Finite-frequency power system reduction

Abstract

In this paper, the finite-frequency model order reduction of large-scale power systems is studied. Two computationally efficient algorithms are presented for this purpose. The algorithms use the iterative rational Krylov subspace framework of the H2-model reduction and construct a reduced order model which ensures a superior approximation accuracy within the frequency region which contains the local, interarea, and/or interplant oscillations. The algorithms also preserve the slow and poorly damped poles of the original power system model by using a hybrid approach to combine the modal preservation property of modal truncation and superior approximation accuracy of moment matching. Accuracy within the desired frequency region is achieved by using the generalized Kalman-Yakubovich-Popov (GKYP) lemma-based frequency-dependent extended realizations. The performance of the proposed algorithms is tested on practical numerical examples and a comparison with the well-known techniques is presented to highlight the efficacy of the proposed algorithms.