Reduced Order Modeling Using Singular Perturbation Method

Abstract

Model order reduction has been done of linear time invariant (LTI) systems by utilization of Singular Perturbation Method (SPM). In this method Eigen values of full order original system are divided into "slow" and "fast" subsystems depending on the dominating and non-dominating Eigen values. First of all the system matrices A, B, C and D are transformed into Controllable Canonical Form (CCF) as the partition matrix A22 must be non singular. The result obtained using proposed method is almost same as that of full order system and comparable with other popular methods of model order reduction. Singular perturbation method also retains the stability of original higher order system. The effectiveness of this technique is illustrated by taking one example.