriptor 11.4 Setting half-word adder 8�6 Setting adder ~'4 Writein of adder 6.6 ~ Writein of half-word adder 5.5 Setting saving mask 4.9 Jump wi.th return 4.9 ~ Settir~g index register 4.5 - Multiplication with floating point 4.0 Setting descriptor with connective 4.0 Addition of half-w~rd whole numbers 3.5 Addition with floating point 2�8 Index register writein 1�8 Setting descriptor 1�8 End of cycle 1.6 Suutractinn with floating point 1.5 Multiplication of half-word whole numbers 1.3 W~ite zero into ward . 1.1 Other instructions $�4 100 FOR OFFIC(AL t1SE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2447/02109: CIA-RDP82-00850R000500434418-4 FOR OFFI(:IAL USE ONLY The most used we,re the ins~ructions ~or ~oxming and w~iting i.n descriptox registers. They are used mainl.y in trans~erring parameters. Ttte ~orma,tion i,nstruction is rather slow, which has a negative effect on the running speed o~ FORTRAN programs. Let us point out also that virtual operaticns made up only 0.3 percent. Conversion of a number from a whole number into a floating was encountered 41 times and from a floating inC: a whole 5 times. The reverse subtraction instruction m~de up 20 percent of all cases of subtraction with a floating point. It was possible to use the shift instruction in 32 percent of instances of the multiplication and division . of whole numbers. The writein of a zero made up 15 percent of all word writein operations. y Bibliography 1. Shtarkman, V.S. "Svravnitel'nyy analiz translyator FOREX" [Comparative Ana- lysis of FOREX Compiler], Preprint No 129, USSR Academy of Sciences Institute of ApYlied Mathematics, 1978. 2. Shtarkman, Vik. S. "Lokal'naya optimizatsiya ob"yektnoy programmy v trans- _ lyatore Foreks" [Local Optimization of an Ubject Program in the FOREKS Com- piler], Preprint No 149, USSR Academ~~ of Sciences Institute of Applied Mathe- matics (IPM AN SSSR), 1979. , 3. Mikhelev, V.M. and Vershubskiy, V.Yu. ASTRA. Novyye vozmozhnosti yazyka" - [ASTRA: New Possibilities of the Language], Preprint No 61, IPM AN SSSR, 1976. 4. Chaykovskiy, M.G. et al. "Avtokod dlya tsentral'nogo protsessora sistemy AS-6" [Autcode for the AS-6 Central Processor System], ITEF-153, Moscow, 1976. 5. "Draft Proposed ANS Fortran (X3J3/76)," SIGPLAN NOTICES, Vol 11, N~ 3, t~ar 1976. 6. "Programmirovaniye na yazyke Assemblera YeS EVM" [Programming in the YeS Com- _ puter Assembly Language], Moscow, Statistika, 1976. 7. Tsikritzis, D. and Bernstayn, F. "Operatsionnyye sistemy" [Operating Systems], Moscow, Mir, 1977. ~ 8. Gris, D. "Konstruirovaniye kompilyatorov dlya tsifrovykh vychislitel'nykh - mashin" [L~esigning Compilers for Digital Computers], Moscow, Mir, 1975. 9. Aho, A.V., Johnson, S.C. and Ullman, J.D. "Code Generation for Expressions with Cotmnon Subexpressions," JACM, .'ol 24, No 1, Jan 1977. 10. Wirth, N. "The Design o~ a PASCAL Compiler," SOFTWARE - PRACT~CE AND EXPERI- ENCE, Vo1 7,, No 4, Oct 7.977,. 11. Bayakavskiy, Xu.M., V~yukova, N.T.r Galatenko, V.A. and Khodu7.ev, A~$. "Programmirovaniye na yazyke POPLAN" [Programming in PO~LAN1, ~~I~ AN SSSR, 1976. COPYRIGHT: Izdatel'stvo "Nauka", "Px'og'rammirovaniye", ].~)81 883]. CSO: 1863/45 101 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY ~ TJDC 681.3 SYSTEM OF DTALOGU~ PREPARA'~~ON OF TASKS ~OR UNTPIED SERTES COI~iTTERS Moscow PROGRAP'~lTROVANIYE in Russian No 5, Sep-Oct 81 (manuecript received 2 Sep 80) PP 68-73 [Article by V.V. Khanykov, A.V. Rybakov and N.V. Anan'ina] [Text] In this paper a description is given of an instrument aystem for working with a desigu library for Unified Series (YeS) computers in the operating system. ' The development, debugging and utilization of software in a YeS operating system assume the heavy use oP an assignment control language (YaUZ). The complexity of the software, the English language mnemonics and the "machine" orientation of the - YaUZ cause a great number of errors in assignments. The appar.atus of catalogued procedures only partly facilitates the user's situation. For the development of YeS computers the problem of facilitating communication ~ between the user and the YeS operating system assignment control system on the national level has become pr~ssing. ' In this paper a solutinn based on the creation of a psychologically natural input language is discussed and a description is given of the implementation of a pre- processor for the dialogue preparation of assignments for YeS computere. The rejection of a dialogue assignment remote input (DUBZ) system was caused by the ~ fact that the instruction language used in it requires a great number of input characters in describing an assignment and is oriented toward the English language. In addition, when working with a DUBZ system the possibility of parallel-series input is lacking and the effectiveness of the work itself is determined to a great extent by the configuration of the system and the number of users working simul- _ taneously. The features of the development of the input YaUZ by means of a preprocessor are illustrated in�the example of the creation of a fu~ction-oriented language for - working with a design library [1]. T~e procedur'e for speci~ying function atate~ ments is as ~ollows: An analysis is made of user's actions in the development and debugging of routines by means of a design library. = 102 ~ . ~ - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE OvLY In the us.er's waxk typi~a~, action~ axe ~ing7.ed out which axe usual7.r indep~ndent of the compu[er and are determined by~fihe s.pec~~i;cs o~ the axea o~ analys~s. Each typical action or combinat~on of th~m is assigned some #unct~on^oriented sta*_ement wherehy the variable parts of these statements correspond to real ob~ects on which the user plans actions in his assignment. The following can be placed under the heading of t}rptcal actions o~ a user when working with a design library: designation of a speci4ic library ~'rom a great number of permissible libraries in the s~rstzm, selection oP a specific section of a library, and the set o~ ~unctional operations relating to manipulating a selected ob~ect. The minimum set of function-orietzted statements for working with a design library must include the following statements: start af assignment, notation in design library, translation from library, editing o~ contents of library, start execution, circulation of contents of library for different machine media, etc. xhe variable parr,s in these statements specify the names of assignments, libraries and objects in the system. For convenience of the user's working with the system a set of service functions must be provided for the creation of the design library and for maintaining it in the working state. In addition, the service functions provide the user with all the necessary information regarding the state of the d~sign library and the system as a whole. The procedure for analyzing ~eS con~puter YaUZ statements for the user's work with the design library and for the formation of function-oriented statements is dis- cussed in [2]. ~ A set of Unified Series operatin~; system assignment control language. (YaUZ OS YeS) standard statements performing similar actions is specified for each function- oriented statement. The software converting function-oriented statements into standard assignment control language statements performs the role of a preprocessor #or the YeS operating system. The purpose and functions of the steps of preparing and performing an assignment in the process of the creation of software are ditferent. At the first step the user must formulate the sequence of actions in the assignment and describe it in the assignment contr~l language. At the second step these actions are impleniented in the computing environment oP a Ye5 camputer. The authors sugggest that minicomputers be used in the first step. For this purpose a dialogue assignment preparatio:z (DPZ) system has been deve.loped for YeS computers. The main objective of the system is for the user to be able to use function-oriented statements in the dialogue mode for working with the design library, as we11 as to offer him timely consultation in working with the assignment control language. Here it is necessary to make possible the recognition of .�un:.Ci.on--oriented atate- ments regardless o~ the national language. The structure of the software ~f the dialogue assignment preparation system was _ ehosen on the basis of this ob~ective (fig 1). The system can be in one of two 103 ~'OR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R040500030018-4 FOR OFFICIAL USE ONLY modes- "Assi.gnment Prepara~ion" ox "Cot~sul.Xation~~--~hexeby the ~~xst ~,s ~he bas~c mode. - - - _ _ /lpotpa~+nra ~ e~a~�~A 1~ ~ 8uanoaa /lpnapaMMa KoNCynemupyro- KuNmponq u{aA ~oacucmt- � om4rmos 2~ ` ~++a 5) CucmeMnaie npozpnn~nia tt- 7tKCmo4an uN NOpOI(UU ORCPQ' MOpUN Bqll KOM- ma6nuyer 3~ mapas AqJ u cynemupyroWt~r~ = gaKynitymoB 4~ na8cucmaMa � Figure 1. Key: 1. Dialogue maintenance routine 4. Assignment control language state- 2. Respanse check routine ~ ment and document generation rou- 3. Svstem tables tine ~ 5.` Consultation subaystem 6. Text information for con~ultation subsystem When it is necessary to obtain information the uaer can establish the "Consulta'tion" mode. For rhis he must begin the input of a xeaponse to any dialogue asaignmeat preparation question with a special character. In the system is provided the issuance of individual informatian on the purpose and syntax of variable information ~ontained in #unction-oriented statements, on the algorithm for usex's actione in the development and deDugging of ~outines, and also on working with the dialogue aesignment preparation system ~.tsel~. The changeover to the "Assignment ~'reparation" mode occurs automatically alter the end of a response to a user~s quest~on. The extstenCe o~ the two taodee makes it - u ~ 104 FOR OFFICIAL ~ iSE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY possible for an untrained us.ex nat on1y~ to p~~pare packages o~ asa~gnments inde~ pendently, but a~so Co learn the YeS computex o~erating system a~s~gnment contr~l language. Not the syntactic~.l analysis of functi.on-oriented statements ~ut selection according to the type o~ ~unc~ion (selecCion ~ram a 13.st) is used in the dialogue assignment preparation system. The table o~ ~unctfional operations on the screen of the display is shown in fig 2. This*method of implementing the recognitian of function-oriented statemPnts makes possible maximum independ~nce from the national language and re- duces the number of errors in writing an assignment. Tn addition, the problem of informing the developer regarding statements used in the dialogue assignment pre- paration system is solved automatically in this case. , PRTNTOUT TEXT GET INFORMATION CLEAR AR;A ON DISK PRINT HEADING OF LIBRARY PRINT HEADING OF LIBRARY PRINT HEADING OF LIBRARY CL SL RL COMt'RESSION OF LIBRARY SL COMPRE`~SION OF LIBRARY RL COMPRESSION OF LIBRARY CL REM~VAL FROM LIBRARI SL REMOVAL FROM LIBRARY RL REMOVAL FROM LIBRARY CL _ TRANSLATE INTO PL/1 TRANSLATE INTO PORTRAN TRANSLATE INTO ASSEMBLY LAN- GUAGE CHANGES PRINTOUT VTOC EXECUTE CATALOGUING IN LIBhARY SL OPTTMIZA~ION OF ASSIGN- CATALOGUING IN I.IBRARY CL MENT END OF FORMATION OF PRINT ENTIRE LIBRARY INDEPENDENT OPERATION . ASSIGNMENT , Figure 2. [Designations SL, RL, CL and VTOC are in the Roman alphabet; the rest is in Russian.] After the user selects a fL:-~ction the system initiates a dialogue, specifying re- fining questions in thz natural language relating to objects and the attributes of specific statements. The dialogue between the user and the d~,3logue aesignment preparation system is constructed on the basis of rul~es for the use of function- oriented st.atements. The us~r's responses in the "Assignment Preparation" mode pass tests based on the - requirements of the syntax o~ the YeS onerating system assignment control language. In the case of an incorrect response on the part of the user the dialogue assign- - ment preparation system diagnoses the error and sugg~$ts that he x~pe~t the iriput o� a response. On1y correct responses are entered in the appropriate table of th~ ciial~gue assignment preparation system. A table with the list o~ functions lights up on the display~s screen each time zCter tiie end oE the input of the required information relating to the preceding 105 FOR OFFI('IAL [J~E 01~lLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 NOR (]~F'NICfAL U~E ON~..Y �unction. This u~kes it possihl.e to construct packages o~ assignments nQt accoxd-- ing to a strict algarithm but in keeping ar~th the usex~s x~quirements. The purpose o~ the "Optimizati.on," "Get Tn~ormation" and "Tndependent Operation" - functfiot:s of the dialague assignment preparation s}*stem must be emphasized espec.ial- 1y. When it is necessary~ to optimize an assignment with respect to the time for execution on a YeS computer and/or wtth respect to the distribution of the magnetic disk storage, the "Optimization"' ~unctton must be indic.ated as the last function in creation o~ the package. The "Get Tn~ormation'' functian accomplishes the output to the display~'s screen and alphanumeric printer o~ the basic key concepts used in working GrTth the dialogue assignment preparat~on sy~stem. The "Tndependent Opera- tion" function makes it possible for the user to form directly on the display's screen the YeS operating syster.i assignment control language statements needed by him. Thereby the conditions are created for describing unique assignment steps, which makes it possible to realize the multifunctional capabilities of a YeS c~m- puter operating system. . Upon termination of the formation of an assignment the dialogue assignment prepara- tion system, based on the information stored in system tables, constructs YeS operating system assig,nment control language atat~ments and the documentation for the package created. Instructions for the computer operatar and a verbal descrip- tion of assignment steps are included in the documentation. An example of a dialogue between a user and the dialogue assignment preparation ~ system is shown in fig 3. A.z assignment for a YeS operating system (fig 4) was formed as the result of the sqstem's operation. The parallel-series input af in- formation is provided in the dialogue assignment preparation system for the ex- perienced user. This makes it possible for the user to input, in response to the first question, all the necessar.y information relating to the function selected. DP7. [dialogue assign- ment preparation sys- tem]: DPZ SYSTEM WORKING. ENTER NAME OF ASSIGNMENT. USER� RrIBOTA [in Roman alphabet] DPZ:� ENTER CLASS OF ASSIGNMENT. USER: ~ DPZ� CLASS OF ASSIG'.ZMENT DETERMINES PRIORITY OF RUNNING OF ~ ASSIGNMENT. SPECI??IED BY ROMAN CHARACTERS FROM A TO 0. USER: A DPZ : ENTER Nt1ME OF LIBRARY . US~R: PRO BA DPZ: ERROR IN ENTRY OF LIBRARY NAME. REPEAT ENTRY. USER� PROBA [in Roman alphabet] * * * * Then a tat~le with a J,i,st o~ ~unctians is output onto the d~,s~pl.ay ~ s screen. The usex indicates by means a~ the cursor the ~unetion "Cataloguing in ST~." * * * * [Continued on f~llowing page] 106 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFF[CIAL USE ONLY , DPZ: ~N~~R NAM~ Q~ ~EXT. USER: ~EX'~ [ in ~nsl,ish] DPZ: ~NT~R DATA. USER: TEX'~: PROC~DUR~ OP~TONS('MA~N~; [3:n Eng~ish) ~ ~ * ~ USER: ~ * ~ ~ - A tab7.e with a list oP ~unctions is output onto the display's screen. - The user indicates by means of the cur~or fihe function "Translate into PI,/1.. * * * ~ DPZ: ENTER NAME Ok' MODULE. _ USER: TEXT [in English] - * ~t * * . , A table with a list of functions is output onto the display's screen. The user indicates by means of the cursor the function "End of Formation of Assignment." Figure 3. //RABOTA JOB MSGLEVEL=(1,1),CLASS=A [in English] - CATALOGUING IN LIBRARY [in Russian] SL //RAB0001 EXEC PGM=IEBUPDTE,PARM=NEW //SYSPRINT DD SYSOUT=A //SYSUT2 DD DSNAME=PROBASL, - UNIT=SYSDA,DISP=MOD,VOLUME=SER=OSVALD //SYSIN DD DATA ~ ADD NAME=TEXT,LIST=ALL,LEVEL=OO,SOURCE=0 NUMBER NEW1=0,INCR=1 T~X'P: PROC OPTIONS(MAIN); , TRANSLATE INTO PL/1 [in Russian] _ f/~~' STEP1 - ~RANSLATTON [in Russian] //RABOOU2 EXEC PGM~I~M~AA,PARMANODECK,LOAD,STZE=52K~, . [Continued on following page] - 107 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/42/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY /f REGION~52K //SYSPRINT DD SYSOU~~A //SYSUT3 DD DSNAME~&&SYSUT3,UN~T~SYSDA, SPACE=(80,(250,250)), ACB=BT,KSTZE~8~ //SYSLIN DD DSNA~F~&&I,OADS~~,A~SP~ CMOD,PAS~),UNTTaSXSSQ, SPACEa(80,(250,10o)) //SYSUTI DD DSNAMEQ&&SY'SUTI,UNTT~SX`SI~A, _ SPACE=(1024,(60,60) � CONTTG), SEP=(SYSUT3,SY'ST~TN),DCBQBLKSTZE~1024 //SYSLIN DD DSNA'MEqPROSASL(T~XT), UNIT=SYSDA,DTSP~OT~D,VOI~'UMEq5~~~+OSVAT,D STEP2 - CATALOGUTNG IN ZIBRARY RT., [in Rtisstan] //RAB0003 EXEC PGM=TEWL,,PARM='NCAL',COND~(8,LT,'RAB0002) //SYSPRINT DD SYSOUT~A //SYSLMOD DD DSNAt'IE=PROBASL(TEXT), UNIT=SYSDA,DTSP=OLD,VOLUME=SER=OSVALD //SYSUTI DD DSNAMEn&&SYSUTI,UNIT=SYSDA,SPACE=(1024,(100,25)), - SEP=SYSLMOD, DCB=BLKSTZE=1024 //SYSLIN DD DSNAME=&&LOADSET,DTSP=(OLD,DELETE),UNIT~SYSSQ Figure 4. The dialogue assignment preparation system is implemented as a set of individual subroutines written in FOKAL. Therefore the addition of a new or the exclusion of an old function in the structure of the dialogue assignment preperation system does not present special problems and can be performed by the user. The following set of hardware is needed for the normal operation of the dialogue assignment preparation (DPZ) system: a minicomputer of the "Elektronika" type (or a CM-3 or CM-4) wlth an expanded on-line memory unit; a VIDEOTON-340 display; an alphanumeric printer of the DZM-180 type; and an IZOT 1370 magnetic disk storage. In conclusion let us formulate the distinctive features of the system suggested for the dialogue preparation of assignments for YeS computers. 1. The use of the national language in the dialogue is conducive to freeing the user from remembering complicated YeS computer as~ignment control language mnemonics and to reducing the time for tihe'preparation of an assignment. 2. The existence of two modes makes it possible for the user to obtain a con- sultation at any moment of t~.me in working w3th the system. 3. Checking o~ the correctnes~ of usex's responses e~.i,minates sy~?Ztactical errors in the prepaxation of an assignment, which is conducive to the e~~icient uti7.~za- tion of machine time ~ox YeS computers. 108 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R040500030018-4 FOR OFFICIAL USE ONLY 4. The number o~ chaxacters to be input b}* ~ u$ex in wox'king ~.th the APZ system is reduced dxas~ical.ly~. 5. Adaptation o~ ~he DPZ sys~em to tfie 1eve1 0~' th,e user~s px'ofess~;ona1. training is made possib~e by the euipl.oyment o~ a seTial or para7.1e1-ser~a7. data input mode, - as well as by means o~ the "independent Operation" ~unction. 6. The functional nature o~ subroutines makes it pos~sible to expand and develop the = DPZ system without a cardina7, change in its structure. 7. The tabular organization of data in the sys~em makes the following possible: the entry of changes in intormation entered withoufi waiting for the end of the formation of a package; the e�ficient construct3on o~ the sequence of assignment ' steps in a package; and the creation of uniform documentation conforming precisely to the assignment package, simultaneously with construction of the package itself. Bibliography 1. Baker, F.T. "Chief Programmer Team Management oP Proauct~on Progra~ning," IBM SYST. J., Vo1 11, 1972, p 56. 