State feedback linearisation-based neural network adaptive controller for a class of uncertain SISO non-linear systems

Abstract

In this paper, a neural network (NN) direct adaptive control algorithm is presented for a class of uncertain SISO non-linear systems. In the proposed approach there is no sign constraint on the system control gain and/or on its derivative as done in the literature. The NN approximates an ideal controller in feedback linearisation form based on an estimate of the control error signal used in the adaptive laws derivation. An estimated value of the control error is provided by a fuzzy inference system composed of a set of rules determined heuristically from information related to the history of the output tracking error. Lyapunov direct method is then used to prove the global exponential boundedness of all the signals involved in the closed loop and hence the stability of the system. Simulation results demonstrate the effectiveness of the proposed approach.