MODEL ORDER REDUCTION VIA STEP AND FREQUENCY RESPONSE MATCHING

Abstract

Three approaches to reduce the order of linear time invariant system are given. In first approach the original system is transformed into another form of state space . representation by a suitable linear transformation, such that the truncation of the, cost coordinates from the state, space representation leads, to an approximated system so that the cost function of original system is preserved. In the second approach point to point matching method is used. In this method the frequency response of the reduced order model is close enough to the frequency response of the original model in the bandwidth of interest. In the third approach order reduction is done via balanced realization-a frequency weighting technique method, by transforming the original system into another balanced state space representation and truncating the less important stats from energy point of view.