2. Anan'ina, N.V., Rybakov, A.V. and Khanykov, V.V. "Rationalizati~n of a User's Work Grith a YeS Disk Operating System in the Development of Programs," ELEKTRONNAYA TEKHNIKA, Ser. 9, No 1, 1980, p 35. COPYRIGHT: Izdatel'stvo "Nauka", "Programmirovaniye", 1981 8831 ~ CSO: 1863/45 109 - N'OR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ON~.Y UAG 681.3.06 ORGANTZATTON TN DISPAI~ OPERATING SYSTII~i OF DETERMINAT~ OUTPUT OF INFORMATION C1VER EPZTIRE FIELD OF OUTPUT UNITS OF MULTIMACHIN~ COMPUTIPIG COMPLEX Moscow PROGRAI~IIROVANTYE in Russian No 5, Sep-Oct 81 (manuscript received 28 Apr 80) pp 88-91 [Article by V.P. Petlinskiy and V.F. Tyurin] [Text] Questions are discussed, relating to the software simulation by means of an operating system of the total field of output units in a multimach~ne computing complex. A method is suggested for determinate output based on dividing the single common output waiting line into a number of separate waiting lines. The implemenCation of this metliod in the DISPAK operating syatem ia described. In the development of operating systems (OS's) for a multfmachine computing com- plex (MVK) traditional strategies and principles are often use3 for arganizing the operation of both the entire system and of individual components of it in a single machine. This results in the fact that many additional possibilities of the MVK - are unintentionally constricted or are not utillzed at all, both from the viewpoint of conveniences for the user and of the more intelligent organization of running a solution to problems, and fro~ the viewpoint of the effectiveness of the utiliza- tion of equipmer.t. For example, the traditional strategy for outputing information in the single- machine variant of the DTSPAK operating system is buffering in the calculation process, formation of a request for output and placing it in a single common waiting line for the end of the task and attending to requests, uniformly formulated and indistinguishable from the viewpoint of the operating system, from the common waiting line after the end of the task. Let us note that a request fnrmulated for output, after being placed in a common waiting line, doea not carry any additional information regarding its belonging to a specific task of the user. This sort of organization o~ output has become widespread in many operating systems, since it makes it possib~.e.easily to implement a multiprogram calculation mode and is class- ified as system outpu~ [1~. With system output di~~erent kinds oP in~carmat~.an intended ~or output to vari,ous types of pexipheral output units are distributed in the DISPAK operating system into their own output waiting line [2], and in the OS/MFT, OS/M11T and YeS [Unified Series] opera~ing system [3, 4] into thei.r own output class, similar to the ou+:put waiting line in the DTSPAK operating system. One or more output units orient%d - 110 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY - toward a specific typ.e of in~ozmation to be qutput are connected to ~ach output waiting 1.ine. ~hus~, in the ou~put buffer i.t ~s possibl,~ tq o7rgan?:z~ as many waiting l.ines as the types o~ autpufi units speci~ied ~,n gertexat~on o~ ~he operat~ ing system in a given equi,pmen~ con~tgurat$on. ~ Tn the DISPAK opexating sys~em ~:t i:s~ possi,ble to i,ndica~e ~our types of output units in generation o~ the operating sy~tem for a stngl.e computer: the ATsPU- 128/2M [alphanumeric pr~nter] t}rpe for rapid pr~ntout (up to two); the PT-80 type Por punched cards; the PI.-8Q type for punched tape; and the CALCONlP type for a graph plotter [5]. Ih the process of this study in the output buf_fer it was possible to organize up *_o four ~ndi~ridual output waiting lines for the number of types of output units. Let us discuss the procedure for attendin~ to each of these waiting lines. Re- quests from the waiting line are selected for service according to the rule "first come, first served" (FI~'0 [first in, first out] [6]). If not ~ust one unit works with the waiting line in question (output waiting line for rapid print- out), then units are selected for operation in cyclic order, by turns [2]. It is - obvious that in the latter case according to the arrangement described above for the organization of system output, t.he possibility of tying in the output from a specific task to a specific unit--determinate output--is eliminated. The same output strategy was used also in DISPAK operating system versions at work in BESM--6 multimachine computing complexes. Meanwhile the existence of a common output buffer for the MVK makes it possible to consider the combination of all output units of the complex as a common output field (a maximum of eight ATsPU-128/3M's, four PI-80's, four PL-80's and four CALCOMP-type graph plotters), but not Grith the hardware switching of units [7] for operating with a required computer of the complex, but by means of software switching or the software simulation of this switching. With this more flexible strategies for utilizing each imit in the total field are possible. The need for these strategies is occasioned by the large flow of information output f~om a11 computers of the complex requiring determinate output in the total field, as well as by the untraditional methods of processing output information in the dialogue terminal systems undergoing intense development in recent times. These strategies must make the following possible: ~'he establishment of priority service according to the "last come, �irst served" rule (LIFO [last in, first out] [6]) for specific categories of users. The putting together of a continuous pri.nted output listing (a roll) for the tasks of one or more subdivisions of users for a specific time period---the ad- ministration o~ outpu~. Distribution o� the output of users o~ the te~eprocessing sys~em avex xemqte user stations fuxnishe~ with xapid pr~.nting units. Storage of the output o~ speci~ic tasks o� users on a disk (tape) in the ~orm o~ an output file, but without actual printout for the purpose o~ subsequent pxocess- ing in the dialogue mode at a terminal. 111 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY Witti tt~e output o.f especi~t~.7.y ~raXuab7.e, inPormat~;on (e.~., i,n the machi,ne dupllca- tion of document$~ion on a unit wi:fih high pxi:nting qual,~,ty~) , khe ~end~ng o~ it to a uni.t set as.~de ~ox thi$ puxpo~e. Checking the opexati4n o~' the requi,red outpnt unit by* means o~ test tasks while solving other prob7.ems in lceeping ~th a set ~chedu~.e iri:thout disxupting the usual solving mode. For the purpose o~ enabl,ing the capabi],ities 3.ndtcated above, the output strategy in the DTSPAK operating system dE~scrihed above was changed. One of the main prin- ciples on which ~the new strategy is based is the separation of the single common output waiting 7.ine ~or each type o~ information finto a common line and a number of separate output waiting lines (a total of 32 waiting lines for each type of in- formation). The second important principle is the introduction of software-simu- lated switching of any of the output units of the total output f3eld of the MVK for working with only one of the dutput waiting lines. With this output strategy each separate waiting line receives requests for the output of specific tasks of users, and each waiting line is served by one (or more) preindicated unit. Additional tables of two types placed in the common disk storage have been introduced for the purpose of implementing this kind of service. The first type of table is represented by tables for the selection of requests (from a combination of key characters of the user's code) and for the distribution of requests selected to the required output waiting line. In the�same tables information is stored on the order for placing a request in the waiting line for service (FIFO or LIFO). The second type of table--the unit switching table-- contains information on which waiting line is served by a specific unit of the MVK's total output field. Tables of both types contain check information which is used for checking their contents from the entry of a random code both with computer hardware errors and with possible software errors. The employment of the method described above for organizing determinate system output has a number of important advantages over the alternate method implementing these possibilities through the system of working with a single common output waiting line. It is obvious that in this case each request sent for buffering to the common waiting line must carry an identifier of the unit out of the entire combination of units assigned to this waiting line which will receive the actual output. Later, when the common outgut waiting line is dumped, this identifier is used for referencing to a specific unit. � ' It is possible to single ~out three key advantages of the output strategy used. Fir~t, the tabulax method o~ distributing in~orsnation over separate waiting lines makes it possible to achieve great ~lexibility in possible distribution variants - at a lawer cost, si:nce there is no longer a need to revise operating s}rstem rou- tines for the purpose o~ ad~usting l;or ~he variant selected. T~or the required - distribution it is su~~iciE:nt to fi11 in the tables in the extexnal storage anew, which is easily acliieved by using service utilfities. 112 ~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAL USE ONLY Second, in the di,stx.tbu~fon o~' add~essed ~requeats ~xam a common W~iC~.ng 7.~ne the need arises. of xe~x'iev~ng, sorting and accessing rec~ues~ts ~xom the ~ta~tin~ ~ine - accordi^g to indi.vi:dual ~dentiffi,exs o~ unfts. The z~dditi4n oP thes~ a~.gorithins c~n increase to a grea~ ex~ent the amount o~ xoutines ~,n tRe apex~ting system. - Mea*~-ahile the speci:fi:cs o~ the runn~ng o~ opexating sy~ste~n xoutfines (strict re-- quirements with regaxd to the capz~c:~ty~ o~ ttie memor}r used) redu~e ar~d 3n our case totally e1im~.nate ~he possibility o~ us~ng algorithms of tli~,s sort. Algorithms with preliminary tabular distribution with subsequent si.mple serial access3ng are implemented considera~l}* more compactlp. Third, in our case overhead costs in the operating system i:or the star~up of schedulers of the operation o~ external units are reduced, since it is known be- ~ forehand whether there are requests for operation in the waiting line to which - the external units of a given computer of the multimachine complex have been - assigned. In the variant which is the alternate of this, startup of the external - unit operation scheduler must be carried out whenever the common line is not empty, since its functions include the sorting of requests and referencing to the - required unit. It can also be stated that for the output strategy employed the protection of in- formation from unsanctioned access and the LTFO rule for privileged requests for output are implemented easily. All 32 output waiting lines have a different status. The waiting line with the number 1 is the co;mnon waiting line. It receives all requeats for output w~ich , are not distributed by means of selection tables to any separate waiting line. Waiting lines with numbers b~ginning with 2 and ending with 20 are transient or on-line planning waiting lines. The aec~uence for the placement of requests in these waiting lines and the serving of requests from these lines can be altered pn line on the basis of up-to-the-minute requirements. The waiting line with the number 13 is used for the distribution and accessing from it of requests for test output, for testing the proper working order of the required unit of the MVK's total output field. The waiting line with the number 14 is used for buffer- ing requests for output which are later to be teleprocessed~ The copying of in- formation from this waiting line onto a teleprocessing disk (tape) is provided for in the system. The placement of requests into this waiting line is possible both by the usual method--by means of reques~ selection-distribution tables-- and through an extracode for storage of the output file in a teleprocessing disk ~(rape). The storage of these files in a gener.al archive of output teleprocessing files has been implemented at the present time. The organization of personal ~~rchives of use~'s output files is to be provided for later. This method can be ;used as an alternative metliod of the sheet-by-sheet interrogation of output in- formation described in [8). Distribution into this waiting line ~or administra- _ tive requirements also has been provided for tasks whj.ch end accidentally. Waiting lines with the numbers 15 and 16 are reserved ~or serving remote teleprocessing -stations ~urnished with raT~id printout ~xom output units o~ the total MVK field. Waiting 1.ines numbering 17 through 32 axe adminiatrative. The placement of re- quests for output and the c~utput o~ inEormation from theae waiting lines axe accomplished automatically. Furthexmoxe, in units set aside for sexving admini-- strative waiting lines the mode of outputing printouts for user su6divisions has 113 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY been implemented. Al.l the in~ormation of a schedu].~ ~ar a~a~x:~y~ 7.ong C~me period (up to 2 to 3 24 h p~x~:ods) is a.ssigned ~.n advance and i:s t~toxed ~:n sy~stem tables and ~cales. ~his~ ~n~oxl~C~on d~texmines groups of usexs to he sexved in this mode, the units. on which the sex~ring o~ adxninistxat3;ve w~a~:ting lines is implemented, ~ and the ~ol~owing oxdex o~ pexiods ~ox sexving user sul~div~.sions. The protection o~ in~or~fiion di,stril~uCed into any separate waiting line (through an access kep) and pratection ~xom ~.ts redistxi:l~ution into another wai:ting line _ are provided ~or $n the system. The ~ame proCectfion is provided also for units attached for serving a specific waiti~ng 1ine. Of course, the formation and correction of tables and scales for all five types of waiting lines is a labor-ir~tensive process which is difficult to perform by hand - without errors. The SYeRB (Servis Razgruzki Bufera [buffer dumping service]) - dialogue system was developed for facilitating the entire set of operations re- lating to enabling determinate output capabilities in the total field of MVK output units. This system includes a number of utilities which make it possible to obtain information on all waiting lines in a form convenient for perception on the screen of a display, to control on-line waiting lines, to organize archives of output files, to form, edit and enter into system tables a schedule for administrative waiting lines, and to set and cancel access keys for waiting lines. An instruction language is used for communication with users of the SYeRB system. Protection from unsanctioned access to the system through a key is provided for in the SYeRB system. Proposed as a further development of the system described is the creation of a combination of utilities for working with archives of user output files for the = purpose of their dialogue or package editing and output in processed form to the required unit. In conclusion let us mention that all the above-described capabilities of the DISPAK operating system and SYeRB system have not been realized in operating systems known to us both for the BESM-6 computer and computers of the YeS type [4, 9]. On the other hand, the organization of system output in keeping with the system described in this article can be used with success both for BESM-6 MVK's and YeS MVK's, as well as for MVK's of greater capacity. - Bibliography 1. Madnik, S. and Donovan, Dzh. "Operatsionnyee sistemy" [Operating Systems], - Moscow, Mir, 1978. 2. Zel'dinova, S.A., Koshkina, L.V., Ozorin, Xu.V�> '~yuxin, V�~� and Shulepov, N.I. "Structuxe and ~unctioning o~ the UIS~AK Opexa~ing System," TRUDX CHP~, No 136, Chelyabinsk, 1973. 3. Kattsan, G. Qperatsionnyye sistemy (pxagmaticheskiy podkhod)'~ [OPexating 11 roach~, Moscow, Mir, 1976. Systems, a Pragmatic App 114 EOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC'IAL USF. ONLY 4. T.'eledov. G.V� and Ray~ov, I~1D, ~'V'v~deni:ye ~v OS X'~S ~YM~~ I~~xpducti.on to ~ the Un~.~ied S~ries Cam~uter Opexating S~stemJ' Stati,s~~.ka~ 7,977. 5. T}rurin, V�F. , ed. 'bperatsionnaya s~.~tesaa D~S~AK d1,ya S~S2~1~6 (sf.ste~nmomu ' prograunnistu i operato~ru)" [DTS~'AK Opexatfng Sy~stem ~ox ~he BES1~I~6 (~or System Programmers and ^Jp@xators)], Mosco~r, T1~.' AN SSSR [USSR Academy of Sciences Tnsti:~ute of Applied Mathema~tcs], 1973. 6. Tsikritzis, D. and Bexnstaysi, "Opexafisionnyye sistemy" [Operating Systems]y - Moscow, Mir, 1,978. 7. Enslou, F.G. "Mul~tiprotsessornyye sistemy i para11e1'nyye vychisleniya" - [Multiprocessor S}rstems and Parallel Compufiations], Moscow, Mir, 1976. 8. Zel'dinova, S.A., Paremskiy, M.V. and Tyurin, V.F. "Nekotoryye bazovyye voz- mozhnosti OS DISPAK" [Some Basic Capabilities of the DTSPAK Operating System], Moscow, IPM AN SSSR, 1976. 9. Lomidze, O.N. and Silin, I.N. "Automated System for Buffering Results of the Calculation of Problems Utilizing Magnetic Disks in the 'AUBNA' Operating System for the BESM-6 Comt:uter," PROGRAMMIROVANIYE, No 3, 1978, p 56. COPYRIGHT: Izdatel'stvo "Nauka", "Programmirovaniye", 1981 ' 8831 CSO: 1863/45 li . i ' i i ~ 115 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY PUBLI~ATIONS UDC 629.7.017.2 CONTROL ALGOR7THNLS FOR SPAC~CRAFT Moscow UPRAVLENIYE DVIZHUSHCHIMISYA OB"YEKTArII NA OSNOVE ALGORITMA S MODEL'YU in Russian 1981 (signed to press 17 Apr 81) pp 2-8, 227-232 [Annotation, introduction, bibliography and table of contents from book "Control of Moving Objects Based on an Algorithm with a Model", by Igor' Mikhaylovich Sidorov, Lyudmila Yevgen'yevna Goncharova and Valeriy Georgiyevich Lebedev, Izdatel'stvo "Mashinostroyeniye", 1063 copies, 232 pages] . [Excerpts] Annotation This book reviews a new class of algorithms with models that are realizable in on- board computers included in the control contours of moving objects such as flying craft, the rolling stock of high-speed surface transportation systems, aad other complex mechanical systeru~. Considers algorithms with models that have a high level of adaptability to external disturbances and to a situation of limited infor- ma.tion. The potential of the algorithms described is demonstrated through the exa.mple of a number of problems of orienting and stabilizing controlled ob~ects. - The book is intended for engineers specializing in dynamics and control of flying craft. ' Introduction The use of onboard digital computers in the control systems of rockets and space- craft raises a number of new queations, among which are the desigaing and building of the computers themselves and their asaociated it~put and output data convertors. Another set of problems comprises the synthesis of control algorithms that can be realized on such computers and analyais of the quality of the processes of regula- tion in the closed system of the ob~ect and the regulator. This book will consider onl.y the questions of synthesizing algorithms to stabilize ob~ects with complex dynamic schemes. . - Including an onboard computer in the control circuit opens up new possibilities in ~ building stabilization systems that would be difficult to realize on the bas~s of analog equipment. The use of onboard digital machines makes it possible to. replace a number of stages required to snythesize a stabilization system with analog ele- ments by forming a computer procedure for the stabilization algorithm. In this sense formulation of the algorithm, which ultimately is a program for realization on the onboard digital computer, should be viewed as a atage in designing the stabilization system. ~ 116 FOR OFFICIAL USE ONLY ' APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 HOR OFFICIAL USF ONLY But despite the fact that the traditional process of design is replaced by the selection of an algorithm where an onboard digital computer (ODC) is used, this does not mean that the problem of synthesizing this algorithm can be solved by using a formalized computing procedure. On the contrary, construction of a stab- ilization algorithm that is realizable on an ODC opens up the possibility of malcing f uller use of the characteristics of the dynamic scheme of the object being regu- Iated. In this book the stabilization algorith~s are constructed for complex objects which include the equa.tions of connected-in oscillators in their dynamic schemes. For objects whose equations of movement correspond to the movement of an object as a "solid state," the use of an ODC in the structure af the stabilization system can also produce a certain effect, related to the feature of using computing technology. In this case questions such as the variability of the parameters of the stabiliza- tion system, drift, and the like, which are very important for analog technalogy, are practically excluded from considc~ration. In an ODC the restructuring of the parameters of the algorithm or a switch from one algorithm to another can be accomplished quite simply. These advantages are ~ common to the construction of any type of stabilization system for an ob~ect when ODC's are used. As for the probZem of synthesizing the stabilization algoritluns, the principles of construction and methodology for selecting the parameters of the stabilization alg~rithm have been worked out for ob~ects with simple dynamic schemes, and therefore it makes sense to use analog algorithms that have been transferred to ODC's. In this case well-known methods of the theory of discrete systems can be used for analytic study of the closed system, the object and the regulator. A greater impact should be expected from the use of ODC's to stabilize large objects with complex dynamic schemes. In this case the use of ODC's is 3ustified both from an economic standpoint and considering the fact that the introduction of new tech- nology should lead to expanded possibilities of solving new problems. Stabilization algorithms realized on ODC's make it possible to more fully meet the requirements made by the ob,ject for a stabilization aystem than where analog tech- nology is used. The requirements which the object makes for its stabilization system should be worked out on the basis of a comprehensive study of disturbing forces and moments acting on the object, analysis of interference in the senaors that determine the coordinates of the object, and a detailed description of the dynamic acheme of the ob~ect. Compiling the dynamic scheme of the ob~ect includes solving probl.ems of hydrodynamics that describe the oscillations of the liquid in the fuel tanks, prob- lems of elasticity theory which describes flexural oscillatione of the body of the object, and a description of the dynamics of the actuating organs of the stabiliza- tion system with due regard for their basic nonlinearities. It m3y be necessary to consider additional questions, for example to study the longitudinal oscillations of the object and the interrelationship between oscillations of the object in the _ stabilization planes and longitudinal oscillations. v The requirements made of the stabilization system are formulated on the basis of ex- perienGe studying ob~ects of different clasaes. Numerous monographs have aummar- ized the results of these studies [1, 16]. 