Structure-preserving model reduction of physical network systems by clustering

TL;DR

This paper presents a clustering-based model reduction technique for physical network systems that preserves structural properties and provides explicit error bounds, applicable to both first- and second-order systems.

Contribution

It introduces a novel clustering method based on almost equitable partitions that maintains physical properties and offers explicit H2-norm error estimates.

Findings

Explicit error bounds for model reduction are derived.

The method extends to second-order physical network systems.

Clustering preserves key physical and structural properties.

Abstract

In this paper, we establish a method for model order reduction of a certain class of physical network systems. The proposed method is based on clustering of the vertices of the underlying graph, and yields a reduced order model within the same class. To capture the physical properties of the network, we allow for weights associated to both the edges as well as the vertices of the graph. We extend the notion of almost equitable partitions to this class of graphs. Consequently, an explicit model reduction error expression in the sense of H2-norm is provided for clustering arising from almost equitable partitions. Finally the method is extended to second-order systems.