A Novel Bayesian Robust Model and Its Application for Fault Detection and Automatic Supervision of Nonlinear Process

Lin Luo, Lei Xie, Uwe Kruger, Khalid Alzebdeh, Hongye Su

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In the past decade, Gaussian mixture regression (GMR) has become one of the most successful methods used for the monitoring of multivariate systems. One of the major limitations in GMR is its lack of robustness to outliers, since the estimates of the means and the precisions can be severely affected by atypical observations. Another issue is the infinite trouble, which is using maximum likelihood to fit a Gaussian mixture model (GMM). To resolve these issues of GMR-based methods, a novel Bayesian robust regression (BRR) is proposed for the automatic supervision of a nonlinear process. Specifically, a prior is placed on the mixture component in order to identify outliers in the process. Moreover, a separate precision weight is used for the inverse covariance matrix to address the infinite trouble. The predictive distribution obtained by the variational Bayesian inference is a Student's t distribution, which makes the model more robust. The theoretical findings are fully supported by an empirical study performed on an artificial data set. Moreover, the comparisons of monitoring results on a simple nonlinear system and the benchmark Tennessee Eastman process demonstrate that the BRR-based process monitoring method is superior to the classical methods, such as principal component analysis (PCA) and kernel principal component analysis (KPCA).

Original languageEnglish
Pages (from-to)5048-5061
Number of pages14
JournalIndustrial and Engineering Chemistry Research
Volume54
Issue number18
DOIs
Publication statusPublished - May 13 2015

Fingerprint

Fault detection
Principal component analysis
Monitoring
Process monitoring
Covariance matrix
Maximum likelihood
Nonlinear systems
Students

ASJC Scopus subject areas

  • Chemical Engineering(all)
  • Chemistry(all)
  • Industrial and Manufacturing Engineering

Cite this

A Novel Bayesian Robust Model and Its Application for Fault Detection and Automatic Supervision of Nonlinear Process. / Luo, Lin; Xie, Lei; Kruger, Uwe; Alzebdeh, Khalid; Su, Hongye.

In: Industrial and Engineering Chemistry Research, Vol. 54, No. 18, 13.05.2015, p. 5048-5061.

Research output: Contribution to journalArticle

@article{1fab151e59234f3981ebaa360fcadc9f,
title = "A Novel Bayesian Robust Model and Its Application for Fault Detection and Automatic Supervision of Nonlinear Process",
abstract = "In the past decade, Gaussian mixture regression (GMR) has become one of the most successful methods used for the monitoring of multivariate systems. One of the major limitations in GMR is its lack of robustness to outliers, since the estimates of the means and the precisions can be severely affected by atypical observations. Another issue is the infinite trouble, which is using maximum likelihood to fit a Gaussian mixture model (GMM). To resolve these issues of GMR-based methods, a novel Bayesian robust regression (BRR) is proposed for the automatic supervision of a nonlinear process. Specifically, a prior is placed on the mixture component in order to identify outliers in the process. Moreover, a separate precision weight is used for the inverse covariance matrix to address the infinite trouble. The predictive distribution obtained by the variational Bayesian inference is a Student's t distribution, which makes the model more robust. The theoretical findings are fully supported by an empirical study performed on an artificial data set. Moreover, the comparisons of monitoring results on a simple nonlinear system and the benchmark Tennessee Eastman process demonstrate that the BRR-based process monitoring method is superior to the classical methods, such as principal component analysis (PCA) and kernel principal component analysis (KPCA).",
author = "Lin Luo and Lei Xie and Uwe Kruger and Khalid Alzebdeh and Hongye Su",
year = "2015",
month = "5",
day = "13",
doi = "10.1021/ie503633p",
language = "English",
volume = "54",
pages = "5048--5061",
journal = "Industrial & Engineering Chemistry Product Research and Development",
issn = "0888-5885",
publisher = "American Chemical Society",
number = "18",

}

TY - JOUR

T1 - A Novel Bayesian Robust Model and Its Application for Fault Detection and Automatic Supervision of Nonlinear Process

AU - Luo, Lin

AU - Xie, Lei

AU - Kruger, Uwe

AU - Alzebdeh, Khalid

AU - Su, Hongye

PY - 2015/5/13

Y1 - 2015/5/13

N2 - In the past decade, Gaussian mixture regression (GMR) has become one of the most successful methods used for the monitoring of multivariate systems. One of the major limitations in GMR is its lack of robustness to outliers, since the estimates of the means and the precisions can be severely affected by atypical observations. Another issue is the infinite trouble, which is using maximum likelihood to fit a Gaussian mixture model (GMM). To resolve these issues of GMR-based methods, a novel Bayesian robust regression (BRR) is proposed for the automatic supervision of a nonlinear process. Specifically, a prior is placed on the mixture component in order to identify outliers in the process. Moreover, a separate precision weight is used for the inverse covariance matrix to address the infinite trouble. The predictive distribution obtained by the variational Bayesian inference is a Student's t distribution, which makes the model more robust. The theoretical findings are fully supported by an empirical study performed on an artificial data set. Moreover, the comparisons of monitoring results on a simple nonlinear system and the benchmark Tennessee Eastman process demonstrate that the BRR-based process monitoring method is superior to the classical methods, such as principal component analysis (PCA) and kernel principal component analysis (KPCA).

AB - In the past decade, Gaussian mixture regression (GMR) has become one of the most successful methods used for the monitoring of multivariate systems. One of the major limitations in GMR is its lack of robustness to outliers, since the estimates of the means and the precisions can be severely affected by atypical observations. Another issue is the infinite trouble, which is using maximum likelihood to fit a Gaussian mixture model (GMM). To resolve these issues of GMR-based methods, a novel Bayesian robust regression (BRR) is proposed for the automatic supervision of a nonlinear process. Specifically, a prior is placed on the mixture component in order to identify outliers in the process. Moreover, a separate precision weight is used for the inverse covariance matrix to address the infinite trouble. The predictive distribution obtained by the variational Bayesian inference is a Student's t distribution, which makes the model more robust. The theoretical findings are fully supported by an empirical study performed on an artificial data set. Moreover, the comparisons of monitoring results on a simple nonlinear system and the benchmark Tennessee Eastman process demonstrate that the BRR-based process monitoring method is superior to the classical methods, such as principal component analysis (PCA) and kernel principal component analysis (KPCA).

UR - http://www.scopus.com/inward/record.url?scp=84929358953&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84929358953&partnerID=8YFLogxK

U2 - 10.1021/ie503633p

DO - 10.1021/ie503633p

M3 - Article

VL - 54

SP - 5048

EP - 5061

JO - Industrial & Engineering Chemistry Product Research and Development

JF - Industrial & Engineering Chemistry Product Research and Development

SN - 0888-5885

IS - 18

ER -