117 FOR OFFiCIA.L USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY . Most of the published works devoted to the theory of optimal control concentrate their attention on construction of an optimal law of stabilization which insures minimum quadratic deviations of the coordinates of the object given the statistical characteristics of interference and the external forces and moments acting on the ob~ ect . Experience with the study of the dynamic schemes of various objects shows, however, that when a stabilization system is being built , in addition to influences on the selection of parameters of the system of external actions and interference the chief difficulties in selecting the structure and parameters, which are determ~ning for the makeup of the control system, depend on internal factors. The characteris- tics of the object's dynamic scheme, which includes a series of connected oscil- lators, are determining. The requirements made of a stabil.ization system vary in nature and often conflict with one another. Therefore, the process of designing the algorithm involves suc- cessively meeting a series of requirements, and cannot be reduced to the problem of cons~ructing an algorittun that optimizes a chosen all-embracing criterion in some particular way. The structure of a stabilization system using an ODC is set forth most fully in [2], where it is shown that the ODC permits more effective eynthe~is of a stabilization system for objects with complex dynamic schemes than does the use of continuous analog units. This work is able to trace the analogy in the calculations of dis- crete and continuous systems; this makes it possible to use the experience gained in the use of continuous systems for analysis and synthesls of discrete control sys- tems. The amplitude-phase frequency characteristics of the discrete stabilization algorithm obtained in this way are close to the corresponding characteristice of an algorithm realized on continuous units. It should be underlined that the method of synthesizing a stabilization system for ob~ects with complex dynamic schemes that is presented in work [2] is based on successive satisfaction of the requirements made of the stabilization system. In the initial stage of design the basic parameters of the stabilization system are selected: the amplification factor and differentiation time constant for the ob~ect whose dynamic system has been described in simplified terms by solid state equations without considering the attached oscillators. The parameters af the stabilization system are selected so as to insure the necessary quality of regulation processes when the object is acted on by disturbing forces and moments. The determining fac- tor in selecting the values of the parametere of the stabilization system is not the random components of the forces F~nd momenta, but rather the standard set of dis- turbing influencea. This aet describes the extreme values of the forces and moments, the gradients of disturbing influences, fixed initial conditions, the constant values of forces and moments resulting from maximum possible misalignment of the thrust of the engines, non:;yumaetrical diatributions of weight within the ob~ects, and the like. In the next stage the requirements for the amplitude-phase fequency characteristic of the stabilization system are fulfilled on frequencies that correspond to the oscillations of the attached oscillators. The specific features of a stabilization system rea].ized on continunus units and the fairly dense spectrum of frequencies and spread of the parameters of the dynamic ~ 118 ~ FOR QFFICIA~. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY system of the object away from nomimal values, which is an especially important consideration when designing a stabilization system, result in a situation where it is usually not possible to fulfill all the requirements of the ob~ect for its stabilization system. A compromise decision must be accepted. The most striking example in this respect is the case of nonstabilizability [2]], where the object makes conflicting demands of the stabilization system for neighboring freque:~cies. In this case the work of the stabilization system leads to unstable oscillations on.the frequency corresponding to the oscillations of one of the oscillators. In - other words, the stabilization system, which is expected to insure stability of movement of the object, cannot handle its assig~ments and additional design ele- ments, oscillation dampers, must be introducted to disperse the energy of the oscil- lations which is pumped through the stabilization system. The problems that arise during design of the control system are made more complex when building ob~ects that have multiple purposes and perform a whole set of tasks simultaneously, and also when the absolute dir~ensions of the ob~ect are increased. When the disturbed motion of such an ob~ect is described by a system of differen- tial equations, thE number of attached oscillators included in the dynamic scheme increases. The spectrum of frequencies of the system becomes denser and the ampli- tude-phase frequency characteristic of the stabilization system must meet certain requirements on each of the frequencies correaponding to oscillation of an attached oscillator. It should be observed that the desc:ription of the ob~ect by a dynamic scheme becomes less reliable as the ob~ect becomes more complex and its dimensions increase. This means not only an increase in possible deviations of the actual parameters of the ob~ect from those given in the system of equatians, but a~.so ~he fact that the physical premises included in the dynamic scheme of the object may not be entirely correct. The number of att~ched oscillatars included in the dynamic scheme for such a complex object is som2what indefinite. The requirements for the phase characterisiic of the stabilization system on the frequencies of particular attached oscidlltors also may not be adequately sub- stantiated. The principal goal of the present book is to show that the use of onboard digital computers in the control contour opens iip new npportunities for constructing stabi- _ lization algorithms of more complex structure, which can be used to support the regulation processes of contemporary space craft. Let us formulate the basic guidelines to employ in making up the stabilization - algorir_hm. During construction of the stabilization algorithm the basic principle, which is successive fulfillment of the series of requirements made by the ob~ect of its _ control system, is preserved, as in work [2]. It is unwise to pose the problem of _ synthesizing an algorithm that optimizes some global criterion, because many dif- ferent requirements made of the stabilization system must be considered. These conditions difier in nature, reflect highly diverse physical processes, and can be reduced to a general criterion only in artifical terms. The stabilization algorithm should be synthesized as a set of simpler elements so that to meet a specific condition imposed on the control system all that must be _ done will be to select the parameters of the corresponding element. It is also 119 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY essential that the selection of the values of these parameters have minimum impact on solving the other problems that arise during formulation of the object`s stabilization system. The multiplicity of tasks which are arising as space tech- nology ~evelops require building space craft of different classes and with differ- ent purposes. No single control algorithm can be designed, of course, to solve the problems of control and stabilization for such ob~ects. Nonetheless, the process of designing a stabilization system must be based on the general approach criterion. Greater complexity of the dynamic scheme of the object entails greater complexity in the structure of the stabilization algorithm. When synthesizing stabilization algorithms for complex objects it is necessary to consider this basic point as the design becomes more complex, the reliability of the description of the d3!namic system decreases. This circumstance reflects an objecCive trend in the develapment of space technology and results from the fact that it is necessary to use new design concepts and, consequently, new physical phenomen3 whose mathem~tical descriptions have not been adequately developed are possible. If the effect of introducing new technology is to permit a broadening of the capa- bilities of the control syatem, the stabilization algorithm realized on an ODC must take account of the inadequte reliability of the dynamic scheme of the ob~ect, and the increased complexity of the structure of the algorithm should be directed to greater use of the elements of identification of the structure and parameters of the object and to introducing elements of adaptation in the algorithm. There are also design requirements made of the the stabilization algorithm on the ODC. The algorithm that stabilizes the launch vehicles and space craft in the recovery and correction segments with the sustainer engines working must not contain iterative procedures. The algorithm should not include such problems as, for exam- ple, determining the roots of the characteristic equation, inverting matrixes, cal- culating special functions, and the like. It is most acceptable to reduce the stabilization algorithm to finite difference equations. This statement about the design features of the algorithm does not apply to such specific problems of space craft control as guidance problema, turns around the center of mass, and ~rientation problems, in other words to those cases where there are significant time intervals between engagement of the engines. For these problems it is often possible to use iterative proceaures to solve problems of iden- tifying the dynamic scheme of the object. Conclusion In conclusion we will set forth a few principles that can be used during work to synthesize algorithms to control the movement of a space craft. The system to con- trol the angular motion of a multifunctional space craft must have a broad r3nge of tasks, including orientation and stabilization of the object, performance of turns in space, guidance and approach during docking, and numerous others. During an extended space flight there may be changes i~ the configuration of the ob~ect, the composition of ineasurement and actuating organs, ~nd the mass, moment, strength, and other characteristics of the craft. Therefore, when constructing particular subaysteme that solve particular problems, their interaction in the overal structure of the control system of the space craft must be taken into account. During docking, for example, control of angular motion must be combined with shaping the approach tra~ectory, whereas the angular stabili- zation of the launch vehicles depends very little on the characteriatics of the 120 � FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 - FOR OFFICIAL USE ONLY trajectory in the active segment. Constructing the contxol system as a set of essentially autonomous functional units designed to perform particular tasks ma.y lead to significant problems in performing the flight of the space craft. Including an onboard digital computer in the control system makes it possible to construct the system on the heirarchical principle, because the central processor processes and selects the inforniation coming from lower-level subsystems and also produces and distributes control signals among these subsystems. Considering the many different tasks now being given to the control systems of long-lived space craft, the problem of constructing algorithms that are realizable on ODC's is a _ central challenge in synthesizing the control system. The basic requirement made of algorithms to control the movement of space craft is to perform their task with due regard for possible changes in the characteristics of the object ard in the composition of ineasurement and actuating organs. Optimal performance of the given maneuver by the ob,ject is desirable, but it should be ac- complished so as to fulfill this requirement. The degree of departure from the optimal solution is determined by the skill of the designer, The method of constructing a stabilization algorithm that has been proposed in this book may be used to synthesize algorithms to control other types of movement by space craft, in particular for problems of guidance and turning the space craft in space. It should be kept in mind that the method of constructing an adaptive algo- rithm with a model is not a formalized pro cedure, but rather depends on the essen- tial features of the problem being solved. The problem of adap ting the structure and parameters of the algorithm is solved by analyzing and iden tifying information that describes the movement of the ob~ect. The dynamic scheme corresponding to the angular movement of contemporary space craft is so complex that it is difficult to follow the traditional .approach to con- struction of an adaptive control system based on identification of the structure and parameters of the ob~ect's dynamic scheme and then modifying the structure and para- metei~s of the cnntrol algorithms, It has been shown in this book, through the exam- ple of constructing a stabilization algorithm, that identification involves expand- ing the observed signal into its constituent parts. The control action is formed on the basis af analysis of each of the components of the observed signa.l. In this case the algorithm is constructed so that there is no need for detailed study of the question of the causes of oscillating components or obtaining precise quanti- tative descriptions of them. The appearance of an oscillating component in the ob- served signa]. may be the result of flexural oscillations of the body or oscillations of the liquid filler, but this does not affect the responae of the adaptive algo- rithm, which is directed to damping the oscillating component. The purpose of this work has been to show how the algorithm is shaped, its ability to adapt, and a certain plasticity which is manifested in the process of synthesiz- _ ing the algorithm. It is assumed here tha t the degree to which the designer under-- stands the characteristics of the ob~ect's dynamic scheme and the designer's under- standing of the potential of the algorithm ma.y significantly affect the choice of a model, the selection of the type of filtering elements, and the organization of the computing process in the ODC. Choice of the model is conditioned on the mech- anical characteris tics of the control prob lem. The characteristics of the filter- ing elements, in turn, are determined by the requirements made by the ob~ect for 121 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY amplitude-phase frequency characteristic of the control system and by the ability to reorganize the structure of the algorithm in the adaptation mode. In principl the algorithm can also use more powerful filt.ers, but we should point out that our attempts to use recursive filters and smoothing weighted functions of the Gibbs multiplier type did not pro~*e useful in constructing an algorithm with adaptation. This pointed to the more general principle that any more refined testing tool is more critical of unforeseen changes in the structure and parameters of the control object. The book has deliberately emphasized the fact that it is not wise to formulate a rigidly regimented method of constructing the control algorithm, similar to the - solutions to mathematical problems. The principle of construction of the control algorithm must make it possible to employ the creative capabilities of the builders of the control system, their understanding of the dynamic characteristics of the object and their perception of the constraints and demands made of the control sys- tem that have not been formalized in the mathematical statement of the problem. FOOTNOTES 1: K. A. Abgaryan, and I. M. Rapoport, "Dinamika Ra.ket" [Rocket Dynamics], Moscow, _ "Mashinostroyeniye," 1969, 376 pages. 2. V. D. Arens, S. M. Fedorov, and M. S. Khitrik,~"Dinamika Sistem Upravleniya - Raket s Bortovymi Vychislitel'nymi Mashinami" [Dynamics of the Control Systems - of Rockets with Onboard ComputerQ], Moscow, "Ma.shinostroyeniye," 1976, 272 pages. 3. N. N. Bogolyubov, and Yu. A. Mitropol'skiy, "Asimptoticheskiye Metody v Teorii Nelineynykh Kolebaniy" [Asymptotic Methods in the Theory of Nonl~near Oscillations], Moscow, "Fizmatgiz," I.963, 412 pages. - 4. "The Onboard Digital Computers of Contemporary Launch Vehicles and Space Craft," VOPROSY RAKETNOY TEKHNIKI, 1970, No 7, pp 3-18. _ 5. F. R. Gantmakher, "Teoriya Matrits" [Matrix TheoryJ, Moscow, "Nauka.," 1967, 576 pages. 6. E. I. Gitis, "Preobrazovateli Informatsii dlya Elektronnykh Tsifrovykh ~ Vychislitel'nykh Ustroystv" [Data Convertors for Electronic Dig3tal Computers], Moscow, "Energiya," 1975, 447 pages. 7. L. Ye. Goncharova, B. I., Rabinovich, and I. M. Sidorov, "Construction of a Control System with a Strindard MAdel," DAN SSSR, 1973, Vol 213, No 5, pp 1037-1039. 8. E. Gumbel', "Statistika Ekstremal'nykh Znacheniy" [The Statistics of Extreme Values], Moscow, "Mir," 1965, 232 pages. 9. R. Kalman, and R. B'yusi, "New Results in Linear Filtration and Prediction Theory," TRUDY AMERIKANSKOGO OBSHCHESTVA INZHENEROV-MEICHANIKOV, SER. D., Vol 33, - No 1, IL, 1961. 122 FOR OFFICIAL USE ONLY . APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 - FOR OFFICIAL USE ONI,Y ~ 10. N. I. Kiselev, and I. M. Sidorov, "Statistical Estimate of the Global Ex- tremum," AVTOMATIKA I VYCHISLITEL'NAYA TEKHTiIKA, 1974, No 4, pp k5-49. 11. G. Korn, and T. Korn, "Spravochnik po Matematike. Dlya Nauchnykh Rabotnikov i Inzhenerov" [Handbook of Ma.thematics. For Scientists and Engineers], M~oscow, "Nauka," 1977, 832 pages. 12. N. A. Lifshits, V. N. Vinogradov, a.nd G. A. Golubev, "Korrelyatsionnaya Teoriya Optima.l`nogo Upravleniya Mnogomernymi Protsessami" [Correlation Theory of Optimal Control of Multidimensional Processes], Moscow, "Sovetskoye radio," 1974, 328 pages. 13. A. I. Lur'ye, "Nekotoryye Nelineynyye Zadachi Teorii Avtomaticheskogo Regulirovaniya" [Some Nonlinear Problems of Automatic Regulation Theory], ~ Moscow-Leningrad, GITTL, 1951, 216 pages. 14. V. G. Lebedev, and I. M. Sidorov, Algorithm for Turning a Space Craft with Elastic Elements at an Assigned Angle," KOSMICHESKIYE ISSLEDOVANIYA, Vol 12, Vyp 5, 1974, pp 797-799. 15. Dzh. Kh. Lening, and R. G. Bettin, "Sluchaynyye Protsessy v Zadachakh Avtomaticheakogo Upravlaniya" [Random Processes in Automatic Control Problems], Moscow, "Isnotrannaya literatura," 1958, 387 pages. 16. G. N. Mikishev, and B. I. Rabinovich, "Dinamika Tonkostennykh Konstruktsiy s Otsekami, Soderzhashchimi Zhidkost [Dynamics of Thin-Walled Besign Elements with Compartments Containing Liquids], M~scow, "Mashinostroyeniye," 1971, 563 pages. 17. Ye. P: Popov, "Prikladnaya Teoriya Protaessov Upravleniya v Nelineyniykh Sistemakh" [Applied Theory of Control Processes in Nonlinear Systems], Moscow, "Nauka," 1973, 584 pages. ~ 18. Ye. P. Popov, I. M. Sidorov, and I. P Korotayeva, "Division of Movements by the Harmonic Linearization Technique and Its Use to Synthesize Nonlinear Systems," in "Metody Sinteza Nelineynykh Sistem Avtomaticheskogo Upravleniya" [Methods of Synethsizing Nonlinear Automatic Control Systeme), M,oscow, "Mashinostroyeniye," 1970, pp 71-86. . 19. V. S. Pugachev, "Teoriya Sluchaynkh Funktsiy i Yeye Primeneniye k Zadacham ~ Avtomaticheskogo Upravleniya" [The Theory of Random Functions and Its Appli- cation to Problems of Automatic Control], M~oscow, "Fizmatgiz," 1960, = 652 pages. 20. B. I. Rabinovich, "Study of the Stability of Systems with Multiple Degrees of Freedom," IZVESTIYA AN SSSR, SER. TEKHNICHESKAYA KIBERNETIKA, 1964, No 4, pp 159-169. 21, B, I. Rabinovich, "Vvedeniye v Dinamiku Raket-Nositeley Kosmicheskikh Apparatov [IntroductLon to the Dynamics of the Launch Vehicles of Space Craft], Moscow, "Mashinostroyeniye," 1975, 416 pages. 22. A. P. Razygrayev, "Osnovy Upravleniya Poletom Kosmicheskikh Apparatov i = Korabley" [Fundamentals of Controlling the Flight of Space Craft and Ships], M~oscow, "Mashinostroyeniye," 1977, 472 pages. - 123 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 HOR OFFICIAL USE ONi.Y 23. I. M. Rapoport, "0 Nekotorykh Asimptoticheskikh Metodakh v~eorii Differentsial'nykh Uravneniy" [Some Asymptotic Methods in the Theory of Differential Equations], Moscow, Izd. AN SSSR, 1954, 287 pages. 