A Simplified Approach for Biased Model Reduction of Linear Systems in Special Canonical Form

Abstract

A new method is proposed for the modelling of high-order linear continuous time systems represented in special canonical form. New algorithms are developed to generate low order biased models which retain both the initial time-moments and Markov parameters of the original high-order system to give better approximation in both transient and steady state parts of the time response. The procedure is computationally simple and direct and avoids the necessity of matrix inversions and multiplications of the original high order system matrices. The new method avoids the calculation of the time moments and Markov parameters of the original high order system beforehand unlike many of the available methods for the model reduction in canonical form. The proposed method is digital computer oriented and very useful for the analysis and design of high order systems in special canonical form. The method is illustrated through typical numerical examples.