Comparisons of Subspace Identification Methods for Systems Operating on Closed-loop

Authors:	Qin S. Joe, The University of Texas at Austin, United States Lin Weilu, The University of Texas at Austin, United States Ljung Lennart, Linkoping University, Sweden
Topic:	1.1 Modelling, Identification & Signal Processing
Session:	New Results in Closed-loop Identification
Keywords:	closed-loop identification, subspace algorithms

Abstract

In this paper, we analyze two recently proposed closed-loop subspaceidentification methods, referred to as innovation estimation method and whitening filter approach respectively. The similarity and difference between them areinvestigated in detail. It turns out that all closed-loop subspace identificationmethods can be classified as one-step, two-step, or multi-stage projection methods.A SISO closed-loop simulation shows that to identify a consistent model thewhitening filter approach might require longer future and past horizons than theinnovation estimation method.