24. B. V. Raushenbakh, and Ye. N. Tokar', "Upravleniye Oriyentatsiyey Kosmicheskikh - Apparatov" [Control of the Orientation of Space Craft], Moscow, "Nauka," 1974, 600 pages. 25. V. M. Rogovoy, and S. V. Cheremnykh, "Dinamicheskaya Ustoychivost' Kosmicheskikh Apparatov s ZhRD" [Dynamic Stability of Space Craft with Liquid-Fuel Rocket Engines], Mascow, "Mashinostroyeniye," 1975, 152 pages. - 26. I. M. Sidorov, and N. N. Balashova, "Control of the Movement of a Space Craft Entering the Atmosphere at Escape Velocity," KOSMICAESKIYE ISSLEDOVANIYA, 1973, Vol 11, No 3, pp 388-396. 27. I. M. Sidorov, and L. Ye. Goncharova, "The Stability of Nonstationary Control Systems with Nonlinear Correction Devicea," in "Problemy Navigatsii i Avtomaticheskogo Upravleniya" [Problems of Navigation and Automatic Control], Izd. VINITI, 1969, Vyp 1, pp 45-52. 28. I. M. Sidorov, L. Ye. Goncharova, and V. I. Prokhorenko, "Recognition of the Spectral Structure of Control Processes," in ibid., pp 53-59. 29. I. M. Sidorov, and I. P. Korotayeva, Study of Nonlinear Systems in the Case of Establishing ~tao-Frequency Processes'," IZV. AN SSSR. SER. TEKI~iICHESKAYA KIBERNETIKA, 1969, No 5, pp 148-158. 30. I. M. Sidorov, and I. P. Korotayeva, "Study of the Structural Stability of Mechanical Systems with Many Degrees of Freedom in the Presence of a Correct- ing Device," IZV. AN SSSR. SER. TEI~IIQICHESKAYA KIBERNETIKA, 1965, No 5, pp 183-187. 31. I. M. Sidorov, L. B. Krangacheva, and V. G. Lebedev, "Construction of a Stabilization Algorithm for a Deformable Space Craft Using an Onboard Digital Computer," KOSMICHESKIYE ISSI.EDOVANIYA, 1973, Vol 11, Vyp 3, pp 388-396. 32. I. M. Sidorov, and V. V. Timofeyev, "Analyaie of ~tao-Frequency Regimes of Os- ciJ.lations in a Stabilization Syatem," IZV. AN SSSR. SER. TEK~IldICHESKAYA ~ KIB1?RNETIKA, 1978, No 1, pp 174-182. 33. I. M. Sidorov, and S. V. Cheremnykh, "One Method of Studying the Stability of Regulable Syatems," INZH. ZHURNAL. MEKHANIKA 'TVERDOGO TELA, 1967, No 2, pp 74-80. ~ 34. N. V. Smirnov, and I. V. Dunin-Barkovskiy, "Kurs 1'eorii Veroyatnostey i Matematicheskoy Statistiki dlya Tekhnicheskikh Prilozheniy" [Course in Prob- ability Theory and Mathematical Statistics for Engineering Applications], Moscow, "Nauka," 1965, 347 pages. 35. A. V. Solodov, "Metody Teorii Optimal'nykh Sistem v Zadache Nepreryvnoy Lineynoy Fil'tratsii" [Methods of the Theory of Optimal Systems in the Problem of Continuous Linear Filtration], Mpscow, "Nauka," 1976, 252 pages. 124 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY - 36. K. Spidi, R. Braun, and Dzh. Gudvin, "Teoriya Upravleniya. Identifikatsiya i Optimizatsiya Sistem Upravleniya" [Control Theory. Identification and Optimization of Control Systems], Mnscow, "Mir," 1973, 214 pages. 37. Yu. T. Tu, "Tsifrovyye i Impul'snyye Sistemy Avtomaticheskogo Upravleniya" [Digital and Pulsed Automa.tic Control Systems], Moscow, "Ma.shinostroyeniy~," 1964, 703 pages. 38. Firmen and Tite, "Digital Stabilization System of the Titan 3S Launch Vehicle," VOPROSY RAKETTTOY TEKHNIKI, 1970, No 7, pp 42-58. 39. Yu. A. Tsurikov, "The Stability of One Dynamic System," INZH. ZHURNAL. MEKHANIKA TVERDOGO TELA, 1966, No 2, pp 193-195. ~ 40. Ya. Ye. Tsypkin, "Teoriya Lineynykh Impul'snykh Sistem" [Theory of Linear Pulsed Systems], Moscow, "Nauka.," 1963, 968 pages. 41. F. Chaki, "Sovremennaya Teoriya Upravleniya" [Contemporary Control Theory], - M~oscow, "Mir," 1975, 376 pages. 42. V. A. Shatalov, S. N. Seletkov, and B. S. Skrebushevskiy, "Primeneniye EVM v Sisteme Upravleniya Kosmicheskimi Apparatami" [The Application of Computers in Systems to Control Space Craft], Mpscow, "Mashinostroyeniye," 1972, 240 pages. Table of Contents Intr.oduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Chapter One. Objects of Regulation . . . . . . . . . . . . . . . . . . . . . 9 l.l. Equations of Disturbed Rocket Motion . . . . . . . . . . . . . . . 9 1.2. Functional and Structural Schemea of Stabilization Systems with Onboard Digital Computers . . . . . . . . . . . . . . . . . . . . 19 1.3. Selecting a System of Sensors . . . . . . . . . . . . . . . . . . . 25 1.4. Characteristics of the Dynamics of a Stabilization System with an Onboard Digital Computer . . . . . . . . . . . . . . . . . . . . 27 - Chapter 2. Methods of Studying Stability and Transitional Processes 32 2.1. Reducing a System of Equations to Canonical Form. Approximated Formulas for Roots . . . . . . . . . . . . . . . . . . . . . . . . 32 2.2. Criterion of Stabilizability of the Ob~ect of Regulation 43 2.3. Construction of the Solution Envelope. Determining the Ampli- tude of the Limit Cycle . ~ . . . . . . . . . . . . . . . . . . . . 48 2.4. Consideration of the Variability of the Coefficients of Equations of Disturbed Mfltion . . . . . . . . . . . . . . . . . . . . . . . . 57 2.5. Ztao-Frequency Oscillations in Nonlinear Stabilization Syatems 66 Chapter 3. Exa.mples of the Application of Optimal Control Theory to Stabili- zation Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.1. The Case Where the Ob~ect of Control Is Described by Solid _ State Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.2. Optimal Control with Consideration of Additional Degrees of Freedom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 125 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY Chapter 4. Stabilization Algorithm with Model . . . . . . . . . . . . . 90 4.1.. Sequence of Solving the Problem of Synthesizing a Stabili- zation Algorithm . . . . . . . . . . . . . . . . . . . . 90 4.2. Selecting the Method of Smoothing . . . . . . . . . . . . . . 95 4.3. The Case of Approximating the Input Signal with a Parabola by the Least Squa.res Method . . . . . . . . . . . . . . . . . 102 4.4. Scheme of the Algorithm with Model. Variation I...... 110 4.5. Finite Difference Equa.tion of Variation I of the Al~orithm. . 118 4.6. Calculation of the Roots of a Closed System Using the Transfer Function of the Algorithm with Mndel 128 4.7. Consideration of Aerodynamic Moment in the M~del 138 4.8. Construction of an Algor�ithm to Stabilize the Center of Ma.ss and Consideration of the Dynamics of Actuating Organs.. 141 4.9. Examples of Stabilization Processes Using the Algorithm - with Mndel . . . . . . . . . . . . . . . . . . . . . . . . . 147 Chapter 5. Algorithm for Active Influence on Oscillations of Attached Oscill.ators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 5.1. Formulating the Requirements Made by the Ob~ect of the Amplitude-Phase Frequency Characteristica of the Stabili- zation Algorithm to Insure Stability of the Closed System 154 5.2. Structure of the Algorithm with Model When Actively Influ- encing Oscillatinns of Attached Oscillators. Variation II. . 163 5.3. Finite Difference Equation of the Algorithm. Variation II. . 172 5.4. Two Examples of Constructing a St~bilization Algorithm 175 5.5. M~odification of the Algorithm in Variation II 181 5.6. System of Precise Orientation of the Ob~ect around the Center of Mass Considering Elasti~ Oscillationa 189 Chapter 6. Elements of Adaptation in the Algorithm with Model 196 6.1. Principle of Constructing an Adaptive Algorithm with - `M:3c1. Variation ZII . . , . . . . . . . � � � ~ � � � . � � 196 6.2. Realization of the Adaptive Algorithm on an Onboard Computer. 200 Chapter 7. Analysis of the Work Capability of a Stabilization System. . 205 7.1. Effect of Deviation in the Coefficients of the Model from Feasible Coefficients on Stability ~of the System 205 7.2. Study of the Noise Suppression of the Stabilization System. Nonlinear Element of the Saturation Zone Type 213 7.3. Nonlinear Element of the Type of the Relay with a Zone of Nonaen,sitivity . . . . . . . . . . . . . . . . . . . . . . . 220 _ 7.4. Optimization of External Diaturbances . . . . . . . . . . . . 224 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Footnotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 COPYRIGHT: Izdatel'stvo "Mashinostroyeniye", 1981 11,176 _ CSO: 1863/50 � 126 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USl: ONLY UDC 519.86: 681.3 r1ACHINE MATHEMATICAL MODELING Moscow REALIZATSIYA MATEMATICHESKIKH MODELEY NA EVM in Russian 1981 (signed to press 22 Jun 81) pp 2-5~ 143-144> 168-174 [Annotation, foreword, excerpt from chapter 6.1, bibliography and table of contents from book "Machine Realization of Mathematical Models" by Viktor Aleksandrovich Leont'yev, Izdatel'stvo "Energiya", 7,000 copies, 175 pages] [Excerpts] Annotation. ~ The problem of controlling a set of programs based on analysis of a real-time "demand-production model is examined. The criteria f~r effectiveness selected are those associated with model dimension, accuracy in modeling and the amount of nonpr~~ductive use of machine time. Effective algorithms are derived for solving. allocation and traveling salesman problems. The book is intended for engineering and technical personnel engaged in the developmenC of models and algorithms, and also for statistical analysis of economic efficiency in automated control systems. Foreword. The further upsurge in the national economy and the more complete satisfaction of the public's material and spiritual needs can be accomplished primarily through growth in the efficiency of social production and the acceleration of the scientif~c and technical progress [1,2]. A special place is now assigned to improvements in labor productivity, the rational utilization of material resources and manpower, and economical opening up and exploitation of natural resources. And in this business, an exceptionally ' imp~rtant part is being played by improvements in the planning mechanism and the ~,rga~izational forms of management, and alsn in economic accounting based on the development of state anc~ sector automated control systems and special mathematical models for solving the most urgent and complex socioeconomic problems. The fundamental principle in building control systems and large models is the SyStC=ill~i approach whose essence has been described in [3, page 455]: - 127 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC1At USE ONLY 1) formulating goals and clarifying their hierarchy before.the initiation of any activity associated with control; this includes decisionmaking. 2) obtaining maximinn effect in the sense of achieving the goals set with minimum cost by comparative analysis of alternative pathways and methods of achieving the goals and making the appropriate choice; 3) quantitative analysis (quantification) of goals and the methods and facilities used to achieve them, based on a broad and comprehensive evaluation of all possible and planned results of activity. Thus, the systems approach is a generalized, organizational-management principle characterizing the highest stage in the development of the control process over a given class of objects and it arms the developer and researcher with a powerful, standardized tool for analyzing and synthesizing systems that differ in terms of functions, character and structure and gives them a scientifically sound "template" for action. In accordance with this approach, during the first stage individual optimization problems and sets are resolved, special mathematical methods and approaches are worked out for problem solving and methods for analysis and measurement of the status of ob~ects, and particular control problems are studied, and so forth. It is with the solving of precisely these questions, and also with the methods for - realizing and computing operating conditions for machine models of complex processes, that this book deals. The basic problems selected by the author are selection of dimension for the ob~ects modeled, accuracy in modeling and minimization of nonproductive use of machine time. In accordance with this, methods and pathways have been worked out for achieving - set goals: an algorittun for selecting model dimension; a method for metareductions [metod.metaprivedeniy] for solving allocation problems, to which the problem of model.ing accuracy is reduced; algorithms for solving traveling salesman problems in enhanced dimension, in which a rational sequencing organization is set for the operation of units in the program set. It should be noted that the need to solve the problem of model dimension using machine methods has now come to a head; but work devoted to this question can be found neither in the Soviet nor the foreign literature. The method of inetareductions, based on iterative maximization of dual evaluation of allocations within a problem as a problem in linear programing with specific matrix constraints may be of interest to specialists in the the theory of algorithms and computing methods. In a number of instances, the features of this method have required the introduction of nontraditional terminology. The approaches developed to contrulling th~ realization of program sets are oriented largely on the application of economic dynamics in the models. However, problems of selecting model dimension and minimizing nonproductive use of machine facilities can be applied to a broad class of models. The method of inetareductions in discrete programing presented is not linked with the features of any particular models and can be used in all cases where the control pr.oblem for the process being considered is reduced to a problem of allocations. 128 FOR OFFICIAL t~SE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY The author expresses his deep gratitude to doctors of technical sciences Yu.M. - Fatkin and I.V. Goryachev, candidates of physicomathematical sciences V.A. Antipov, V.L. Arlazarov and Ye.A. Dinits, and candidates of technical sciences A.L. Uzdemir and I.I. Bronshteyn, who participated ~t various stages in examining and discussing the material forming the basis of the book; and also to T.N. Makarova and T.S. Rukina for their help in preparing the material for print. The author expresses his special thanks to candidate of technical sciences V.N. Buslenko for his important criticisms and advice, which have been taken into account in the final version of the book. Enquiries and recommendations should be sent to: 113114, Moscow, M-114, Shlyuznaya naberezhnaya 10, "Energiya" Publishing House. Chapter 6. Technical Means for Automating the Acquisition of Statistical Data. In this chapter we consider special devices that free the machine user from the need to carry out laborious, nonproductive work on acquisition, logging and initial processing of statistical i.nput data (that is, determination of matrix distances), and conduct certain procedures that have traditionally been done directly in the computer. Thus, in the devices described, a number of algorithms are derived for initial data processing, leading to a reduction in the volume of input data and ~thus to a reduction in computing operations. In other cases the devices are designed to carry out operations following which the search for optimal plans is accelerated. The devices described can be used in control of industrial output production, for example, filler (for drawing synthetic fibers), for analyzing specimens in studies of inetallic structures and geological samples, for doing electron microscope studies ~ on the quality of stored foodstuffs, and also in all cases where a large number of small-dimension objects in a study have heating temperatures, degrees of illumination, color and other attributes that differ sharply from the levels in the background against which the objects are placed. . From a design viewpoint, the devices are a set o~ standard eleGtronic com- ponents, which execute measurement and logic functions and which are controlled with the aid of a(control) program unit and memory. 2'he unit should be thought of as either a general-purpose or a control computer. BIBLIOGRAPHY - 1. "Materialy XXVI s'yezda KPSS" [Materials on the 26th CPSU Congress], Moscow Politizdat, 1981, p 223~ 2. Brezhnev, L.I. "Velikiy Oktyabr i progress chelovechestva. Doklad na sovmestn.om torzhestvennom zasedanii Tsentral'nogo Komiteta KPSS, Verkhovnogo Soveta SS~R i Verkhovnogo Soveta RSFSR v Kremlevskom Dvortse s'yezdov 2 noyabrya 1977 goda" [The Great October and Human Progress. Reported Delivered at a Joint Ceremonial Sessio~n of the CPSU Central Committee, USSR Supreme Soviet and RSFSR Supreme Soviet in the Kremlin Palace of Congresses, 2 November 1977], Moscow, Politizdat, 1977, p 30. 129 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 NOR OFFIC.IAI. USE ONLY 3. Gvishiani, D.M. "Organizatsiya i upravleniye" [Organization and Control], Moscow, Nauka, 1972, p 536. 4. "Metodologicheskiye voprosy postroyeniya imitatsionnykh sistem: Obzor/S.V. Yemel'yanov, V.V. Kalashnikov, V.I. Lutkov, B.V. Nemchirov" [Methodological Questions in Buildizlg Simulation Systems. A Collection of Works by S.V. Yemel'yanov, V.V. Kalashnikov, V.I. Lutkov and B.V. Nemchirov.], edited by D.M. Gvishiani and S.V. Yemel'yanov, Moscow, International Center for Scientific and Technical Information, 1978, p 88� - 5. "Ekspertnyye metody v sistemnykh issledovaniyakh" [Evaluation Methods in Systems - Researchj edited by USSR Academy of Sciences academician D.M. Gvishiani and corresponding member of the USSR Academy of Sciences S.V. Yemel'yanov, in "Sbornik trudov VNII sistemnykh issledovaniy" [Collection of Works by the All- Union Scientific Research Institute for Systems Research], Moscow, USSR State Committee for Science and Technology, 1979, 4th editi.on, p 88� 6. Yu.M. Fatkin, L.N. Blycskina, J.N. Dmitrieva et al. "Hierachical imitation demand-supply model. Preprints IFAK 75, Sixth Triennial World Congress, Boston/ Cambridge, Massachisetts, pt III D, session 62; Urban, Regional and National Planning, 62.5. 7. Avanesov, Yu.A. "Prognozirovaniye sprosa v roznichnoy torgovle" [Predicting Demand in the Retail TradeJ, Moscow, Ekonomika, 1975, p 103. 8. Likhtenshteyn, V.Ye. "Modeli diskretnogo programmirovaniya" [Models in Discrete Programing], Moscow, Nauka, 1971, p 328. n 9. Balas, E. An Additive Algorithm for Solving Linear Programs with Zero-0ne Variables," OPERATIONS RES., 1965, Vol 13, No 4, pp 517-546. 10. Semenova, V.A. "Opredeleniye ratsional'nogo varianta spetsializatsii sel'sko- khozyaystvennykh predpriyatiy v rayone" [Determination of the Rational Variant in Rayon Specialization of Agricultural Enterprises] in "Ekonomiko-matematicheskiye metody planirovaniya i analiza sel'skokhozyaystvennogo proizvodstva" [Economic- Mathematical Methods in Planning and Analyeis of Agricultural Production], Novosibirsk, Nauka (Siberian Branch), 1969, pp 93-115. ~ 11. Granberg, A.G. "Sel'skoye khozyaystvo v modeli optimal'nogo territorial'no- proizvodstvennogo compleksa" [Agriculture in a Model for the Optimal Territorial- Production Complex] in "Economic-Mathematical Methods in Planning and Analysis of AgriculturalProduction," Novosibirsk, Nauka (Siberian Branch), 1969, - pp 53-73. 12. Dudkin, L.M. and Vakhutinskiy, I.Ya. "Vzaimouvyazka optimal'nykh planov raznykh urovney upravleniya metodami iterativnogo agregirovaniya" [Coordination of Optimal Plans at Different Control Levels Using the Methods of Iterative Sequencing] Moscow, Nauka, 1975, p 256. 13. Dinits, Ye.A. "Algoritm porazryadnogo sokrashcheniya nevyazok i transportnyye zadachi" [An Algorithm for Bit Serial Contraction of Discrepancies and Transport Problems] in "Issledovaniya po diskretnoy matetnatike" [Studies in Discrete Mathematics], Moscow, Nauka, 1973, pp 46-57. 130 , FOR OF~'ICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR ~FFICIAL USE ONLY 14. Mikhalevskiy, B.N. "Model' narodnogo khozyaystva" [A Model of the National - Economy] in "Slovar'-spravochnik: Matematika i kibernetika v ekonomike" [A ~ Han~book of Mathematics and Cybernetics in Economics], Moscot~, Ekonomika, 1971, p 97. ~ 15. Dinits, Ye.A. and Kronrod, M.A. "Odin algoritm resheniya zadachi o naznachenii" [One Algorithm for Solving the Allocation Problem], DOKLADY AN SSSR, 1969, - Vol 189, No 1, pp 23-25. - 16. Entsiklopediya kibernetiki [Encyclopedia of Cybernetics], Vols 1-2, Kiev, Edited . by the Ural'sk Soviet Encyclopedia, 1975. p 374. 17. Bellman, R. and Endzhel, E. "Dinamicheskoye progr.a~nirovaniye i upravleniya v chastnykh proizvodnykh" [Dyna~nic Programing and Control in Partial Derivatives], Moscow, Mir, 1974, p 207. 18. Chernous`ko, F.L. and Banichuk, N.V. "Variatsionnyye zadachi mekhaniki i upravleniya" [Variation Problems in Mechanics and Control], Moscow, Nauka, 1973, p 238. 19. Korbut, A.A. and Finkel'~hteyn, Y~i.Yu. "Diskretnoye programmirovaniye" [Discrete Programing], Moscow, Nauka, ?972, p 368. 20. Kantorovicha L.V, and Gorstko, A.B. "Optimal'nyye resheniya v ekonomike" [Optimal Decisions in Economics], Moscow, Nauka, 1972, p 231. 21. Zangvili, U.I. "Nelineynoye programmirovaniye" [Nonlinear Programing], Moscow, Sovetskoye Radio, 1973, p 311. 