Fault monitoring of nonlinear process based on kernel concurrent projection to latent structures

Author

Rongrong Sun ; Yunpeng Fan ; Yingwei Zhang

Author_Institution

Key Lab. of Integrated Autom. of Process Ind., Northeastern Univ., Shenyang, China

fYear

2014

fDate

28-30 July 2014

Firstpage

5184

Lastpage

5189

Abstract

In this paper, a fault monitoring method based on kernel concurrent projection to latent structures (KCPLS) is proposed. It is the purpose of the article to effectively detect the faults which are happening in all of the spaces. Considering the problems of the conventional PLS model, the residual spaces of X and Y is further decomposed to establish a more accurate relation between input and output. The non-Gaussian is considered as most of the industrial process is nonlinear. KCPLS detection is applied to vehicle battery industrial processes. The results of simulation show the effectiveness of the proposed method.

Keywords

fault diagnosis; fault tolerant control; nonlinear control systems; principal component analysis; KCPLS; fault detection; fault monitoring; kernel concurrent projection to latent structures; nonlinear process; vehicle battery industrial process; Batteries; Eigenvalues and eigenfunctions; Kernel; Monitoring; Principal component analysis; Vectors; Vehicles; Fault Detection; KCPLS Detection; Kernel Concurrent Projection to Latent Structures (KCPLS); Residual Spaces;

fLanguage

English

Publisher

ieee

Conference_Titel

Control Conference (CCC), 2014 33rd Chinese

Conference_Location

Nanjing

Type

conf

DOI

10.1109/ChiCC.2014.6895823

Filename

6895823