SYSTEM IDENTIFICATION USING MULTILAYER NEURAL NETWORK

Abstract

A number of efforts have been made to address the issues relating identification of systems. In spite of such attempts, constructive procedures similar to those available for linear systems are not available for nonlinear systems. Hence, identification and control is a formidable problem. The properties of neural networks like flexibility, learning ability and ability to approximate any nonlinear function motivate us for their use in the field. In this dissertation, identification along with control of dynamic systems using multi-layer feed forward neural network is carried out in MATLAB and C++. Simulation results for identification results indicate that the trained neural network behavior approximates the behaviour of actual plant. Also, the simulation results for control indicate that the output of the plant follows the output of the reference model closely