Data-driven approximation and reduction from noisy data in matrix pencil frameworks

TL;DR

This paper introduces a data-driven method combining matrix pencil techniques and signal matrix models to learn reduced-order models from noisy time-domain data for LTI systems, demonstrated on a building model.

Contribution

It presents a novel approach integrating matrix pencil frameworks with SMM for noise-robust model reduction from time-domain data.

Findings

Effective noise handling in model reduction

Successful application to a building system

Improved accuracy over traditional methods

Abstract

This work aims at tackling the problem of learning surrogate models from noisy time-domain data by means of matrix pencil-based techniques, namely the Hankel and Loewner frameworks. A data-driven approach to obtain reduced-order state-space models from time-domain input-output measurements for linear time-invariant (LTI) systems is proposed. This is accomplished by combining the aforementioned model order reduction (MOR) techniques with the signal matrix model (SMM) approach. The proposed method is illustrated by a numerical benchmark example consisting of a building model.