Frequency-based model order reduction via genetic algorithm approach

Abstract

A frequency-based model order reduction (MOR) via genetic algorithm (GA) approach is presented in this paper. An exogenous autoregressive model with a smaller dimensionality, which can mimic the full order model, maybe obtained using the GA MOR approach. For a general MOR, the GA predicts the elements of the system state matrix [A] defined in a state space representation along with the elements of the [B] and [C] matrices of the reduced order model. As a frequency-based MOR technique, the GA predicts only the elements of the [B] and [C] matrices of the reduced order model while [A] is set in the modal form. The proposed GA model order reduction approach is compared to recently published work for method evaluation.