Model order reduction for MIMO linear dynamical networks via Reaction-Diffusion transformation

Abstract

In this paper, we propose a model order reduction method for MIMO linear dynamical networks, where a large number of subsystems interact according to a network. In this method, a spatially one-dimensional reaction-diffusion structure, which can be efficiently exploited even for large-scale systems, is fully utilized. We give a simple algorithm as well as a computable error bound in terms of the H∞-norm.