22. Yudin, D.B. and Gol'shteyn, Ye.G. "Zadachi i metody lineynogo progratnmirovaniya" - [Problems and Methods in Linear ProgramingJ, Moscow, Sovetskoye Radio, 1964. . _ p 736. 23. "Statisticheskoye izucheniye sprosa i potrebleniya" [Statistical Studies of = Demand and Consumption], Moscow, Nauka, 1966, p 25S. 24. Lur'ye, A.L. "Problemy optimizatsii ekonomicheskoy sistemy" jProblems in Optimizing the Economic SystemJ, Moscow, ~~foscow State University, 196y, pp 3-25. 25. K~ssc~v, V.V. "riezhotraslevoy balanc" [The Intersector Balance]. Moscow�, Ekonomika, 1966, p 223. 26. Ivanov, V.V. "Voprasy t~chnosti i effektivnosti vychislitel'nykh algoritmov (Obzor dostizheniy v oblasti kibernetiki i vychislitel'noy tekhniki)" [Questions of Accuracy and Efficiency in Computing Algorithms. A Review of Achievements in the Field of Cybernetics and Computer Technology], Kiev, Ukrainian 5SR Publishing Commission, 1969, p 135. 27. Sergiyenko, I.V. "Metody ~rganizatsii vychislitel'nogo protsessa na vychislitel'nykh mashinakh. Obzor" [Methods of Organizing Che Computing Process in Computers. A Review], I~:iev, Ukrainian SSR Publishing Commission, 1971, p 1G2. 131 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAI, l1SH ONLY 28. Lipayev, V.V. and Yashkov, S.F. "Effektivnost' metodov organizatsii vychislitel'nogo protsessa v ASU" [Efficiency in Methods for Organizing the Computing Process in Automated Control Systems], Moscow, Statistika, 1975, p 255. 29. Papkov, V.I. "0 statisticheskom issledovanii programm bystrodeystwyushchikh vychislitel'nykh mashin" [Statistical Studies on Programs in High-Speed Computers], in "Vychislitel'naya tekhnika" [Computer Equipment], Leningrad, Energiya, 1972, Second edition, pp 61-67. 30. IVANOV, V.V. "Nekotoryye voprosy optimizatsii vychisleniy" [Questions of Computing Optimization] in "Matematicheskoye obespecheniye ETsVM" (Software for Digital Computers], Kiev, Ukrainian SSR Publishing Commission, 1972, pp 149-172. 31. Kavalerov, G.I. and Mandel'shtam, S.M. "Vvedeniye v informatsiannuyu teoriyu izmereniy" [Entry in Information Theory of Measurements], Moscow, Energiya, _ 1974, p 375. 32. Rozenberg, V.Ya. "Wedeniye v teoriyu tochnosti izmeritel'nykh sistem" [Entry in the Theory of Accuracy in Measuring Systems], Moscow, Sovetskoye Radio, 1975, p 304. 33. Sorenkov, E.I.; Teliga, A.I. and Shatalov, A.S. "Tochnost' vychislitel'nykh ustroistv i algoritmov" [Accuracy in Computing Devices and Algorithms], Moscow, ` Mashinstroyeniye, 1976, p 200. - 34. Venetskiy, I.G. and Kil'dishev, G.~. "Teoriya veroyatnostey i matematicheskaya statistika" [Probability Theory and Mathematical Statistics], Moscow, Statistika, 1975, p 264. 35. Gmurman, V.Ye. Rukovodstvo k resheniyu zadach po teorii veroyatno~tey i matematicheskoy statistike" [Handbook for Solving Problems in Probability Theory and Mathemattcal Statistics], Moscow, Vysshaya shkola, 1975, p 333. 36. Demidovich, B.P.; Maron, I.A. and Shuvalova, E.Z. "Chislennyye meeody analiza" - jNumerical Analytical Methods], Moscow, State Publishing House of Physics and Mathematical Literature, 1963, p 400. 37. Yudin, D.B. and Gol'shteyn, Ye.G. "Lineynoye programmirovaniye" [Linear Programing], Moscow, Nauka, 1969, p 424. 38. Gol'shteyn, Ye.G. and Yudin, D.B. "Zadachi lineynogo programmirovaniya transportnogo tipa" [Transportation-Type Problems in Linear Programing], - :~Soscow, Nauka, 1969, p 384. ' - 39. Kun, G. "Vengerskiy metod resheniya zadachi o naznacheniy" [The Hungarian Method for Solving the Allocation Problem] in "Metody i algoritmy resheniya transportnoy zadachi" [Methods and algorithms for Solving the Tranaportaion Problem], Moscow, Gosstatizdat, 1963, pp 92-107. 132 ~ FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPR~VED F~R RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 40. Florian, M. and Morton, M. An Experimental Evaluation nf Some Methods of Solving the Assignment Problem." CANADIAN OPERATIONAL R.ESEARCH SOCIETY, Ottawa, 1970 Vol 8, No 2, pp 101-108. . 41. Gavett, T.W. and Plyter, Norman V. "The Optimal Assignment of Facilities to Locations by Branch and Bound." OPERATIONS RESEARCH, 1966, Vol 14, No 2, pp 210-232. 42. Klein, M. "A Primal Method for Minimal Flow Costs with Applications to the Assignment and Transportation Problems" MANAGEMENT SCIENCE, 1967, Vol 14, No 3, pp 205-220. 43. Garfinkel, R.S. "An Improved Algorithm for the Aottonesk [sic] Assignment Problem." OPERATIONS RESEARCH, 1971, Vol 19, No 7, pp 1747-1751. 44. Leont'yev, V.K. "Algebraicheskaya struktura nekotorykh zadach diskretnogo programmirovaniya" [Algebraic Structure of Some Problema in Discrete Programing], in "Problemy Kibernetiki" [Problems of Cybernetics], 26th edition, Moscow, Nauka, 1973, pp 277-290. 45. Little, Dzh.; Murti, K.; Suini, D. and Kerel, K. "Algoritm dlya resheniya zadachi o kommivoyazhere" [An Algorithm for Solving the Traveling Salesman, " Problem], EKONOMIKA I MATEMATICHESKIYE METODY, 1965, No .l, pp 94-107. ' 46. Leont'yev, V.A. "Metod metaprivedeniy dlya zadachi o naznacheniyakh. Tezisy dokladov na VI Vsesoyuznoy konferentsii po ekstremal'nym zadachami" [A Method of Metareductions for Allocations Problem. Theses of Reports at the 6th A11- Union Conference on Extremal Problems~, Part II, Tallin, Estonian SSR Academy of Sciences Publishing Commission, 1973, pp 17-18. ' 47. Leont'yev, V.A. "Metod metaprivedeniy k resheniyu nekotorykh zadach kombinatornogo - tipa" [A Metareduction Method for Solving Certain Problems of the Combinatorial Type], KIBERNETIKA, 1974, No 7, Abstract 7G621, p 69. 48. Ore, 0. Teoriya grafov" [Graph Theory], Moscow, Nauka, 1968, p 352. 49. Popecku, Dorina Comsa. "An Algorithm for Solving the Transportation Problem. Economic Computation and Economic Cybernetics Studies and Research." Bucharest, 1976, No 2, pp 41-51. 50. Verkh~vskiy, B.S. "Simmetr.ichnyye mnogoindeksnyye transportnyye zadachi" [Symmetric Multiple Index 7'ransportation Problems] in "Sbornik: Problemy optimal'nogo planirovaniya, proyektirovaniya i upravleniya proizvodstvom" [Collection of Works: Problems in Optimal Planning, Design and Control in Production), Transactions of a Moscow State University Theoretical Conference, Moscow, Moscow State University, 1963, pp 482-498. 51. Khutsishvili, I.G. "Ob odnoy zadache razmeshcheniya" [An Allocation Problem], Sakartvelos SSR Metsniyerebata Akademiks moambe. Reports of the Georgian SSR Academy of Sciences, 1971, Vol 63, No 3, pp 557-559. ~ 133 FOR OFFICIAL t1SE O1~ILY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/42/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 52. Anisimov-Spiridonov, D.D. "Metod lineynykh vetvleniy i nekotoryye ego primeneniya v optimal'nom planirovanii" [A Method of Linear Branching 119�Some of its Applications in Optimal Planning], Moscow, Nauka, 1964, p 53. Anisimov-Spiridonov, D.D. "Metody i modeli bol'shikh sistem optimal'nogo planirovaniya i upravleniya" [Methods and Models for Large Systems in Optimal Planning and Control], Moscow, Nauka, 1969, p 360. 54. Gol'shteyn, Ye.G. "Transportnaya zadacha i yeye obobshcheniya" [The Transportation Problem and its Generalization] in "Sbornik: Metody i algoritmy resheniya transportnoy zadachi" [Collection of Works: t�iethods and Algorithms for Solving the Transportation ProblemJ, Mosr_ow, Gosstatizdat, 1963, pp 3-34. S5. Dualer, P. and Galler, B. "Primeneniye metoda privedennykh matrits k resheniyu obshchey transportnoy zadachi" [Applicati~n of the Reduced Matrix Method in Solving the General Transportation Problem] in "Methods and Algorithms for Solving the Transportation Problem," Moscow, Gosstatizdat, 1963, pp 115-121. 56. Revenko, V.L. and Oleyarsh, G.B. "A Problem in Allocation," KIBERNETIKA, 1972, No 2, pp 69-70. 57. Yefimov, A.I. and Maksimilian, S.V. "A Four-Index Transportation Problem" in "Matematicheskiye metody v ekonomik~' [Mathematical Methods in Economics], Kishinev, Shtinitsa, 1971, 3d editiun, pp 3-34. - 58. Staykova-Penkova, N. "TrTTansactions~ofntherHigheraEconomicsTlnstituteX Transportation Problem], Sofia, I, 1970 (1971), pp 1-19. 59. Pierskalla, W.P. "The Multidimensiunal Assignment Problem," OPERATIONS RESEARCH, 1968, Vol 16, No 2, pp 422-431. 60. Haley, K.B. "The Solid Transportation Probletn," OPERATIONS RESEARCH, 1962, Vol 10, No 4, pp 448-463. 61. Corban, A. "Un model multidimensional de transport" [A Model for Multi- - dimensional Transportation], Studii si cercetari matematice, Bucharest, 1972, 24, No 7, pp 1019-1082. 62. Junginger, W. "Ueber die Loesung des dreidimensionalen Transportproblems" [Solving the Three-Dimensional Transportation Problem], Diss., Dokt. Naturwiss, Univ. Stuttgart, 1970. 63. Leue. C. "Methoden zu Ordnungsproblemen" [Methods in Sequenc154 162blems], = Angewandte Information, Braunsc::weig, 1972, Vol 14, No 4, pp 64. Miller, C.E.; Tucker, A.W. and Zemlin, R.A. "I:lteger Programming Formulation of Traveling Salesman Prablem," J. ASSOCIATER FOR COMPUTING MACHINES, 1960, Vol 7, No 4, pp 326-329. 65. Bellmore, M. and Nemhauser, G.L. "The Traveling Problem [sic]: a Survey." OPERATIONS~RESEARCH, 1968, Vol 16, No 3, pp 538-558. 134 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY 66. Burkov, V.N. ar~d Lovetskiy, S.Ye. "Methods for Solving Extremal Problems of the Combinatorial Type (a Review)" AVTOMATIKA I TELEMEKHANIKA, 1968, No 11, pp 69-93. 67. Lawler, E.L. and Wood, D.E., Branch-and-Bound Methods: a Survey," OPERATIONS RESEARCH, 1966, Vol 14, No 4, pp 699-719. 68. Bellman, R. "Primeneniye dinamicheskogo programmirovaniya k zadache o kommivoyazher" [Application of Dynamic Programming in the Traveling Salesman Problem], in Kiberneticheskiy sbornik" [Collection of Works on Cybernetics], - Moscow, Mir, 1964, pp 219-222. 69. Lin, S. "Computer Solution of the Tr aveling Salesman Problem," BELL SYSTEM � TECHNICAL JOURNAL, 1965, Vol 44, No 10, pp 2245-2269. 70. Webb, M.H.J. "Some Problems of Producing Approximate Solutions to Traveling Salesman Problem with ]iundreds of Thousande of Cities," OPERATIONAL RESEARCH = QUARTERLY, 1971, Vol 2'l, No 1, pp 49-66. , 71. Eastman, W.L. "Linear Programming with Pattern Constraints," Ph. D. dissertatioii, Harvard, 1958. 72. Shapiro, D. "Algorithms for the Solutions of Optimal Cost Traveling Salesman Problem," Sc. D. Thesis, Washington University, St. Louis, 1966. ~ 73. Bellmore, M. and Malone, J.C. "Pathology of the Traveling Salesman Subtour- - Elimination Algorithms," OPERATIONS RESEARCH, 1971, Vol 19, No 2, pp 278-307. 74. "Sbornik: Metody i algoritmy resheniya transportnoy zadachi" [Collection of Works: Methods and Algorithms for Solving the Transport ation Problem], Moscow, Gosstatizdat, 1963, p 151. - 75. Kofman, A. and For, R. "Zaymemsya issledovaniyem operatsiy" [Operations Studies], Moscow, Mir, 1966, p 279. 76. Barachet, L.L. "Graphic Solution of the Traveling Salesman Problem," OPERATIONS RESEARCH, 1957, Vol 5, No 6, pp 841-845. 77. Karg, L.L. and Thompson, G.L. "A Heuristic Approach to Solving Traveling S:ilesman Probl.ems," MANAGEr1ENT SCIENCE, 1963-1964, Vol 10, No 2, pp 225-248. 78. Leont'yev, V.A. "Postroyeniye na zadannom mnozhestve tochek gamil'tonova tsikla, blizkogo pa dline k naukratchayshemu" [Building on a Given Number of Points in a Hamiltonian Cycle Close to the Shortest Len gth] in "Aktual'nyye voprosy tekhnicheskoy kibernetiki" [Ur gent Problems of Cybernetics Equipment], Moscow, Nauka, 1972, pp 244-248. 79. Author's Certificate No 331406 (USSR). A Computing Device for Solving the Symmetric Traveling Salesman Problem/V.A. Leont'yev,Published in BYULLETEN IZOBRETENIY, 1972, No 9. 135 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500034418-4 FOR OFFICIAL USE ONLY 80. Author's Certificate No 385279 (USSR). A Device for Solving the Symmetric Traveling Salesman Problem/V.A. Leont'yev. Published in BYULLETEN IZOBRETENIY, 1973, No 25. ' 81. Kheld, M. and Karp, R.M. "Primeneniye dinamicheskogo programmirovaniya k zadacham uporyadocheniya" [Application of Dynamic Programing in Ordering Problems] in "Kiberne~icheskiy sbornik" (A Collection of Works on Cybernetics], . Moscow, Mir, 1964, No 9, pp 202-218. � 82. Author's Certificate No 397915 (USSR). A Device for Sampling Oblique Ares on a Graph/V.A. Leont'yev. Published in BYULLETEN IZOBRETENIY, 1973, No 37. $3. Author's Certificate No 432550 (USSR). A Device for Scanning Two-Dimensional Parameter Fields/V.A. Leont'yev. Published in BYULLETEN IZOBRETENIY, 1974, No 22. . 84. Author's Certificate No 427351 (USSR). A Device for Analyzing Point Source Images/V.A. Leont'yev. Published in BYULLETEN IZOBRETENIY, 1974, No 17. 85. Author's Certficate No 402735 (USSR). A Method for Determining Distances Between Objects/V.A. Leont'yev. Published in BYULLETEN IZOBRETENIY, 1973, No 42. - Contents Page Foreword 3 � 6 Introduction " " Chapter 1. The Statistical-Optimization Problem of Selecting Model 12 - Dimensions 12 1.1 What model dimension is 17 1.2 How to select problem dimension .21 : 1.3 An algorithm for solving the dimension selection problem 1.4 A numerical example of dimension selection for problems and memory distribution in the program r.~its 30 1.5 Conclusions 31 Chapter 2. Statistics in Error and Modeling Control for the Accuxacy Criterion 32 2.1 A brief digression on the accuracy problem 32 36 2.2 Calculating error in the functional 40 2.3 Practical calculation of modeling error 2.4 Reduction of the statistical problem in control over modeling 44 accuracy in assignment problem 2.5 Conclusions 49 Chapter 3. The Method of Metareductions in Assignment Problem far Statistical Problems of Optimizing Modeling Accuracy 51 3.1 Status of the question of assigment problems 51 3.2 Calculating in.itial evaluations 52 3.3 Elementary metareduction 54 3.4 Complete metareduction 62 3.5 Studying the structure of�metareduction cost matrix 68 136 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL US~ ONLY 3.6 Evaluating the number of iterations when computing an optimal plan 73 3.7 Building optimal plans on metareduction cost matrixes 75 3.8 A numerical example in solving assigment problem with the _ metareduction method 80 3.9 Questions of machine realization of the metareduction method 83 3.10 Conclusions 87 Chapter 4. The Metareduction Method Applied in the Assignment Problem for Generalizing the Statistical Problem of Optimizing Modeling Accuracy 88 4.1 Formal description of the three-index assignment problem 88 4.2 Equivalent cost matrices 91 , 4.3 Elementary metareduction 93 4.4 Complete metareduction 99 4.5 Properties of inetareduction cost matrices 106 4.6 Calculating evaluation ~=L and search for optimal plan on the metareduction cost matrix 109 4:7 Conclusions 112 - Chapter 5. Statistics in Nonproductive Use of Machine Time and Controlling the Process of Minimizing It 114 5.1 Formulation of the problem 114 5.2 Calculating evaluations of minimal pathways 119 5.3 Branch in multiple decision trees 120 5.4 Cost matrix conversion in multiple plan branch 123 5.5 Excluding matrix elements in order to improve decision evaluation . in branch 124 5.6 One approach to increasing dimension in solved traveling salesman problems 125 5.7 Features of symmetric traveling salesman problems 131 5.8 Heuristic approaches to solving traveling salesman problem 134 5.9 Conclusions 141 Chapter 6. Equipment for Automating Statistical Data Acquistion 142 6.1 General reductions 142 6.2 A computing device for solving symaetric traveling salesman problem 144 6.3 A device for solving symmetric traveling saleaman problem 149 6.4 A device for sampling oblique ares on a graph 153 6.5 A device for scanning two-dimensional parameter f ields 156 6.6 A device for analyzing point source images 161 6.7 A method for automatically determining distances between braph.loci 165 Bibliography 168 COPYRIGH'T: Izdatel'stvo "Energiya", 1981 9642 CSO: 1863/53 - 137 _ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY UDC 658.512.2 AUTOMATION OF EXPLORATORY DESIGN (ARTIFICIAL INTELLIGENCE IN MACHINE DESIGN) ~ Moscow AVTOMATIZATSIYA POISKOVOGO KONSTRUIROVANIYA (ISKUSSTVENNYY INTELLEKT V ' MASHINNOM PROYEKTIROVANII) in Russian 1981 (signed to press 18 May 81) Pp 2-5, 298-302 _ [Annotation, foreword, conclusion of book "Automation of Exploratory Deaign (Artificial Intelligence in Machine Design)" by Aleksandr Ivanovich Polovinkin, Nikolay Konstantinovich Bobkov, Genrikh Yazepovich Bush, Valentin Georgiyevich Grudachev, Aleksandr Mikhaylovich Dvoryankin, Sergey Andreyevich Kudryavtsev, Petr Matveyevich Mazurkin, Vasiliy Vasil'yevich Merkur'yev, Mark Abramovich - Moldavskiy, Sergey Arnol'dovich Nikolayev, Gennadiy Sergeyevich Oshchepkov, Emma ~Pavlovna Sarkisova, Oleg Ignat'yevich Semenkov, Anatoliy Nikolayevich - Sobolev, Yevgeniy Arsen'yevich Smirenskiy and Yuliy Tsezarevich Faytel!son, � Izdatel'stvo "Radio i svyaz 10,000 copies, 344 pages] , . [Text] Annotation The new area of automation of exploratory design is explained, with primary attention devoted to heuristic and machine methods for finding new technical treatments. ' .The book is intended for engineers involved in developing and applying methoda of finding new design treatmenta, as well as other specialiats. Foreword ' - Intensive research and development has recently been underway for the creation and practical application of syatems for sutomating planning, design and technological production preparation for varioua classes of technical devices. The primary purpose of theae systems is to reduce the amount of~time and labor required to develop articles and, most importantly, to improve their quality. The goal of practically all developiuent work is to create and produce articles on a level with the best models in the world. However, this goal ia most often not achieved, since obtaining such articles requires inventing a complex of coordinaEed, as well as highly efficient, new technical treatments. This requires 138 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 MOR OFFICIAL USE ONLY synthesizing and analyzing numerous treatment versions, which for many reasons is difficult or impossible without using computers. At the same time, formalizing and programming the process of searching for (synthesizing) new technical treatments involves the statement and resolution of problems of technical creativity, and causes significant difficulties. ~he first automatic design systems made it possible to automate all stages and operations (making engineering analyses, finding required data, optimizing parameters of assigned technical treatments, comparing drawings and other technological planning documentation, testing experimental models, etc.), except for selecting improved and new planning and design treatments. Only recently have convincing results been obtained both here and abroad which make it possible to resolve the problem of automating exploratory design, or (according to the foxeign literature) to use artificial intelligence in machine design. We can say that, on the basis of results wb~ich have been achieved in automating exploratory design, development has begun on second-generation automatic design systems with improved cr~ative potential. There is - justification to assert that in the near future only second-generation automatic design systems which include exploratory design subsystems will in most cases make it possible to create new articles on a level with the wor?d's best models. The automated design system is one of the main and most important areas in which exploratory design methods are used. However, as will be shown below, this is not its only sphere of application. This book disseminates the long years of results achieved by the collective of authors, who have been developing the problem of automating exploratory design. The first three sections of the book explain individual, and we might say universal, methods which can be used independently from one another (or in various combinations) to solve various exploratory design problema in the area of instrument building, machine building and construction. Part 1 presents heuristic methods aimed at applications not involving computers; the possibility of using computers is indicated at the same time. S~:ctions 2 and 3 present special machine methods whose use without computers ~s inadvisable or impossible. Section 4 presents the conception of the creation of exploratory design sub- ~ystems for second generation automated deaign systems. The present book is not primarily a scientific monograph, but rather a practical aid for developing and apply.ing exploratory design methods and - systems. Besides specialists in the development and utilization of automated design systems, this aid will be of interest for engineers in all specialities, and accordingly to students at technical training schools who are interested in methods of finding new design treatments, and for inventors, patent holders and other specialists who are interested in problems of technical creativity and artificial intelligence. In this connection, instead of providing theoretical and _ experimental foundation for the methods, the book only provides references to the appropriate literature; brief bibliographic references are given rather than a 139 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY substantive review and analysis of the work of other authors; because of limited space, only the basic construction ideas are given for many algorithms, with references to the literature for more detailed explanation; many examples are limited to the initial data and end results, omitting the intermediate derivations. The problem in question has required integrated coverage of a broad group of interrelated problems, which resulted in detinite difficulties in the exposition. Considering that this is the first attempt to write such a book, the authors request to be forgiven in advance if some places are not ent~rely successful, and will be glad to receive any co~ents and remarks to improve the book. The authors are extremely grateful to Academicians V.M.Glushkov, G.I. Marchuk, I.F. Obraztsov and B.N. Petrov, to Ukrainian SSR Academician V.S. Mikhalevich, to USSR Academy of Sciences Corresponding Members M.A. Gavrilov and G.S. Pospelov, to G.P. Sofonov, President of the Central ~ommittee of the All-Union Society of Inventors and Rationalizers, to professors Yu. B. Borodulin, B.F. Goryunov, N.G. Zagoruyko, G.P. Zakharov, E.K. Kalinin, Yu.V. Kapitonova, I. Myuller, V.A. Myasnikov, A.I. Petrenko and D.A. Pospelov, as well as to engineer Yu.F. Morozov for their valuable critiques and for supporting research on the problem, - which undoubtedly helped to improve the manuscript. The authors thank the leadership of the Mariyskaya CPSU oblast committee and the Yoahkar-Ola CPSU city committee, primarily comrades V.P. Nikonov and G.N. Vodovatov for their years of assistance, which facilitated formation of the collective and supported its _ fruitful work. The authors also express their gratitude to official reviewers Professor O.N. Trifonov ~nd Docent A.Ya. Medvedev, who made a number of valuable comments on the manuscript which helped to improve portions of the book significantly. Conclusion This book presents three heuristic methods (Chapters 1-3) which have been tested in practice and which mosC designers can use without computers to solve many exploratory design problems. It is indicated simultaneously that these methods - can be formalized and programmed in moat cases, thus significantly increasing their efficiency. Special notice should be given the generalized heuristic method (Chapters 2,3) which can be used as the basis for obtaining a large number of modifications of specialized heuristic methods aimed at individual classes of technical systems and (or) types of exploratory design problems. Moat machine methods for exploratory deyign conaist essentially of isolating the functions of technical systems and their elementa and finding means for implementing these functions. In this connection, a methodology is developed and presented (Chapter 4) for analyzing the functions of technical systems and the construction of their functional structures the foundation for synthesizing physical operating principles and technical treatments. 