ORDER REDUCTION OF LINEAR DYNAMIC SYSTEMS USING PADE APPROXIMATION

Abstract

In the analysis and design of a high order linear time invariant system, it is often necessary to get a low order model so that it may be used for ' an analogue or digital simulation of the system, or for designing controllers for the system. In this dissertation, we have discussed the Classical Pade Approximation Technique for model reduction of Single-input Single-output Linear Time-Invariant Dynamic Systems for continuous time case., A serious drawback of reduced order models obtained by this technique is that the reduced order model may be unstable (stable) for stable (unstable) high order systems. Seven alternative variants have been described for obtaining stable reduced order models from the high order stable systems. The performance of the reduced order models obtained by Pade approximation technique and other methods have been compared with that of the original system. A computer program for the Classical Pade approximation technique has also been developed in C language. The program for some of the techniques such as Factor Division Method, Truncation method and Differentiation method have also been developed for obtaining stable Pade approximants. The Pade approximation technique has been applied for design of a PID controller for the system model. The design procedure is based on the concept of reference model matching using Pade Approximation Technique. The performanc4the high order plant followSeffectively that of the reference model.