The paper considers a novel method of setting a neural networks controller that takes part in the control of a dynamic plant with unknown parameters. The uncertainties are usually overcome by using sliding mode for controller with a switching input signal. | Journal of Automation and Control Engineering Vol. 4, No. 1, February 2016 An Algorithm of Setting the Weights of a Neural Network Controller Dmitro P. Kucherov and Volodimir N. Dikhtyarenko National Aviation University, Kiev, Ukraine Email: d_kucherov@, vosha@ Andrei M. Kozub National University of Defence of Ukraine, Kiev, Ukraine Email: kozubtanja@ There are methods based on the full information about the actuator and load parameters, and adaptive approaches which enable the system to operate when its parameters vary in a wide range or allow lack of a priori information about these parameters that is the most common case. The sharp change of parameters and disturbances misbalances the control system, which operates quite well under average designed conditions, and the goal of control may not be attained at the same time. Pre-eminently in such cases the adaptive approaches should be used An approach based on constructing neural network controller can be referred to these approaches. It provides correction of the controller parameters in order to optimize its performance under current operating conditions. The advantages of this type of control can be the lack of a reference model, the ability to operate with the perturbing effects of different nature and ease of technical implementation. In this regard, the problem of synthesis of dynamic plant neural network control in the condition of parametric variation of the actuator, load and disturbing factors is seen as relevant. Known approaches of the adaptive systems implementation are based on the methods of adaptive systems constructing, such as methods of constructing extreme and neural network systems. The feasibility of using a particular method of system construction is determined by the characteristics of the control plant and its operation’s conditions. Thus, in the case of extreme dependence of the object parameters of the control signal and continuous parametric variation of the .
