Title :
A new method for determining PCA models for system diagnosis
Author :
Benaicha, Anissa ; Mourot, Gilles ; Guerfel, Mohamed ; Benothman, Kamel ; Ragot, José
Author_Institution :
Res. Unit ATSI, Nat. Eng. Sch. of Monastir, Monastir, Tunisia
Abstract :
In this paper, a new method is proposed to determine the structure of PCA models for system diagnosis. This method based on the principle of variable reconstruction determines PCA models in order to optimize detection and isolation of simple and multiple faults affecting redundant or non redundant variables. This new method has been validated by a simulation example of a nonlinear system.
Keywords :
fault diagnosis; optimisation; principal component analysis; PCA models structure; fault detection optimisation; nonlinear system; nonredundant variables; system diagnosis; variable reconstruction principle; Covariance matrix; Data models; Eigenvalues and eigenfunctions; Fault detection; Indexes; Principal component analysis; Signal to noise ratio; PCA; fault detection and isolation; number of principal components; sensor fault; variable reconstruction;
Conference_Titel :
Control & Automation (MED), 2010 18th Mediterranean Conference on
Conference_Location :
Marrakech
Print_ISBN :
978-1-4244-8091-3
DOI :
10.1109/MED.2010.5547762