140 FOR OFFICIAL U5E ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 F()R ()NNII"'IA1, 11~F: ONI.Y The book presents machine methods for ex~loratory design and their application for the basic three types of search problems singled out above: physical operating principles (Chapter 5), technical treatments for a given physical operating - principle (Chapter 6,7), and optimal parameters of a given technical treatment - (Chapter 9,10). The most effective and promising methods are those of synthesizing physical operating principles, as well as the direction of using mathematical programming approaches and methods to solve problems of finding new technical treatments (Chapter 8,9),(and, obviously physical operating principles), which is illustrated with the example of solving problems of synthesizing the optimal forms of technical system elements (Chapter 11). Although there is no doubt about the advisability of using the individual methods, combined and integrated utilization of these methode is more effective in designing certain classes of technical systems. In this connection, the conception of creating exploratory design subsystems for automated design systems is presented (Chapter 12,13). These subsystems are aimed at solving all three types of exploratory design problems, and have the appropriate software, hardware and information support. Special attention is devoted to substantiating the efficiency of exploratory design subsystems (Chapter 14). The results which have been achieved with regard to machine implementation of individual methods and using these methods as the basis for creating the first experimental exploratory design subsystem are now relatively primitive and inefficient, and if we consider the substantial amount of work on preliminary data preparation, they are not always competitive with traditional "manual" technology based on the trial and error method. Current attempts to automate exploratory design can be compared with the creation of the first artillery pieces, which were far inferior to the bow and arrow in terms of range, accuracy and rate of fire, as well as the first automobiles, which were significantly inferior to horse-drawn carriages and wagons in terms of capacity, speed and . -drivability. In this connection, the main result today consists of recognizing the requirement for automating exploratory desi�;n, providing theoretical and experimental proof of the possibility of solving this problem and providing a substantiated statement ~of practically useful goals and tasks. While it took several hundred years in the Middle Ages to convert primi.tive artillery pieces to efficient weapons, and it took several decades for the automobile to develop in the late 19th and early 20th centuries, if we consider the increasing rate of technical progress, we can ~assert that the road from today's primitive exploratory design systems to - efficient systems in the last quarter of the 20th century will be covered in :10 or 15 years. Let us attempt to formulate the basic directions of work which will facilitate the final formation and extensive practical application of machine technology for exploratory design. First of all, the approaches and methods for solving individual problems in the area of automating exploratory design which are 141 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 NOR OFFI('IA1. i1SF. ONI.Y considered in the book need to be improved (and even replaced with principally new, more efficient ones). In addition, we note the most important new directions of theoretical and experimental research and practical development~ 1. The practical application of individual methods and systems (or subsystems) for exploratory design involves extremely laborious work to create data bases, i.e., to prepare information and transfer it to machine media. This is especially clear, for example, in constructing a general AND--OR tree of technical treatments and then maintaining and updating it. In this connection, one of the main directions of the work consists of automating data preparation, or more precisely creating methods, algorithms and programs which reduce significantly the amount of labor required. 2. The quality of the'input data is very important in creating data bases. In addition, keeping in mind the fact that many information files are to a great extent invariant and will be used repeatedly in different areas of technology, it becomes necessary to develop standard forms for describing information for exploratory design systems. We have in mind the description of patent _ information, scientific discoveries, new physical (including chemical and biological) effects and laws, as well as new substances and construction materials, construction elements, etc. The standard forms for describing various units of information should basically be filled out by the authors who create the information, which will ensure high quality and facilitate essentially "automatic" development of data bases. 3. There now exist methods for automated solution of individual types of problems: 1) synthesis of new operating principles; 2) synthesis of new technical treatments for a given operating principle with the availability of several known technical treatments; 3) determination of optimal parameters of technical treatments. The following important problems, which facilitate complete automation of exploratory design, remain unresolved in the sense of efficient computer application: - using new physical operating principles (for which there exists no developed technical treatment) as the basis for designing acceptable technical treatments; statement of problems of optimizing parameters, and selecting a method of solution, for new technical treatments. 4. The basic body of information in the technical sciences consists of descriptions of fairly precise mathematical models (far calculating and analyzing certain technical treatments), and their theoretical and experimental ~ basis. New physical operating principles and new technical treatments have been evaluated intuitively, and occasionally experimentally, in engineering practice. In connection with automation of exploratory design, it has become necessary to develop methods of mathematical modeling for evaluating new physical operating 142 ~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY principles and technical treatments. The precision of these methods must facilitate the proper selection of only the best of several physical operating principles (or severai technical treatments), i.e., it may be significantly less precise than the traditional mathematical models used to - determine the numerical values of. the parameters of planned technical systems. 'S. Now that the first experimental autonomous exploratory design systems have been created (which have small data files, severely limited sets of solvable ~problems and small configurations of software and hardware), the next problem is ~to create the first practically useful efficient exploratory design subsystems which are organically interfaced with automated design systems, i.e., the problem of creating a second-generation automated design system [50]. Besides eliminating the shortcomings inherent in the first experimental systems, these subsystems should also devote special attention to service software, which makes it possible for designers to master a new tool quickly and easily. 6. Preliminary analysis has indicated that more than half of the software and information and methodical support used in exploratory design subsystems is invariant to the class of technical systems being designed. In this connection, the important problem arises of creating standard software and information and methodical support for exploratory design subsystems. These directions reflect the current level of knowledge and understanding of the problem in question, and naturally.do not pretend to be exhaustive even allowing for other directions and problems formulated in the book in the course of presenting individual problems. However, regardless of the limited nature of this knowledge, the primary determining direction has become obvious: reinforcing human intellect with the help of machines.. The creation of automated systems for exploratory design and-their integration with automated design systems, automated experimental and scientific research systems and information retrieval systems for ~atent and other scientific-tect~nical information will make i.t possible to facilitate the highest possible rate of technical progress and to maintain the status of a technically developed country. COPYR~GHT: Izdatel'stvo "Radio i svyaz"', 1981 6900 CSO: 1863/63 143 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFF[CIAL USE ONLY DATA PROCESSING EQUIPMENT ~ Moscow TEKHN'iCHESKIYE SREDSTVA OflRAHO'~KI INFORMATSII in Russian 1981 (signed to ~ press 17 Feb 81) pp 2~ 3Z9-320 ~ [Annotation and table of contenta of l~ook "Data Procesaing Equipment~~, by Vasiliy Niicolayevtch Kriushin, Ntkolay Matveyevicfi Surin, Valeriy Pavlovich Chuprikov and Nina Grip,,or'yevna Chernyak, Izdate~'stvo "Finansy i statistika", 12,000 copies, 320 pages] [Excerpts] This book~has been authorized by the USSR Ministry of Higher and Secondary Specialized Education as a tex~book for students at higher educa- tional institutions who are studying in tfie specialization "Organization of ~ Mechanized Processing of Economic DaCa." Annotation This textbook describes the operating characteristics, structural and achemati.c diagrams, and primary units of contemporary keyboard, keypunch, and small elec- tronic computer machines. It presents the metiiodological foundations of select- ing the set of hardware for automated control systems and organizing technical . servicing of such systems. Considerable attention is devoted to procedures and methods of work on~the machines and tfie metiiodology of programniing solutions to economic problems. The book can be used by a broad range of practical workers in the field of the application of computer teclinology. Table of Contents . , . . . . . . . . . . . . . . . 3 ~ Foreword . . . . . � � � � � � � � � � � � ' . . . . . . . . . . . Part One. Keyboard Computing Machines . . . . . . . . � 5 . � 8 Chapter 1. Adding and Computing Machines . . . . . . . . . . . . . . : : $ 1.1. General Description . . . . . . � � � � � � � � ~ � � � ' ' ' ' 1.2. Adding Machines . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.3. Computing Machines . . . . . . . . . . . . . . . . . . . . . . . . 13 Chapter 2. Sookkeeptng and Invoice Machines . . . . . . . � . � � � � � : : � 20 2.1. General Description . . . . . . . . . . . . . . . . . . . . . ' � 20 , 2.2. Class 170 Askota Bookkeeping Machine . . � � � � � � � � � � � � � 21 144 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 FOR OFFICIAL USE ONLY 2.3. Zoyemtron Electronic Invoice Macliine . . . . . . . . . . . . . . . 33 2.4. Iskra Electronic Invoice-Bookkeeping Ma.chine . . . . . . . . . . . 42 Part Two. Keypunch and Smal? Electronic Computing Machines 53 Chapter 3. General Description of Keypunch Computers . . . . . . . . . . . 53 3.1. The Keypunch Method and Basic Data f'rocessing Operations on Keypunch Computers . . . . . . . . . . . . . . . . . . . . . . . . 53 3.2. Classification of Keypunch Computers . . . . . . . . . . . . . . . 55 3.3. Punchcard and Coding Data . . . . . . . . . . . . . . . . . . . . 58 Chapter Keypunch Machines for Preparing Punchcards . . . . . . . . . . . -63 4.1. Keyboard Punches and Verifiers . . . . . . . . . . . . . . . . . . 63 4.1.1. Tlie PD45-2/1M Keypunch . . . . . . . . . . . . . . . . . . 65 _ 4.1.2. The P80-6/1M Keypunch . . . . . . . . . . . . . . . . . . 71 4.1.3. The PA80-2/1M Alphanumeric Keypunch . . . . . . . . . . . . 73 4.1.4. The KA80-2/1M Alphat?umertc Verifier . . . . . . . . . . . . 82 4.1.5. Work Procedures for Keypunches and Verifiers 86 4.2. Automatic Keypunch . . . . . . . . . . . . . . . . . . . . . . . . 88 4.2.1. The PI80-U Keypunch . . . . . . . . . . . . . . . . . . . . 89 ; 4.2.2. The PR80-U Reproducing Punch . . . . . . . . . . . . . . . 103 ' 4.3. Decoding Machines . . . . . . . . . . . . . . . . . . . . . . . . 115 ; 4.3.1. The RA80-2 Decoder . . . . . . . . . . . . . . . . . . . . 116 4.3.2. The RM80 Decoder . . . . . . . . . . . . . . . . . . . . . 118 ~ Chapter 5. Keypunch Machines for Ordering Arrays of Punch Cards 119 ~ 5.1. Sorting Machines . . . . . . . . . . . . . . . . . . . . . . . . . 119 ~ S.1.1.. The S80(45)-5M Sorting Machines . . . . . . . . . . . . . . 120 ~ 5.1.2. The S80(45)-7 Sorting Machines . . . . . . . . . . . . . . 130 5.1.3. Technique and Methods of Woric on S80(45)-SM and S80(45)-7 Sorters . . . . . . . . . . . . . . . . . . . . . . . . 132 I 5.1.4. The SAE80-3/1M Electronic Sorting Ma.chine 135 ~ 5.2. Collaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 i ~ Chapter 6. Keypunch Computera for Mathematical Data Processing 162 i 6.1. General Description . . . . . . . . . . . . . . . . . . . . . . . 162 ; 6.2. The T-SMiI Digital Tabulator . . . . . . . . . . . . . . . . . . . 163 6.3. The TA80-1 Alphanumeric Tabulator . . . . . . . . . . . . . . . . 209 6.4. Aggregating Tabulators with Automatic Keypunches . . . . . . . . . 260 6.5. Electronic Computing Attachments for Aggretation with Keypunch Computers . . . . . . . . . . . . . . . . . . . . . . . . . . 273 _ Chapter 7. Small ~lectronic Computing Machines . . . . . . . . . . . . . . . 282 - 7.1. General Description . . . . . . . . . . . . . . . . . . . . . . . 282 " 7.2. Primary Units of Computing Complexes . . . . . . . . . . . . . . . 284 145 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000500030018-4 FOR OFFiCTAL USE ONLY Part Three. Methodological Foundations of Selecting the Set o~E Hard- ware for the Automated Control System and Organizing Technicai Servicing 292 Chapter 8. Methodological Foundations of Selecting the Set of Hardware for the Automated Control System . . . . . . . . . . ~ . . � � � � � � 292 8.1. Stages of Work in Designing the Set of Hardware 292 _ 8.2. Methodology for Selecting the Set of Hardware . . . . . . . . . . 295 8.3. Calculating the Quantity of Hardware . . . . . . . . . . . . . . 299 Chapter 9. Organization of Technical Servicing and Repair of Computer - Eq'uipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 9.1. Plann~ng and Organizing Technical Servicing and Repair of Keyhoard and Keypunch Computing Machines . . . . . . . . . . . . . . 307 9.2. Organization and Planning Technical Servicing of Electronic Computers . . . . . . . . . . . . . . . . . . . . . . . . 314 COPYRIGHT: Izdatel'stvo "Finansy i statistika", 1981 11,176 CSO: 1863/51 s 146 FOR OFFICiAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFIC'IA1, IiSH; ONLY HOMOGENEOUS COMPUTER SYSTEMS, STRUCTURES AND DEVICES Moscow ODNORODNYYE VYCHISLITEL'NYYE SISTEMY, STRUKTURY I SREDY in Russian 1981 (signed to press 15 Jun 81) pp 2-11, 197-208 [Annotation, introdu~tion, chapter 9 and table of contents from book "Homogeneous Coc~puter Systems, Structures and Devices", by Eduard Vladimirovich Yevreinov, Izdatel'stvo "Radia i svyaz 10,000 copies, 208 pages] [Extracts] Annotation This book presents the fundamentals of the construcCion of homogeneous computing systems, structures and devices which use a model of the body of computers which is based on principles of parallel execution of operations, variable logic structure, and design homogeneity of the elementa and connections between them. The book is intended for scientific workers and engineers specializing in the area of computer technology and cybernetics. Sixty figures, 3 tables, 21 bibliographic references. Reviewers: Professor S.D. Pashkeyev and Professor Yu. M. Shamayev. ~ Introduction ~ Upon the appearance of the computer, it was clear that computer technology can be used in many areas of human activity. Subaequent research and practical - ~utilization of computers demonstrated that there ia no branch of the national economy in which the use of computer technology would not provide significant 'economic effect . It became clear that, along with power engineering, computer ~technology predetermines tne capabilities of society ta increase labor productivity. ilncreased labor productivity in a particular branch of the economy is c;~termined, first of all, by continual expansion of the sphere of 3pplication of computers and, secondly, by constant growth in the complexity of the problems to be solved. The requirement for computation has now increased to the extent that there is nosa a requirement for qualitatively new mass computing technology which combines high efficiency, reliability and computational economy wiCh convenience and simplicity of utilization. ~ 147 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFF[CIAL USE ONLY Mass computing technology and its technical and economic indicators. Mass computing technoTogy has a great deal in common with service technology. It must provide users with ~omplete computer support, low computational cost and high reliability, convenience and simplicity of utilization and servicing, etc. In other words, the computer must be just as simple and inexpensive as any or_ner everyday item: television sets, tape recorders, etc. No special con.