MODEL REDUCTION OF DISCRETE TIME CONTROL SYSTEMS

Abstract

The complications of the systems make their analysis a rather difficulty and possibly a non desirable task, mainly due to difficult economic and computational considerations involved. This makes the need for using model, which constitutes a good approximation of the original system. The purpose of this dissertation is to develop methods, which preserve time-domain as well as frequency-domain characteristics of original higher-order discrete-time systems and their application for the control of discrete-time systems. The Routh stability method is reported to be simple and powerful one, which carries significance like reduced-order models are stable, provided original system is stable. The overall time and frequency-domain characteristics are retained and it offers computational simplicity. The Routh approximation method on the other hand, posses not only the stability preservation characteristics of the system, but also many desirable features such as model stability, minimum computation and recursive nature of calculation for reduced model of various orders. However, it does not provide good transient response for purpose. A Modified Routh stability method has been proposed. Besides these reduced-order modeling techniques, Modified Routh stability method using p-domain transformation has been reported which patches up shortcomings of bilinear transformation and yields stable reduced-order models provided original system is stable.