Methods of synthesis of neural regulators for adaptive TONFC control systems

Modern development of a science and technics makes more and more high demands to accuracy of control systems of technical objects with nonlinear indistinct characteristics (TONNH). Such technical objects are ionic-plasma installations, high-precision digital watching drives (CSP) metal-cutting machine tools, manipulators for robots. As a whole management methods TONNH classical methods (optimum, extreme, stochastic and. Etc.) are described by a considerable quantity of the differential equations that leads to increase in volumes of calculations and increase in time of their processing. Therefore classical management methods cannot essentially raise speed and accuracy of work TONNH. Last years interest to new scientific directions, in particular to nejrouprawleniü with use of neural networks has sharply raised. Neural networks allow to process effectively the received results, to carry out deep enough analysis static and dynamic characteristics, to trace occurrence of errors and undesirable modes. Advantage of idea of use of neural networks consists in comparative simplicity of optimization of difficult laws of management, passing process of working out of math.