~itions or special training are required to accommodate and use the computer. The requirement fo~ low-cost computation is foremost; providing a capability for augmenting computational resources as the comple:tity of the problems whic:~ are solved increases is also important. The direct relationship between labor productivity and the amount of computing resources and the'social requirement for continually increasing labor productivity lead to a requirement for continued increase in the amount of computer resources with overall cost limitations. Under these conditions, the creation of mass computing technology becomes ona of the most urgent problems at the current stage of social development. The specific computing capacity V~ is the quantitative estimate of mass - computing technology. By analogy with power availability, this indicator characterizes the availability to society of ccmputing resources per individual. The specific computing capacity is determined by dividing the total capacity of all computing resources by the total population of the country, and is measured ir_ the number of standard computational operations executed per unit of time per individual. - Another estimate of mass computing technolo~y is the specific computing cost Cp. This indicator is determined by dividing the total capital expendit~ires required to create the country's computing resources by their total capacity. The specific computing capacity and specific computing cost are the primary indicators which characterize .*_he level of development of computing technology. It is easy to understat~d that the annual expenditures for computing technology cannot exceed some defined portion of the national revenuea (5-10%). Of course, ~ with this limitation it is possible for V~ to increase contin~uously only if there is a corresponding continuous drop in C~. - Analysis of the trends of development of computer technology and requirements for computing resources to solve national economic problems in order to facilitate the required rates of increase in labor productivity allows us to ~ conclude that the specific computing capacity V~ muat approximately double annually up to 1 billion operations per second per individual. Then V~ can double every 10 years. Such an increase in V~ will fully satisfy the require:nents for computing resources, allowing for the rate of population increase in the country. This increase in V~ can be provided if CD is reduced to 1 ruble per 100 operations/second during the first 10 years, to 1 ruble per 148 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R040500030018-4 r FOR OFFICIAI. USE nNLY T 1000 operations,lsecond in the second 10 years, and finally to 1 ruble per million operations/second (Figure I.1). t Co,rubles /(ops/sec) Yo,ops/sec per individual ' 105 ~ ~o co yo . . 10~~ . f0-6 f03 ~ I } f9B0 1990 Z000 years~ ~ Figure I.1. With VQ and CD changing in this manner, three stages can be isolated in the development of mass computing technology. In the firat stage, with VD = = 103--104 ops/sec per person and CD = 0.01 rubles per operation per second, the required level of provision of mass computing technology is achieved and its cost reduced simultaneously by two or three orders of magnitude as compared with the cost of third-generation computing equipment. In the second stage, with V~ = 106 ops/sec arid Cp = 0.001 ruble, the requirement for solving mass problems while further reducing the cost of computer facilities through mass production of same is satisfied. In the thirti sCage, when V~ = 109 ops/sec and Cp = 0.000001 rubles, the requirement for solving complex problems while sharply reducing the cost of computing facilities by awitching _ to a new production technology is satisfied. Unlinked, weakly-linked and strongly-linked problems. All problems which can be solved using computers may be divided into three classes, depending upon the structural singularities of the links: unlinked, weakly-linked and strongly-linked. Unlinked problems represent a set (aggregate) of independent problems, each of which can be solved within some accepted tolerable time using the resources of one computer taken separately. In the simplest case, a set can contain a single problem. If we assume that the acceptable problem solving time is 104 sec (approximately 3 hours), then the maximum complexity of a problem to be solved amounts to 1010 1011 operations for third-generation machines operating at 106 10~ ops/sec. When the set includes problems with maximum complexity of 1011 operations and the problems are not linked, a solution can be achieved either by using the same computer serially, or by several computers operating in parallel. Weakly-linked problems represent a collection of problems which are interrelated in terms of information exchange such that the total volume of interactions for `the entire set of problems does not exceed the volume of computation for a ~ single problem. Obviously, weakly-linked problems cannot be solved directly - within the accepted tolerable time, since their complexity may significantly exceed the capabilities of a single computer, and it may be difficult to use several computers because of the information links between the problems. 149 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR OFFICIAL USE ONLY Furthermore, one important property of weakly-linked problems should be pointed out. Because of the small amount of linkage between problems, there are methods which can be used to transform a weakly-linked problem to a set of unlinked problems requiring a slightly larger amount of computation. _ Strongly-linked problems represenr a set of problem~ which are interrelated in terms of information exchange such that the total amount of interactions for the entire set is bounded by the amount o� computation for one problem and the amount of computation for the entire set. It is impossible to solve such a problem on a single third-generation computer within the tolerable time, and - the lack of inethods for dividing a problem into independent parts makes it impossible to solve these problems using a set of independent machines as well. Unlinked problems are mass problems. In order to satisfy the demand of the national economy for the solution of mass problems it is necessary to facilitate an increase in the overall performance of the country's computer pool; the performance of unified computing resources must be improved in order to provide for solving weakly-linked, and especially strongly-linked, problems. Since complex problems involving a large amount of computation (strongly-linked problems) are encountered significantly less often than simple (unlinked) problems, these two are ciose to one another in terms of the total amount of computation required. In this connection, it is useful ta pose the problem of - developing unified computing facilities which are identically convenient for solving both a large number of unlinked problems using separate computer resources, as well as a small number (within a single problem) of strongly- linked problems using generalized computing resources. Model of computer collective. In order to solve all three classes of problems (unlinked, weakly- and strongly-linked) it is necessary to turn to a . computational model which is qualitatively different from the traditional computer model. A computer model is the product of formalizing the actions of a human cnmputer engaged in problem solving according to predefined formal so lution rules. The computer model is built on principles of serial execution of operations, fixed computational logic structure, and design heterogeneity of the basic sections of the model and connections between them. The computer model provided the basis for the construction of first-, second- and third-generation znachines. An essential shortcoming of the computer model is the presence of a theoretical limit for ~omputational capacity resulting from the finite information transmissiua rate between elements of the model when operations are executed serially. As the theoretical limit is approached, the technical problems involved in creating a computer with the required parameters increase, and costs - also go up sharply. As a result of the foregoing, r~~gardless of a fairly high theoretical computational capacity limit (109 ops/sec), we must limit ourselves to a~technical limit ~~f 10~ ops/sec for third-generation mach ines. 150 ~ FOR OFFICIAL USE ONLY . APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007142/09: CIA-RDP82-40854R040500030018-4 ~ FOR OFFIC[AL USE ONLY t Another shortcoming of the computer model is the relatively large expenditures on software and hardware with limited performance, which results in a high specific computational cost. Even for the most sophisticated third-generation computers the specific cost amount to between 1 and 10 rubles per operation per second, and cannot be reduced significantly. The computer collective model proposed by the author in the late 1950's and early 1960's for constructing efficient high performance computer systems differs in principle. This model is the result of formalizing the computational process executed by a group of computers in solving a unified complex problem. The complex problem is represented as a set of interrelated parts. Each part is solved by an individual computer, and information is exchanged when necessary with other computers which are then solving their own parts. ~he computer collective model is based on principles of parallel execution of a large number of operations, variable logic structure and design homogeneity of the elements and interconnections. When using this model, the computing equipment is represented as an aggregate of computer models combined into a unified computer collective model with the help of information exchange between one another. In solving a complex problem, its component interrelated parts are distributed among the computers and solved in parallel. The structure of the information exchange network is-established in accordance with the schPme of interactions between sub-problems. In solving a compl.ex problem which represents an independent set of simple problems, the latter are distributed among computers for parallel processing. There is practically no exchange between computers. , The computer collective model has no theoretical performance limit thanks to the assumption of a thearetically unbounded increase in capacity by adding additional computers. The variable logic structure and design homogeneity make it possible to achieve high reliability indicators and economy for the computer collective model. Computer facilities based on the computer collective model can be used to solve all three classes of problems. The problem is actually to mass-produce computers which are convenient and easy to use, reliable, economical, and suitable for combination into a computer collective model for solving complex - problems requiring major unified computing resources. The possibility of creating a unified physical-technological base and organizing mass production of a small number of types of computers allows us to hope for an optimal solution Eo the problem of satisfying the demands of the national economy for computer technology. The theoretically unbounded capability for increased capacity of computer devices based on the collective model while simultaneously facilitating conditions for fiass production makes it possible to achieve the required specific computing 151 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFF[CIAL USE ONLY capacity while keeping the required specific cost within the necessary limits in accordance with the demands of the national economy. At the present stage of development of computer technology, a transition to creating computer facilities based on the computer collective model is becoming realistic thanks to the achievements of microelectronics in creating inexpensive, reliable, small processors and microcomputers. It should be noted that the computing resources of an individual.microcomputer are limited. This amplifies further the trend toward constructing computer facilities on the basis of the computer collective model. The difficulties in creating efficient software are also reduced significantly here as a result of constructing distributed means for controlling computati.onal processes. . The changeover to computer facilities consisting of a large number of relatively simple computers of the same type with regular interconnections makes it possible to simplify significantly their development, manufacture, debugging and operation. All of this points toward the advisability of switching over to mass computing facilities constructed on the basis of the computer collective model. � - Homogeneous computing systems, structures and facilities. The technical foundation for implementing the computer collective model is the integrated direction of homogeneous computing systems, structures and facilities which was proposed and theoretically justified by the author in the late 1950's and early 1960's. Now, after more than 20 years' devel~~pment of this direction, it has become obvious that it provides the t,asis for constructing mass computing facilities intended for solving unlinke~"l, weakly- and strongly- linked problems. This direction includes many classes of computing ~acilities: distributed computer systems, concentrated computer systems, ho:~ogeneous computer structures and homogeneous computing facilities. Distributed computer systems represent an a~gregate of spatially separated computers (or computer systems) which are interconriected via communications systems such that the required interchange between system elements is supported. A distributed computing system can operate in th~ computer network mode, in - which a set of independent problems is solved, as well as in a dedicated mode in which a single complex problem is solved using the overall computing resources of the system. The limiting case of a distributed computer system is the unified distributed computer system, which combines all of the computer resources of the c:ountry and uses them to solve both individual - simple problems as well as a single complex problem. 152 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500034418-4 FOR OFFICIAL USE ONLY At the other pole we find homogeneous computing facilities, which represent a set of elementary computers which cannot be simplified any further _ (elementary automata with programmable structure) which are linked together with regular communications. Homogeneous computer facilities provide an ideal basis for constructing computer facilities in a continuous technological process. Thanks to the simplicity of the elements and the regularity of their connections, the computing facility is manufactured in the first stage in a unified technological process. In the second stage, each user uses software ' to set the facility to implement any universal or special-purpose computer which is maximally suited to the singularities of the problem at hand. These two extreme classes of homogeneous computing systems encompass various computing systems and structures which differ in the dimensions and types of elements, and the communications configurations. Homogeneous computer systems, structures and facilities now represent the most promising direction in computer technology: Chey make it possible to eliminate the limitations which third-generation machines cannot overcome with respect to providing high performance for solving complex problems while sharply reducing cost. The changeover to industrial production of homogeneous computer systems, structures and facilities will make it possible to facilitate uniform distribution of problems within directions which have already been assimilated by industry, concentrating primary efforts in creating families of computers and minicomputers on user service quality, and on the operating efficiency of time-sharing computer systems in creating multiprocessor systems. Homogeneous computer systems, structures and facilities should be tasked to solve the problems of achieving high performance and low specific computational cost, as well as facilitating the transition to mass computing technology. 9. APPLICATION OF HOMOGENEOUS COMPUTER SYSTEMS, STRUCTURES AND FACILITIES FOR SOLVING COMPLEX PROBLEMS ~ 9.1. Application of Homogeneous Systems for Solving Problems in Economics The first chapter pointed out that homogeneous computer systems, structures arid facilities are needed to solve all three classes of complex problems: independent, weakly- and strongly-linked. This actually means that homogeneous systems provide universal means for solving both mass problems involving relatively small amounts of computation, as well as individual complex problems whose solution requires the execution of a large number of operations. Since it is necessary to use computer technology in all areas of the national economy, and considering the increasing requirements for the execution of large amounts of computation, homogeneous systems become mass computing facilities. 153 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 HOR OFFICIAI. USE ONI.Y J Under these conditions, it is difficult to enumerate all of the areas in which homogeneous systems can be app.lied effectively in the national economy. Therefore, we shall consider some of the most important ones. Once such area of homogeneous system application is solution of problems in economics, including problems involved in economic management. The difficulties in solving these problems stem from the increase in their complexity due to ' the continuing increase in the number of products produced, the increase in the number of component parts of products, the increase in the complexity of the technological processes involved in their manufacture, and the rapid obsolescense of products. The labor intensity of management problems increase to an even greater extent. The complexity of many economic management problems is determined quantitatively by material flows. An intensive increase in the ties between enterprises is observed as the economy develops. We can conclude from this that the increased complexity of economic management problems does not depend so much upon the number of objects being managed as the number of connections between them. It has been shown on the basis of experimental investigations that the complexity of economic management problems increases faster than the square of the total number of people involved in the economy. As the economy develops, a moment will arrive at which the overall complexity of the management problems will exceed the human resources of the entire society which could be used ta solve these problems. Thus, if the limit of the economic management tasks which one person could solve was reached earlier, which made it necessary at that time to change over to collective management, we are now approaching the limit of the capabilities of the entire society to manage the economy without using technical means. The need for using computers to solve problems of managing the economy is clear from this. According to estimates made in the 1~ate 1960's, effective management of the economy in this country at that time required computing facilities with a capacity of over 300 million operations per second. Unfortunately, the problem of managing the national economy cannot be split into independent problems which could be solved using several hundred isolated computers. This means that homogeneous computer systems which make it posaible to solve complex problems with the required capacity are needed to solve the problems involved in managing the national economy. Solving the problem of managing the economy requires involving hundreda of tinousands of specialists, who are spread about the country, in the computational process. This means that successful solution of the problem of managing the economy requires that we switch over to creating distributed homogeneous computer systems which combine the principles of centralized management and distributed data processing in the proper proportions. 