Fictitious input–output matrices‐based frequency limited model reduction of continuous‐time systems

The frequency‐limited model order reduction (FLMR) problem considers the model approximation within a desired frequency interval. It is observed that some of the existing FLMR methods tend to provide large approximation errors due to uneven distribution of intermediate eigenvalues. Hence, a new FLMR method based on novel frequency‐limited Gramians is proposed in this article to obtain a lower approximation error. The proposed technique gives a stable reduced‐order model due to the unique configuration of the Gramian matrices, which is obtained using intermediate fictitious input and output matrices. A comparative analysis using four numerical examples is provided, and it is found that the presented method produces better results than the prevailing methods.