Dynamical equivalencing of large-scale power systems using Second-Order Arnoldi algorithms

This paper proposes a technique for constructing dynamical equivalent networks of large-scale linearized power systems by using the second-order Arnoldi (SOAR) algorithm. Since the power system can be described by second-order swing equations, it is desired to construct a reduced-order model which can preserve its second-order structure. Inspired by the SOAR algorithm for single input systems, we will extend it with considering multiple inputs. Both block Arnoldi methods and global Arnoldi methods are investigated. System data of the reduced second-order system are obtained by the congruence transformation. Simulations about 50-machine power systems are performed to demonstrate the accuracy of the dynamical equivalent networks.