Adaptive Control of Non-linear System Using Neural Network

Abstract

The paper deals with one of possible methods for control of non-linear dynamical system, namely with the adaptive neural PS controller (ANPSC). Its principle is in on-line refinement of the model system and then in the adaptive adjustment of the controller parameters. The model of the plant is realized with a two-layer feedforward neural network. The control part of the ANPSC has the structure of the classical PS controller, and it is also created in the form of the neural network (simple perceptron). The simulation calculations were performed in MATLAB.