Improved nonlinear PCA for process monitoring using support vector data description

► We study the monitoring of nonlinear processes based on neural networks. ► We incorporate principal curves and RBF networks and simplify nonlinear PCA. ► The Fast Recursive Algorithm determines the RBF network topology efficiently. ► Support vector data description estimates statistic of nonlinear...

Full description

Saved in:
Bibliographic Details
Published in:Journal of process control Vol. 21; no. 9; pp. 1306 - 1317
Main Authors: Liu, Xueqin, Li, Kang, McAfee, Marion, Irwin, George W.
Format: Journal Article
Language:English
Published: Elsevier Ltd 01.10.2011
Subjects:
Fast Recursive Algorithm
Nonlinear principal component analysis
Principal curves
Radial basis function network
Support vector datadescription
Fast Recursive Algorithm
Radial basis function network
Nonlinear principal component analysis
Support vector datadescription
Principal curves
ISSN:0959-1524, 1873-2771
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:► We study the monitoring of nonlinear processes based on neural networks. ► We incorporate principal curves and RBF networks and simplify nonlinear PCA. ► The Fast Recursive Algorithm determines the RBF network topology efficiently. ► Support vector data description estimates statistic of nonlinear PCs effectively. Nonlinear principal component analysis (PCA) based on neural networks has drawn significant attention as a monitoring tool for complex nonlinear processes, but there remains a difficulty with determining the optimal network topology. This paper exploits the advantages of the Fast Recursive Algorithm, where the number of nodes, the location of centres, and the weights between the hidden layer and the output layer can be identified simultaneously for the radial basis function (RBF) networks. The topology problem for the nonlinear PCA based on neural networks can thus be solved. Another problem with nonlinear PCA is that the derived nonlinear scores may not be statistically independent or follow a simple parametric distribution. This hinders its applications in process monitoring since the simplicity of applying predetermined probability distribution functions is lost. This paper proposes the use of a support vector data description and shows that transforming the nonlinear principal components into a feature space allows a simple statistical inference. Results from both simulated and industrial data confirm the efficacy of the proposed method for solving nonlinear principal component problems, compared with linear PCA and kernel PCA.
ISSN:0959-1524
1873-2771
DOI:10.1016/j.jprocont.2011.07.003