A Stable Nonlinear in Parameter Neural Network Controller for a Class of Saturated Nonlinear Systems

Abstract

This paper deals with the tracking problem for a class of uncertain nonlinear systems subjected to actuator saturation constraint. Despite most of proposed control schemes for saturated systems which employed linear in parameter neural networks (LPNNs), in the present work nonlinear in parameter neural network (NLPNN) is introduced to support global approximation property. To compensate the effect of the input saturation constraint an auxiliary system is introduced and the error dynamics are modified based on the auxiliary states. Then, learning rules are achieved based on the back propagation (BP) algorithm and by adding two robustifying terms to the standard BP learning rules the stability of the overall system is ensured via Lyapunov direct method. Finally simulations performed on a “generalized pendulum” nonlinear system to illustrate the effectiveness of the proposed tracking control scheme.