Structure Preserving and Passive Reduced-Order Macromodeling of Second-Order Systems

DOI: 10.5176/2251-1911_CMCGS50

Authors: Yung-Ta Li

Abstract: We develop a new numerical method for model reduction of passive second-order dynamical systems. It is based on a newly developed decomposition, namely, structured quasi-Arnoldi decomposition. Once the structured quasi-Arnoldi decomposition is computed, a previous developed back conversion procedure will be utilized to transform a linear system to a second-order form. In other words, computing the structured quasi-Arnoldi decomposition is a preprocess of back conversion. Without this preprocess, we may need to search all choices to construct a transformation matrix and the method is combinatorially hard.