Stochastic subspace identification guaranteeing stability and minimum phase
| Authors: | Tanaka Hideyuki, Kyoto University, Japan
Katayama Tohru, Kyoto University, Japan |
|---|
| Topic: | 1.1 Modelling, Identification & Signal Processing |
|---|
| Session: | Subspace Methods |
|---|
| Keywords: | Subspace identification, Stochastic realization,Canonical Correlation Analysis, Balanced realization, LQ decomposition, Spectral factorization technique |
|---|
Abstract
This paper presents a stochastic subspace identification algorithm to compute stable, minimum phase modesl for a stationary time-series data. The algorithm is based on spectral factorization techniques and a stochastic subspace identification method via a block LQ decomposition (Tanaka and Katayama 2003). Two Riccati equations are solved to ensure both stability and minimum phase property of resulting Markov models.
