Discrete-time frequency weighted model reduction and application to sampled-data models

Abstract

The problem of frequency weighted optimal model order reduction for discrete‐time systems is considered. Necessary conditions which completely characterize the reduced‐order model are given. The solution of the problem consists of a set of one generalized Riccati equation and two generalized Lyapunov equations all coupled by a projection. The results that were obtained for this problem are used to solve the problem of finding the optimal, fixed‐order, discrete‐time model for a sampled‐data continuous‐time system. Several configurations and cost functions, each corresponding to a certain physical scenario, are considered. Copyright © 1998 John Wiley & Sons, Ltd.