Radiotekhnika
Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS


Тел.: +7 (495) 625-9241

 

Probabilistic Neural network of decision-making of structural control loop of a complex dynamic object

Keywords:

A.G. Volkov – Ph.D. (Eng.), Associate Professor, Financial University under Government of Russian Federation (Moscow)
E-mail: avolkov@fa.ru
V.I. Goncharenko – Dr.Sc. (Eng.), Associate Professor, Director of Military Institute, Moscow Aviation Insti-tute (National Research University); Lead Engineer, V.A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences (Moscow)
E-mail: vladimirgonch@mail.ru
N.G. Zhuravlyova – Junior Research Scientist, V.A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences (Moscow)
M.A. Ikonnikov – Post-graduate Student, Moscow Aviation Institute (National Research University)
E-mail: maxim.ikonnikov@gmail.com


The results of these studies showed that urgent tasks are related to the development of better, and above all, intelligent me-thods of diagnosing complex objects control meeting the requirements of high efficiency and reliability. Traditional estimation algorithms, despite the high versatility, has low operational decision making, especially with increasing completeness of classes of technical states. It is suggested the use of neural network technology for the treatment of control and diagnostic information of complex objects control and synthesis of neural network algorithm for computing the estimates, based on neural network and traditional algorithms. The application of the algebraic approach in solving the problem of synthesis of functional diagnosis algorithm that allows you to simulate multi-layered network of direct distribution, using as a basis neural networks dual-layer network. This makes it virtually unlimited accuracy of separating surface between classes of technical states. The existing organization of control and diagnosis of complex airborne systems of dynamic objects does not have the required urgency of solving the problem of autonomous operational disaster recovery onboard equipment. To address such need to develop a stand-alone closed-loop control structure, based on which the control system is put intelligent control system diagnostics and reconfiguration using neural network logical basis. The results of analysis of existing neural networks and types of neurons, which are allowed to propose a new model of decision-making subsystem based on an assessment of the technical condition of the structure which is based on a four-probabilistic neural network PNN, where as the second layer of a so-called competitive layer. The necessity of using a probabilistic network, which is caused by fuzzy (probabilistic) character input situations. At the same time competing layer calculates the probability of the input vector belonging to a particular class, and ultimately puts fuzzy input vector according to the class, the probability of belonging to that above. This experiment confirmed the feasibility of using probabilistic PNN network type as a model for the decision on the reconfiguration of onboard equipment, under certain statistical data on the distribution of a class of decisions and the corresponding samples of neurons. The developed neural network algorithm for computing the estimates combines the advantage of the traditional estimation algorithms as the versatility and advantage of neural networks – high speed at the required reliability of decisions. The proposed algorithm is implemented in software of functional diagnosis of the onboard equipment of modern aircrafts

References:
  1. Kruglov V.V., Borisov V.V. Iskusstvenny'e nejronny'e seti. Teoriya i praktika. M.: Goryachaya liniya – Telekom. 2001. 382 s.
  2. Polous A.I., Volkov A.G., Goncharov V.V. Osnovy' sistem iskusstvennogo intellekta: Ucheb. posobie. M.: VA RVSN im. Petra Velikogo. 2009. 251 s.
  3. Polous A.I., Volkov A.G. Nejrosetevaya model' kontrolya i diagnostirovaniya sistemy' upravleniya dvizheniem KA // Trudy' NTK MTORE'S im. A. S. Popova. M.: VA RVSN im. Petra Velikogo. 2003. S. 85 – 92.
  4. Arkanov A.V., Goncharenko V.I., Kobzar' A.A. Opredelenie prichin neshtatnogo funkczionirovaniya ballisticheskogo ob''ekta // Izv. Instituta inzhenernoj fiziki. 2012. № 2 (24). S. 8–10.
  5. Goncharenko V.I., Kobzar' A.A., Korneev A.A., Kucheryavenko D.S. Algoritm oczenivaniya xarakteristik ballisticheskogo ob''ekta pri neizvestny'x parametrax ego matematicheskoj modeli dvizheniya i nekachestvennoj izmeritel'noj informaczii // Vestnik Kazanskogo gosudarstvennogo texnicheskogo universiteta im. A.N. Tupoleva. 2011. № 4. S. 128–135.
  6. Goncharenko V.I., Kucheryavenko D.S., Gojdenko V.K., Skorik N.A. Raspoznavanie tipa avarijnoj situaczii pri ispy'tanii bespilotnogo kosmicheskogo apparata na osnove ispol'zovaniya vejvlet-preobrazovaniya // Nejrokomp'yutery': razrabotka, primenenie. 2016. № 1. S. 39–48.
  7. Nechaev V.V., Goncharenko V.I., Rozhnov A.V., Ly'chev A.V., Lobanov I.A. Integracziya komponentov virtual'noj semanticheskoj sredy' i obobshhennoj modeli analiza sredy' funkczionirovaniya // Sovremenny'e informaczionny'e texnologii i IT-obrazovanie. 2016. T. 12. № 3. Ch. 1. S. 187–194.
  8. Ogandzhanyan S.B., Rozhnov A.V., Burmistrov P.A., Lobanov I.A., Tyurin S.A. Tvorcheskie materialy' «kruglogo stola». Ch. I. Retrospektiva i real'naya konkordancziya issledovanij v sfere intellekta // Nejrokomp'yutery': razrabotka, primenenie. 2016. № 1. S. 17–29.
  9. Rozhnov A.V. Tvorcheskie materialy' «kruglogo stola». Ch. II. Sistemnaya integracziya i modelirovanie novy'x e'ffektov v sfere intellekta // Nejrokomp'yutery': razrabotka, primenenie. 2016. № 3. S. 3–12.
  10. Goncharenko V.I., Legovich Ju.S., Ry'vkin S.E. Obrabotka telemetricheskoj informaczii o kinematicheskix parametrax dvizheniya letatel'ny'x apparatov na osnove diskretnogo vejvlet-preobrazovaniya // Doklady' 18-j mezhdunarodnoj konferenczii. Ser. Cifrovaya obrabotka signalov i ee primenenie– DSPA-2016. (Moskva, 30 marta – 01 aprelya 2016 g.). 2016. T. 1. Vy'p. XVIII-1. M.: RNTORE'S im. A.S. Popova. S. 442–447.
  11. Rozhnov A.V., Lobanov I.A., Bimakov E.V. Obosnovanie zadach sistemnoj integraczii i informaczionno-analiticheskoe modelirovanie problemno-orientirovanny'x sistemy' upravleniya na predproektnom e'tape zhiznennogo czikla / Trudy' XII Vserossijskogo soveshhaniya po problemam upravleniya (Moskva). M.: IPU RAN. 2014. S. 7474-7479.
  12. Rozhnov A.V., Gubin A.N., Mixajlov R.A., Belavkin P.A. Stanovlenie vy'chislitel'ny'x sistem i kompleksov voennogo naznacheniya na zare strategicheskix raketny'x vojsk // Nejrokomp'yutery': razrabotka, primenenie. 2010. № 11. S. 62–68.
  13. Legovich Ju.S., Miodushevskij P.V., Rozhnov A.V. Sistemnaya integracziya i uskorennoe maketirovanie smeshannoj robototexnicheskoj gruppirovki na predproektnoj stadii zhiznennogo czikla / Trudy' 16-j Mezhdunar. molodezhnoj konf. «Sistemy' proektirovaniya, texnologicheskoj podgotovki proizvodstva i upravleniya e'tapami zhiznennogo czikla promy'shlennogo produkta» (CAD/CAM/PDM-2016, Moskva). M.: OOO «Analitik». 2016. S. 199–202.

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio