Use of Sparse Principal Component Analysis (SPCA) for Fault Detection

Shriram Gajjar, Murat Kulahci, Ahmet Palazoglu

Research output: Contribution to journalConference articleResearchpeer-review

Abstract

Principal component analysis (PCA) has been widely used for data dimension reduction and process fault detection. However, interpreting the principal components and the outcomes of PCA-based monitoring techniques is a challenging task since each principal component is a linear combination of the original variables which can be numerous in most modern applications. To address this challenge, we first propose the use of sparse principal component analysis (SPCA) where the loadings of some variables in principal components are restricted to zero. This paper then describes a technique to determine the number of non-zero loadings in each principal component. Furthermore, we compare the performance of PCA and SPCA in fault detection. The validity and potential of SPCA are demonstrated through simulated data and a comparative study with the benchmark Tennessee Eastman process.
Original languageEnglish
JournalIFAC-PapersOnLine
Volume49
Issue number7
Pages (from-to)693–698
ISSN2405-8963
DOIs
Publication statusPublished - 2016
Event11th IFAC Symposium on Dynamics and Control of Process Systems Including Biosystems DYCOPS-CAB 2016 - Trondheim, Norway
Duration: 6 Jun 20168 Jun 2016
Conference number: 11
http://dycops2016.org/

Conference

Conference11th IFAC Symposium on Dynamics and Control of Process Systems Including Biosystems DYCOPS-CAB 2016
Number11
CountryNorway
CityTrondheim
Period06/06/201608/06/2016
Internet address

Keywords

  • Multivariate quality control
  • Statistical Process Monitoring
  • Dimension Reduction

Cite this

Gajjar, Shriram ; Kulahci, Murat ; Palazoglu, Ahmet . / Use of Sparse Principal Component Analysis (SPCA) for Fault Detection. In: IFAC-PapersOnLine. 2016 ; Vol. 49, No. 7. pp. 693–698.
@inproceedings{9be8bdb47d5a463495c7f5d5327d0954,
title = "Use of Sparse Principal Component Analysis (SPCA) for Fault Detection",
abstract = "Principal component analysis (PCA) has been widely used for data dimension reduction and process fault detection. However, interpreting the principal components and the outcomes of PCA-based monitoring techniques is a challenging task since each principal component is a linear combination of the original variables which can be numerous in most modern applications. To address this challenge, we first propose the use of sparse principal component analysis (SPCA) where the loadings of some variables in principal components are restricted to zero. This paper then describes a technique to determine the number of non-zero loadings in each principal component. Furthermore, we compare the performance of PCA and SPCA in fault detection. The validity and potential of SPCA are demonstrated through simulated data and a comparative study with the benchmark Tennessee Eastman process.",
keywords = "Multivariate quality control, Statistical Process Monitoring, Dimension Reduction",
author = "Shriram Gajjar and Murat Kulahci and Ahmet Palazoglu",
year = "2016",
doi = "10.1016/j.ifacol.2016.07.259",
language = "English",
volume = "49",
pages = "693–698",
journal = "IFAC-PapersOnLine",
issn = "2405-8963",
publisher = "Elsevier",
number = "7",

}

Use of Sparse Principal Component Analysis (SPCA) for Fault Detection. / Gajjar, Shriram ; Kulahci, Murat; Palazoglu, Ahmet .

In: IFAC-PapersOnLine, Vol. 49, No. 7, 2016, p. 693–698.

Research output: Contribution to journalConference articleResearchpeer-review

TY - GEN

T1 - Use of Sparse Principal Component Analysis (SPCA) for Fault Detection

AU - Gajjar, Shriram

AU - Kulahci, Murat

AU - Palazoglu, Ahmet

PY - 2016

Y1 - 2016

N2 - Principal component analysis (PCA) has been widely used for data dimension reduction and process fault detection. However, interpreting the principal components and the outcomes of PCA-based monitoring techniques is a challenging task since each principal component is a linear combination of the original variables which can be numerous in most modern applications. To address this challenge, we first propose the use of sparse principal component analysis (SPCA) where the loadings of some variables in principal components are restricted to zero. This paper then describes a technique to determine the number of non-zero loadings in each principal component. Furthermore, we compare the performance of PCA and SPCA in fault detection. The validity and potential of SPCA are demonstrated through simulated data and a comparative study with the benchmark Tennessee Eastman process.

AB - Principal component analysis (PCA) has been widely used for data dimension reduction and process fault detection. However, interpreting the principal components and the outcomes of PCA-based monitoring techniques is a challenging task since each principal component is a linear combination of the original variables which can be numerous in most modern applications. To address this challenge, we first propose the use of sparse principal component analysis (SPCA) where the loadings of some variables in principal components are restricted to zero. This paper then describes a technique to determine the number of non-zero loadings in each principal component. Furthermore, we compare the performance of PCA and SPCA in fault detection. The validity and potential of SPCA are demonstrated through simulated data and a comparative study with the benchmark Tennessee Eastman process.

KW - Multivariate quality control

KW - Statistical Process Monitoring

KW - Dimension Reduction

U2 - 10.1016/j.ifacol.2016.07.259

DO - 10.1016/j.ifacol.2016.07.259

M3 - Conference article

VL - 49

SP - 693

EP - 698

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

SN - 2405-8963

IS - 7

ER -