Order Reduction of Linear System Using Clustering, Integral Square Minimization and Dominant Pole Technique

Abstract—A mixed method is proposed for finding stable reduced order models of single-input- single-output large-scale systems using integral square error (ISE) minimization techniques, the clustering technique and retaining one or more dominant pole. The denominator polynomial of the reduced order model is determined by forming the clusters of the poles of the original system and retaining one or more dominant pole, and the coefficients of numerator polynomial are obtained by using the integral square error minimization technique. This method guarantees stability of the reduced order model when the original high order system is stable. The methodology of the proposed method is illustrated with the help of examples from literature.