Title :
On the application of recursive principal component analysis method to fault detection and isolation
Author :
Jaffel, I. ; Taouali, O. ; Elaissi, I. ; Massaoud, H.
Author_Institution :
Lab. of Autom. Signal & Image Process. (LARATSI), Univ. of Monastir, Monastir, Tunisia
Abstract :
This communication suggests an extension of an online fault detection method [1] to fault isolation. This proposed method is titled Recursive Principal Component Analysis based on First Order Perturbation RPCA-FOP and it is based on first order perturbation theory (FOP) and partial PCA models where the eigenvalues and eigenvectors of the covariance matrix are updated taking into account the effect of new acquired data as a perturbation.
Keywords :
covariance matrices; eigenvalues and eigenfunctions; fault diagnosis; fault tolerant control; perturbation techniques; principal component analysis; RPCA-FOP; covariance matrix; eigenvalues; eigenvectors; fault isolation; first order perturbation theory; online fault detection method; partial PCA models; recursive principal component analysis; Chemical reactors; Covariance matrices; Data models; Eigenvalues and eigenfunctions; Fault detection; Matrices; Principal component analysis; FOP; Fault detection; PCA; RPCA-FOP; eigenvalue decomposition; partial PCA;
Conference_Titel :
Systems and Control (ICSC), 2015 4th International Conference on
Conference_Location :
Sousse
Print_ISBN :
978-1-4673-7108-7
DOI :
10.1109/ICoSC.2015.7153296
