Dynamic modelling and control for a class of non-linear systems using neural nets

Abstract

The paper describes a new neural network Controller using an IMC structure (NIMC). The structure is suitable for control of discrete-time SISO systems containing nonlinearities. Two design steps are assumed: (1) the controller is designed for optimal set-point tracking and disturbance rejection or model uncertainty and (2) the controller is detuned for robust performance. Comparative studies between NIMC and a conventional nonlinear adaptive controller is made.<>