Currently, an important technical and theoretical task is to develop methods and
methods for managing complex dynamic objects that use both traditional methods for
controlling dynamic systems (the Pontryagin maximum principle, the Bellman control
synthesis method, the theory of automatic control), and methods based on the training of
artificial neural networks, such as methods with a reference model, predictive neural control,
method for back propagation of an error, etc. Neuropravlenie can be used in the management
of fighters, asynchronous electric drives and computers. To develop intelligent control
systems, methods of artificial intelligence can be combined with the achievements of the
classical theory of optimal control. The article shows the possibility of combining classical
methods of optimal control and optimization methods, such as the Pontryagin maximum
principle for delayed argument systems, dynamic programming methods, etc., with methods
using artificial neural networks.. The use of neural control technologies is caused by the
existence of uncontrolled noises and interference. The advantage of neural networks is the
possibility of their training, with the right choice of the activation function, accounting for
delay in signal transmission between neurons and the formation of an input signal. The aim of
the article is the development and construction of a generalized mathematical model for
controlling a complex dynamic automatic control system using methods of optimal control
theory, optimization methods and neural networks; developing a general hybrid algorithm for
obtaining optimal values of control functions and weighting coefficients of a neural network
that optimize a given functional. The created model can be used for various activation
functions, taking into account the lag and limitations on the control parameters. An algorithm
for constructing a numerical solution is developed depending on the values of the parameters
of the model, the method, and the type of activation functions. At the end of the article the
results of the computational experiment are shown.
1. Galushkin A.I. Neural networks. Fundamentals of the theory. – М.: Hot
line – Telecom, 2012. – 496 с.
2. Terekhov VA, Efimov DV, Tyukin I.Yu. Neural network control systems.
– Moscow: Higher School, 2002. – 183 p.
3. Prokhorov Danil V. Toyota Prius HEV Neurocontrol and Diagnostics. //
Neutral Networcs. – 2008. – № 21. – P. 458-465.
4. Mikrin E.A. On-board control systems for space vehicles. – Moscow:
Bauman MSTU, 2014. – 245 p.
5. Luukkonen Teppo. Modelling and control of quadcopter. URL
http://sal.aalto.fi/publications/pdf-files/eluu11_public.pdf (date of
reference 03/05/2018).
6. Pontryagin L.S. Optimal regulation processes. // The success of
mathematical Sciences. 1959. – Vol. 14. – Issue. 1. – P. 3 – 20.
7. Andreeva E.A., Kolmanovsky V.B., Shaikhet L.Ye. Management of
systems with aftereffect. – Moscow: Nauka, 1992. – 336 p.
8. Andreeva E.A. Management of dynamic systems. – Tver: Tver state
University, 2016. – 188 p.
9. Andreeva E.A., Tsiruleva V.M. Variations calculus and optimization
methods. – Moscow: Higher School, 2006. – 584 p.
10. Bellman R. Dynamic programming. – Moscow: Foreign Literature, 1960.
– 400 p.
11. Rafalskaya N.V., Tsiruleva V.M. Sufficient conditions of optimality in a
problem linear jn phase variables and in model of water purification. //
Application of functional analysis in approximation theory. – Tver: Tver.
state. University, 2001. – P. 108 – 124
12. Yevtushenko Yu.G. Methods for solving extremal problems and their
application in optimization systems. – Moscow: Nauka, 1982. – 432 p
13. Andreeva E.A, Tsiruleva V.M. Mathematical modeling of control of a
dynamic neural network with delay. // Modeling, optimization and
information technologies. – Voronezh: 2018. – Volume 6. №1. 14 c
14. Andreeva E.A., Pustarnakova Yu.A. Numerical methods for training
artificial neural networks with delay. // Journal of Computational
Mathematics and Mathematical Physics. 2002. – T. 42. P. 1383–1391.
15. Andreeva E.A., Pustarnakova Yu.A. Optimization of the neural network
with delay. – // Application of functional analysis in approximation
theory. – Tver: Tver. state. University, 2000. –. – P. 14 – 30.
16. Andreeva E.A., Tsiruleva V.M., Kozheko L.G. The model of fisheries
management. // Modeling, optimization and information technologies. –
Voronezh: 2017. – № 4 (19). – 10 p
Andreeva Elena Arkadievna
Doctor of Physical and Mathematical Sciences, Professor
Email: elena.andreeva.tvgu@yandex.ru
Tver State University
Tver, Russian Federation
Tsiruleva Valentina Mikhailovna
Candidate of Physical and Mathematical Sciences, Associate Professor
Email: vtsiruljova@mail.ru
Tver State University
Tver, Russian Federation
