SINGULAR PERTURBATION METHOD FOR MODEL V REDUCTION IN CONTROL SYSTEM

Abstract

A large number of model order reduction methods have been studied for many years to reduce the order of a large scale system. These reduced order models play very important role for analysis and design purpose of large scale system. These methods have been suggested to obtain the reduced order model in time domain as well as in frequency domain. One of the most popular methods for order reduction is the singular perturbation method (SPM). This method is not a trial and error method and it removes the stiffness of the system and satisfies the boundary conditions of the system. In this thesis first the concept of singular perturbation has been studied with some details and by some chosen examples the procedure for finding reduced order model has been explained. The concept of decentralized control system has been studied on the basis of controller design of full order decentralized system. Singular perturbation method has been used to find the reduced order model and then reduced order controller was designed using this reduced order model. Reduced order controller obtained by singular perturbation method has been implemented to full order decentralized control. A comparison was done to do the performance analysis of the reduced order controller designed by singular perturbation method with the full order controller designed by decentralization method. The results obtained on comparison showed that the robustness and reliability of the reduced-order controller could be verified and checked as to whether it was giving the required approximate results for a lower-cost controller design with loss of some accuracy.