An Innovative Frequency-Limited Interval Gramians-Based Model Order Reduction Method Using Singular Value Decomposition

The widely used frequency-limited Gramians-based model order reduction method for continuous-time systems, originally developed by Gawronski and Juang, yields unstable reduced-order models. To address this significant flaw, many researchers proposed a way to keep the reduced-order model stable. However, under some circumstances, these contemporary approaches also result in an unstable reduced-order model and a significant variation from the original system, leading to a significant approximation error. The present article proposes a new structure for stable continuous-time systems based on limited-interval Gramians. The suggested structure ensures that a low frequency–response approximation error is attained for the specified limited interval and ensures the stability of the reduced-order model. The stated technique provided consistent and accurate results by proving its efficacy when compared to other conventional approaches.