Frequency-weighted and frequency interval Gramian framework-based model reduction using singular value decomposition

It is observed that the reduced-order models (ROMs) by some existing limited Gramians-based techniques deviate significantly from the high-order model, resulting in a large approximation inaccuracy. Therefore, this paper introduces a novel solution for finite-frequency model reduction using new frequency-weighted Gramians by employing balanced truncation. This work provides a novel structure of fictitious input and output matrices, resulting in the proposed continuous-time system Gramians. The suggested strategy provides stable ROMs even when input and output weightings are incorporated into the system. Furthermore, the proposed approach is extended with the frequency-interval Gramians, ensuring stable ROMs. The proposed simulation outcomes are compared with other well-known frequency-weighted and frequency-interval Gramians-based techniques using numerical examples to demonstrate the efficacy of the suggested strategies.