A New Frequency Confined Gramians-based Model Order Reduction Technique

Abstract

Model order reduction (MOR) is an approach that provides a lower-order system for a given higher-order system. Sometimes, specific frequency restrictions constitute a significant focus for practical applications, such as controller and filter designs that lead to frequency-limited MOR. Gwaronski and Juang were the first to propose a method for frequency-limited MOR. However, this method suffers from instability issues. Hence, this paper presents a new algorithm to overcome the instability issue and provide lower approximation error for the specified frequency range than existing methods. Two numerical examples are included to exhibit the benefits of the suggested approach. An eigenvalue analysis for the reduced-order models is also done to show the stability of the obtained reduced models.