Monitoring and Fault Diagnosis using Fisher Discrimnant Analysis

This paper presents a new monitoring and fault diagnosis method based on Fisher discriminant analysis (FDA). Conventional process monitoring and fault diagnosis based on principal component analysis (PCA) has been widely applied to chemical process. However, such PCA-based approach is ill-suited to...

Full description

Saved in:
Bibliographic Details
Published in2007 International Conference on Machine Learning and Cybernetics Vol. 2; pp. 1100 - 1105
Main Authors Xiao-Chu Tang, Yuan Li
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.08.2007
Subjects
Chemical processes
Cybernetics
Data models
Fault diagnosis
Fisher discriminant analysis
Machine learning
Matrix decomposition
Monitoring
Principal component analysis
Process control
Process monitoring
Vectors
Online AccessGet full text

Cover

More Information
Summary:This paper presents a new monitoring and fault diagnosis method based on Fisher discriminant analysis (FDA). Conventional process monitoring and fault diagnosis based on principal component analysis (PCA) has been widely applied to chemical process. However, such PCA-based approach is ill-suited to fault diagnosis. The reason is that this method only build normal data model whereas does not build fault data model. In this paper, based on pair wise Fisher discriminant analysis fault diagnosis method that consider normal and fault data model was presented to illustrate FDA superiority for fault diagnosis. In addition, based on global FDA reducing dimensional technique is also presented to indicate advantage of discriminating data. Then, two simulation examples are given: (1) one is used to demonstrate advantage of FDA for fault diagnosis.(2)the other one is used as data class example. These studies illustrate that FDA is not only an optimal reducing dimensional tool but also more efficient fault diagnosis method.
ISBN:1424409721
9781424409723
ISSN:2160-133X
DOI:10.1109/ICMLC.2007.4370308