Authors:
Funa Zhou
1
;
Tianhao Tang
2
and
Chenglin Wen
3
Affiliations:
1
Henan University/Shanghai Maritime University, China
;
2
Shanghai Maritime University, China
;
3
Hangzhou Dianzi University, China
Keyword(s):
Unknown fault pattern, Multiple faults, DCA, PCA, Fault diagnosis.
Related
Ontology
Subjects/Areas/Topics:
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Intelligent Fault Detection and Identification
Abstract:
As it can avoid the pattern compounding problem of PCA, designated component analysis (DCA) can be used to implement multiple fault diagnosis for a multivariate process. But designated fault pattern must be defined in advance, which limited its application in unknown fault diagnosis. In this paper, a hybrid DCA-PCA method is developed for unknown multiple faults diagnosis. the main idea is: Implement DCA in the first step. Removing the designated fault pattern from the observation data, then implement PCA to the residual, and use the first loading vector as the new fault pattern to extend the fault pattern base. In the third step, implement DCA for the new fault pattern and compute the significance of the new fault pattern. Simulation for data involved 4 faults shows the efficiency of the progressive DCA fault diagnosis method.