Noise-induced bias in PCA modeling of linear system

Author

Cao, Jin ; Gertler, Janos

Author_Institution

Dept. of Electr. & Comput. Eng., George Mason Univ., Fairfax, VA, USA

Volume

5

fYear

2001

fDate

2001

Firstpage

3660

Abstract

The estimation of parameter bias caused by noise is very significant for modeling and identification. The effect of noise in least squares (LS) identification is well known. Recent studies on modeling by principal component analysis (PCA) have raised the problem of noise-induced bias in this framework. In this paper, a systematical investigation of this topic in static linear systems is presented. Analytical expressions of the bias induced by actuator and sensor noise in PCA modeling are derived, and an approach to approximate these expressions is developed for practical use. The theoretical results are verified by simulation results

Keywords

eigenvalues and eigenfunctions; fault diagnosis; least squares approximations; linear systems; parameter estimation; principal component analysis; MIMO system; eigenvalues; estimation bias; fault detection; fault diagnosis; identification; least squares; linear systems; modeling; parameter estimation; principal component analysis; Actuators; Fault detection; Fault diagnosis; Least squares approximation; Least squares methods; Linear systems; Parameter estimation; Principal component analysis; Redundancy; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

American Control Conference, 2001. Proceedings of the 2001

Conference_Location

Arlington, VA

ISSN

0743-1619

Print_ISBN

0-7803-6495-3

Type

conf

DOI

10.1109/ACC.2001.946203

Filename

946203