An Improved Approach for Biased Model Reduction using Impulse Energy Approximation Technique

Abstract

A new method for biased model reduction of high order linear, time-invariant systems is proposed. This paper presents a new method for generating reduced order models that retain some initial time moments and Markov parameters of the high order systems (biased models). The proposed method is based on approximating the impulse energy value of the reduced order model with that of the original system. This advantageous feature guarantees a better impulse response approximation over the existing Routh approximation methods. New algorithms are proposed to generate low order models to retain the stability using only α parameters unlike other methods which need both the α and β parameters. A better contribution to the impulse response energy of the κth order reduced model is made by an appropriate selection of these parameters. The effectiveness and the flexibility of the proposed method is illustrated by two typical numerical examples.

Indexing terms:

• Large scale systems
• Model reduction techniques
• Biased reduced models