Structure preserving model-order reductions of MIMO second-order systems using Arnoldi methods

This paper investigates structure preserving model-order reductions of the MIMO second-order system. By extending the previous SISO second-order Arnoldi (SOAR) algorithm, both block Arnoldi methods and global Arnoldi methods will be investigated. Analytic expressions of system moments and output moments will be derived analytically in terms of the upper Hessenberg matrix. By employing the so-called congruence transformation, the system data of the reduced second-order system will be obtained. Relationships among these coefficients will also be derived. Simulations about practical engineering applications will be performed to illustrate the feasibility and the efficiency of these two classes of model reductions.