INTERNAL MODEL CONTROL USING NEURAL NETWORK

Abstract

The purpose of this thesis is to provide an outline of Internal Model Control strategy and to explain how they can be used in control systems for designing a controller with and without using neural network. First of all, basic concepts of internal model control have been discussed and designed a controller for various types of systems and it is shown that this method has produced better results as perfect set point tracking and disturbance rejection. After it, an another technique, modified IMC has been used for same system transfer functions to compare the result. Then it presents the comparison between IMC and modified IMC has been conducted for different features and disturbance signals. Some basic concepts about the neural network techniques have been discussed. Training a neural network for any input-output data sets has been described in detail. In the last. controller has been designed for continuous stirred tank reactor (CSTR) using neural network. Some other systems have also been used for designing a neural network based IMC controller and discussed the results.