154 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 , FOR OFFICIAL USE ONLY , Thus, the creation of the National Computer Center Network and the construction qf the Nationwide Automated Data Acquisition and Processing System for accounting and managing the national economy will become possible if homo- geneous computer systems are widely distributed throughout all branches of the national economy and joined into a unified national distributed homogeneous eomputer system. The central question in creating a unified distributed homogeneous computer system involves development of a rayon homogeneous computer system. This is because rayon systems are designed to solve problems which arise within administrative rayons, of which there are several thousand in the country. Therefore, rayon problem solving is of a mass nature. Hence, mass computer technology is required which must have both high performance and low cost. Based on the rayon population and the requirement for providing specific computing capacity per person of V~ = 104 operations per secor:d the total capacity of the rayon homogeneous computer system must react; 10~ operations per second with a specific computational cost per operation per second of C~ = 0.001 rubles. The rayon homogeneous computer system encompasses about 1,000 subscribers (enterprises, kolkhozes, sovkhozes and institutions), with connections between them comprising a total of about 1,000 kilometers. The capacity of the communications channels between elementary machines in the rayon homogeneous system must be approximately 1 mbps. Under these conditions, ~ the rayon homogeneous system will achieve a capacity of 109 operations per second regardless of the deployment of the elementary machines within the ! territory of the rayon. ; For a rayon homogeneous system it is advisable to co-locate the computer facilities with the co~aunications facilities; it is most convenient to install , ~ayon homogeneous systems at communications centers, especially at automatic telephone exchanges. Then all subscribers have direct access to the rayon ; system using existing communications channels. The presence of the rayon homogeneous computer system at an automatic telephone exchange makea it possible to place many of the functions of automatic communications (all the way up to implementation of switching) on the computer facilities. Combining computer and communications facilities in a unified data transmission, storage - and processing system makes it possible to form a unified technical and ; organizational base for solving problems involved in managing the economy with - maximum efficiency. In this respect, the rayon homogeneous computer system can be called the primary 'system at the rayon level, similar to the fashiox~ in which the primary network ~is used to build up different communications systems. By combining all of the rayon homogeneous computer systems into a unified nationwide distributed computer s~stem it is possible to obtain a primary data transmission, storage and processing system. The unified rayon homogeneous computer system can be used as the basis for the Unified System of Computer Systems and the Nationwide Automated Data Acquisition and Processing System for accounting and management - of the national economy. 155 FOR OFF[~CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 9.2. Application of Homogeneous Computer System for Solving Complex Scientific and Technical Problem~ The current state of science and technology is characterized by the occurrence of a significant number of complex problems whose solution requires high- capacity computing facilities. There is an increase in the complexity of computation in certain classes of problems, and new classes of problems are also arising in which the solution complexity is even greater. Such problems include problems of computational mathematics, mathematical physics, linear and nonlinear programming, pattern recognition, investigation and design of complex systems, modeling of complex systems, etc. There is now a requirement for solving complex problems involving 1013--1014 operations with 1010--1011 bits of information to be stored. The solution of such problems requires computing facilities with capacities of 109--1010 operations per second. Ordinary computers based on the computer model have a technical performance limit of 10~ operations per second. linder these conditions, complex problems can clearly be solved if we change over to homogeneous computer systems based on the computer collective model and having capacity of 109 operations per second or more. Concentrated homogeneous computer systems are most convenient for solving complex scientific and technical problems: these can be based on modern microcomputers, which are characterized by small size, low cost and high reliability. Let us consider some classes of complex problems and singularities of utilizing concentrated homogeneous computer systems to solve them. Many problems in radar, hydroacoustics, nuclear physics, geophysics, meteorology, medicine, sociology, etc. can be solved using pattern recognition methods. However, high capacity computer systems are required to solve these problems; in addition, the requirements for efficient man-comptiter system information exchange make it necessary to develop methods for inputting and outputting optical and audio information using pattern recognition methods. Pattern recognition problems are dividecl into three types. The firat type involves finding a decision rule in which the expenditures for construction are minimal. The second involves finding a system of features with which the costs associated with recognition errors and with measuring these features are minimized. In problems of the third type it is necessary to find the ver.sion of grouping realizations into patterns which would minimize the expenses involved in using the patterns. Problems of empirical prediction are close to pattern recognition problems. Empirical prediction problems involve developing methods of representing initial data, detecting regularities in a set of input data and representing algorithms - for predicting new regularities. The problem of empirical prediction is one ofthe main parts of the problem of artificial intelligence. 156 FOR OFFICIAL USE ONLY ~ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-04850R000500030018-4 FOR OFFICIAL USE ONLY The solution of pattern recognition and empirical prediction problems involves a large amount of computation. Furthermore, these problems can be represented as an aggregate of intercornected problems of lesser complexity, which makes it relatively easy to construct parallel algorithms which can be implemented efficiently in concentrated homogeneous computer systems. An im~portant class of problems whose solution is of major interest is comprised by problems involved in automating the design of complex articles and modeling complex systems. The problem of reducing planning periods and putting articles dnto production is now becoming especially critical in machine building. This problem can be solved by creating automated systems for planning, technological preparation and management of production on the basis of homogeneous computer systems. One of the planning problems is to develop effective methods of constructing the geometry of the planned object. It should be noted here that this problem is particularly complex in constructing the flow surfaces of flight vehicles, ship hulls and light truck cabs, since the design requirements call for optimal equipment placement and strength while the outside surface must be smooth in accordance with the requirements of hydroaerodynamics and aesthetics. When machine methods are used to solve planning problems, problems arise in developing the mathematical description of complex contours and surfaces, methods of configuring parts into assemblies, and determining the differential and integral characteristics of the surface. These problems can be solved especially effectively using the theory of nonlinear and linear cubic aplines. Spline-function methods have made it possible to create an effective mathematical apparatus for representing contours and three-dimensional surfaces which is universal,.homogeneous and easily reduced to algorithmic form. The particular value of t:hese methods is their convenience in implementing parallel computational processes in homogeneous computer systems. An important area of application of homogeneous computer systems is the investigation of complex object such as microcosmic physical systems, biological and linguistic systems, etc. In terms of properties, all of these objects are complex systems whose main singularity is that they cannot be divided into independent parts. This makes it necessary to examine the entire system as a unified whole. It becomes possible to solve such problems if concentrated homogeneous computer systems are used. � 9.3. Application of Homogeneous Computc~r Systems for Implementing Structural Moclels Constructing a liomoKeneous computer system on the basis of the computer collective model mai:es possible a different approach to the solution of a number of complex problems. The use of homogeneous computer systems is especially promising for the construction of digital models. 'Modeling devices have proven themselves in analog technology; their advantages are that it is simpler to assign the problem solving method, and the solution is 157 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00854R004500030018-4 FOR OFFICIAL USE ONLY accelerated, thanks to the structural representation of the problem in the model. The simplicity of assigning the problem solving method is combined with simplicity in changing the solution scheme during the modeling process. Structural representation of a problem makes it possible to observe the behavior of sections of the model as a function of parameter variations. Homogeneous computer systems can be used to design various automation devices, control systems, communications devices, special-purpose machines and automata. An important circumstance in structural modeling is the cap~bility of evaluating the implementation complexity of a device or machine. Using homogeneous computer systems, we can assign structural models of widely varying special- purpose machines and devices without spending a 1ot of time. This makes it possible to find the optimal structure in accordance with given criteria by trial and error. The modelin~; system can be represented as the combination of a homogeneous computer system and compucer structure or device. A modeling system based ~n a homogeneous computer system has all of the advantages of analog systems in terms of convenience and problem-solving simplicity, but avoids the shortcomings inherent in analog devices, namely low computational accuracy and the requirement of using manual methods for structure assignment. Let us consider some examples of implementing the computer collective model in communications devi~es. One of the first examples of using digital modeling . systems in communications engineering is the microprocessor-based homogeneous control system for existing electromechanical and quasi-electronic switching . systems. Such a system makps it possible to expand the scope and type of services available to subscribers and to improve the technical servicing of , existing electromechanical systems; it also facilitates a gradual transition to purely electronic systems arid the associated possibility of gradual training and retraining of personnel. To these advantages of using microprocessor-based homogeneous control systems we can also add increased reliability, which provides high viability for homogeneous systems, and relatively low cost of control devices, which facilitates mass production of small-nomenclature devices, good hardware repairability and the capability to increase the number of switching fields serviced easily by decentralizing control and adding control modules. Automatic telephone exchange controllers using the comput2r collective model are a natural complement to existing switching syatems, since a switching fi~~ld is nothing more than a certain collec:tive of connectora. In addition, this collective has the same intrinsic functions as a computer collective: connecti.ons are made by homogeneoLS devices (homogeneity), several connections can be made simultaneously (parallel operation); depending upon the number of subscribers serviced, exchange switching systems are built up by combining the necessary number of standard switching field modules. This latter property is analogous to the variable logic structure in the computer collective model. By establishing a one-to-one correspondence between the connector collective model and the computer collective model, we can thus realize automatic telephone exchanges - based on these principles. ~ 158 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY Figure 9.1 shows a functional diagram which illustrates the design principle of an erectromechanical switching system with a ring-type microprocessor- based homogeneous control system. The switching equipment establishes the connections directly. The service equipment provides the necessary signal interface between the switching equipment and the microprocessor-based control system. One characteristic feature of the control system, and incidentally of all homogeneous systems, is that the control system memory is made up of the memory of the individual microprocessors which comprise the system. This control arrangement makes it possible to divide the switching field into separate zones served by separate machines in the computer collective model. When any control system microprocessor malfunctions, its functions are taken over by another processor which is free at that time, and the system continues to function. Further development of this principle Gonsists of dividing the switching field and controller into separate functional modules and constructing automatic telephone exchanges on the basis of the computer collective model which is realized using the modules. In the present version~(Figure 9.2) the telephone exchange architecture is represented by a decentralized homogeneous system. - (1) (1) _ LLcineiro k19P/Pls//JJO!/!/U CK . ~C~~ 2 ~ / ~ \ / ~B~i ~2~ ` / ` Figure 9.1. Figure 9.2. Key: - 1;. Switching system; l. Switchinq system 2. Homogeneous computer system. 2. Homogeneous computer system The exchange consists of identical separate modules. A module is made up of a switch, which makes connections directly, and a microprocessor-based homogeneous control system, which organizes the process by which connections are made. In turn, the microprocessor-based control system consists of a . 159 FOR OFFIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 - FOR 4FFIC'IAL USE ONLY collection of microprocessors which are combined into either a ring or a matrix system. By isolating the required number of microprocessors, it is possihle to provide the required capacity and memor~? size in this case both for making connections and for making various additional services availarle to subscribers, providing the required viability, system repairability, etc. The microprocessor-based control systems of the individual modules are joined in the same manner, and represent the overall controller for the entire automatic telephone exchange. The modules are connected through both the required physical connection and under the control of the overall program which controls the individual modules in the system. - In a special case, the microprocessor-based control aystem of a single module can degenerate into a single microprocessor, but these microprocessors are nonetheless combined into the homogeneous control system of the exchange. BIBLIOGRAPHY 1. Akushskiy, I.Ya. Yuditskiy, D.I. "Mashinnaya arifinetika v ostatochnykh klassakh" [Machine arithmetic in residue classes]. Moscow, Izdatel'stvo "Sovetskoye Radio", 1968. 2. Anisimov, B.V., Chetverikov, V.N. "Osnovy teorii i proyektirovaniya tsifrovykh vychislitel'nykh mashin" [Fundamentals of theory and design - of digital computers]. Moscow, Izdatel'stvo "Vysshaya Shkola", 1970. 3. Balashov, Ye.P., Knol', A.I. "Mnogofunktsional'nye zapominayushchiye ustroystva" [Multifunctional memory devices]. Leningrad, Izdatel'stvo "Energiya", 1972. 4. Glushkov, V.M. "Sintez tsifrovykh avtomatov" [Synthesia of digital ~ automata]. Moscow, Izdatel~'stvo "Fizmatigiz", 1962. ~ 5. Varshavskiy, V.I., et. al. "Odnorodnye struktury" [Homogeneous structures]. Moscow, Izdatel'stvo "Energiya", 1973. b. Golubev-Novozhilov, Yu.S. "MnogomashinnyE kompleksy vychislitel'nykh sredstv" [Multi-machine computer complex~s]. Moscow, Izdatel'stvo "Sovetskoye Radio", 1967. 7. Yevreinov, E.V., Kosarev, Yu.G. "Vozmozhnosti postroyeniya vychislitel'nykh sistem vysokoy proizvoditel'nosti" [Possibility of designing high performance computer systems]. Novosibirsk, Izdatel'stvo SO AN SSSR, 1962. 160 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/42/09: CIA-RDP82-40850R000500430018-4 FOR OFF[CIAL USE ONLY 8. Yevreinov, E.V., Kosarev, Yu.G. "Odnorodnye vychislitel'nye sistemy vysokoy proizvoditel'nosti" [High performance homogeneous computer systems]. Novosibirsk, Izdatel'stvo "Nauka", 1966. ~ 9. Yevreinov, E.V., Prangishvili, I.V. "Tsifrovoye avtomaty s nastranvayemoy strukt~.roy (Odnorodnye sredy)" [Di.gital automata with adjustable structixre (Homogeneous media)]. Mosco~a, Izdatel'stvo "Energiya", 1971. 10. Yevreinov, E.V., Khoroshevskiy, V.G. "Odnorodnye vychislitel'nye sistemy" [Homogeneous cQmputer systems]. Novosivirsk, Izdatel'stvc "Nauka", 1978. 11. Kalyayev, A.V. "Odnorodnye kommutatsionnye regist~ovye struktury" [Homogeneous switching register structures]. Moscow, Izdatel'stvo, "Sovetskoye Radio", 1978. 12. Lazarev, V.G., Savvin, G.G. "Seti svyazi, upravleniye i kommutatsiya" [Communications networks, control and switching]. Moscow, Izdatel'stvo, "Svyaz"', 1973. 13. Pashkeyev, S.D. "Osnovy mul'tiprogrammirovaniya dlya spetsializirovannykh vychislitel'nykh sistem" [Fundamentals of multiprogramming for special- - purp~se computer systems]. Moscow, Izdatel'stvo, "Sovetskoye Radio", 1972. 14. Pospelov, D.A. "Vvedeniye v teoriyu vychislitelinykh sistem" [Introduction to theory of computer systems]. Moscow, Izdatel'stvo "Sovetskoye Radio", ~ 1972. ~ 15. Pukhov, G.Ye., Samaylov, V.D., Aristvo, V.V. "Avtomatizirovannye analogo-tsifrovye ustroystv~a modelirovaniya" [Automated analog-digital madeling dPVices]. Kiev, Izdatel'stvo "T.ekhnika", 1974. 16. Balashov, Ye.P., Smolov, V.B., Petrov, G.A., Puzankov, D.V. "Mnogo- funkt~ional'nye regulyarnye vychislitel'nye struktury" [Multifunctional regular computing struci:urps]. Moscow, Izdatel'stvo "Sovetskoye Radio", 1978. , 17. "Mul'tiprotsessornye vychi.slitel'nye sistemy" [Multiprocessor computer systemsj. Ya.P. Khegaturov, editor. Moscow, Izdatel'stvo "Energiya", I971. 18. Prangishvili, I.V., et. al. "Odnorodnye mit:roelektronnye assotsiativnye protsessory" [Homogeneous microelectronic associative processors]. Moscow, Izdatel'stvo "Sovetskoye Radio", 1973. 19. Prangishvili, I.V., et. al. "Mikroelektronika i odnorodnye struktury dlya postroyeniya logicheskikh i vychislitel'nykh ustroystv" [Microelectronics and homogeneous structures for constructing logical and computer devices]. Moscow, Izdatel'stvo "Nauka", 1967. 161 . ~ FOR OFFICI~L USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2407/02/09: CIA-RDP82-00850R000500430018-4 FOR aFFICIAL USE ON'LY 20. Glushkov, V.M., Myasnikov, V.A., Ignat'yev, M.B., Torgashev, V.A. "Rekursivnye vychislitel'nye mashiny" [Recursive computers]. Moscow, 1977 (ITMVT AN SSSR Preprint) 21. Marchuk, G.I., Kotov, V.Ye. "Modul'naya asinkhronnaya razvivayemaya sistema. Kontseptsiya" [Modular asynchronous developmental sqstem. Conception]. In two parts. Novosibirsk, 1978. (VTs SO AN SSSR Preprint). Table af Contents _ Introduction 3 1. Complex Problems 1.1. Complex systems and problems 12 1.2. Unlinked, weakly-linked and strongly-linked problems 19 1.3. Estimating computational complexity. General requirements for computer facilities 26 2. Computer Collective Model 2.1. General conception of computer collective model 32 2.2. Design principles of computer collective model 38 2.3. Structural and functional diagram of computer collective model 45 2.4. Basic properties of computer collective model 50 2.5. Basic classes of computer collective models and areas of application 55 3. Fundamentals of Canstruction of Homoseneous Computer . S~stems, Structures and Devices 3.1. General conceptio~s of homogeneous computer systems (HCS) 60 3.2. Microstructure theory of HCS 65 3.3. Macrostructural theory of HCS 80 3.4. Basic classes of computer collective models 86 4. Distributed Homogeneous Computer Systems . 4.1. Singularities of DHCS 91 4.2. Computer networks and DHCS 93 4.�3. DHCS model and iunctioning modes 96 4.4. Functional diagram of DHCS and DHCS clasaification 102 4.5. DHCS software 115 4.6. Quasi-distributed computer systems 117 5. Concentrated Homogeneous Computer Systems ~ 5.1. General conception of DHCS 120 5.2. Singularities of development of CHCS technical devices 126 5.3. CHCS theory 138 162 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4 FOR OFFICIAL USE ONLY 6. Homogeneous Computer Structures 6.1..Singularitie s of homogeneous computer structures 146 6.2. Universal homogeneous computer structures 148 _ 6.3. Problem-orien ted homogeneous computer structures 152 7. Homogeneous Computer Facilities 7.1. Singularitie s of homogeneous computer facilities 161 7.2. Design fundamentals of homogeneous computer facilities � 165 7.3. Adjustment of homogeneous computer facility 170 7.4. Physical foundations of homogeneous computer facility design 173 . 7.5. Design of digital devices in homogeneous computer facilities 177 8. Software for Homogeneous Computer Systems, Structures and Facilities 8.1. Basic approaches to software system development 179 8.2. Parallel computation methods, parallel algorithms and languages 182 8.3. HCS software 187 8.4. HCS architec ture 190 - 9. Application of Homogeneous Computer Systems, Structures and - Facilities for Solving Complex Problems 9.1. Application of Homogeneous Systems for Solving Problems in Economics 197 9.2. Application of Homogeneous Computer System for Solving - Complex Scientific and Technical Problems 200 , 9.3. Application of Homogeneous Computer Systems for ~ Implementing Structural Models , 202 Biblio.graphy 206 COPYRIGHT: Izdatel'stvo "Radio i svyaz 1981 6900 CSO: 1863/61 - END - � � 163 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000500030018-4