REDUCED ORDER MODELING & SUBOPTIMAL CONTROL

Abstract

The work presented in this dissertation deals with proposed new method in model reduction and also it deals with an existing method proposed by Vittal Rao and Lamba for suboptimal control The proposed method deals with order reduction of SISO systems in frequency domain. The proposed method is compared with Conventional methods with respect to Step response, Bode plot, ISE and RELATIVE ERROR. A very few methods are available for model order reduction using cross grammain matrix. On %, Aldhaheri has proposed method in time domain the method doesn't guarantee stable reduced order models. In this dissertation a method using frequency domain cross grammain for reduced order modeling is proposed and the results for ISE and RELATIVE ERROR are compared with Balanced truncation and Antoulas method . The step response and bode plots are compared with Balanced truncation . For suboptimal control three performance indices, optimal and suboptimal controller coefficients(controller gains) and optimal and suboptimal eigen valuf have been compared The advantage of the proposed technique is that for a proper band of frequencies the characteristics of the reduced order model are that of the higher order model stability Conditions are also incorporated to insure stability. The proposed technique is explained with two examples. An the existing suboptimal control technique has been implemented by using a fourth order example