Improved detection of incipient anomalies via multivariate memory monitoring charts

Application to an air flow heating system

Fouzi Harrou, Muddu Madakyaru, Ying Sun, Sofiane Khadraoui

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

Detecting anomalies is important for reliable operation of several engineering systems. Multivariate statistical monitoring charts are an efficient tool for checking the quality of a process by identifying abnormalities. Principal component analysis (PCA) was shown effective in monitoring processes with highly correlated data. Traditional PCA-based methods, nevertheless, often are relatively inefficient at detecting incipient anomalies. Here, we propose a statistical approach that exploits the advantages of PCA and those of multivariate memory monitoring schemes, like the multivariate cumulative sum (MCUSUM) and multivariate exponentially weighted moving average (MEWMA) monitoring schemes to better detect incipient anomalies. Memory monitoring charts are sensitive to incipient anomalies in process mean, which significantly improve the performance of PCA method and enlarge its profitability, and to utilize these improvements in various applications. The performance of PCA-based MEWMA and MCUSUM control techniques are demonstrated and compared with traditional PCA-based monitoring methods. Using practical data gathered from a heating air-flow system, we demonstrate the greater sensitivity and efficiency of the developed method over the traditional PCA-based methods. Results indicate that the proposed techniques have potential for detecting incipient anomalies in multivariate data.

Original languageEnglish
Pages (from-to)65-74
Number of pages10
JournalApplied Thermal Engineering
Volume109
DOIs
Publication statusPublished - 25-10-2016

Fingerprint

Principal component analysis
Heating
Data storage equipment
Monitoring
Air
Process monitoring
Systems engineering
Profitability

All Science Journal Classification (ASJC) codes

  • Energy Engineering and Power Technology
  • Industrial and Manufacturing Engineering

Cite this

@article{f39297ca293b4801bef277bc13348813,
title = "Improved detection of incipient anomalies via multivariate memory monitoring charts: Application to an air flow heating system",
abstract = "Detecting anomalies is important for reliable operation of several engineering systems. Multivariate statistical monitoring charts are an efficient tool for checking the quality of a process by identifying abnormalities. Principal component analysis (PCA) was shown effective in monitoring processes with highly correlated data. Traditional PCA-based methods, nevertheless, often are relatively inefficient at detecting incipient anomalies. Here, we propose a statistical approach that exploits the advantages of PCA and those of multivariate memory monitoring schemes, like the multivariate cumulative sum (MCUSUM) and multivariate exponentially weighted moving average (MEWMA) monitoring schemes to better detect incipient anomalies. Memory monitoring charts are sensitive to incipient anomalies in process mean, which significantly improve the performance of PCA method and enlarge its profitability, and to utilize these improvements in various applications. The performance of PCA-based MEWMA and MCUSUM control techniques are demonstrated and compared with traditional PCA-based monitoring methods. Using practical data gathered from a heating air-flow system, we demonstrate the greater sensitivity and efficiency of the developed method over the traditional PCA-based methods. Results indicate that the proposed techniques have potential for detecting incipient anomalies in multivariate data.",
author = "Fouzi Harrou and Muddu Madakyaru and Ying Sun and Sofiane Khadraoui",
year = "2016",
month = "10",
day = "25",
doi = "10.1016/j.applthermaleng.2016.08.047",
language = "English",
volume = "109",
pages = "65--74",
journal = "Applied Thermal Engineering",
issn = "1359-4311",
publisher = "Elsevier Limited",

}

Improved detection of incipient anomalies via multivariate memory monitoring charts : Application to an air flow heating system. / Harrou, Fouzi; Madakyaru, Muddu; Sun, Ying; Khadraoui, Sofiane.

In: Applied Thermal Engineering, Vol. 109, 25.10.2016, p. 65-74.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Improved detection of incipient anomalies via multivariate memory monitoring charts

T2 - Application to an air flow heating system

AU - Harrou, Fouzi

AU - Madakyaru, Muddu

AU - Sun, Ying

AU - Khadraoui, Sofiane

PY - 2016/10/25

Y1 - 2016/10/25

N2 - Detecting anomalies is important for reliable operation of several engineering systems. Multivariate statistical monitoring charts are an efficient tool for checking the quality of a process by identifying abnormalities. Principal component analysis (PCA) was shown effective in monitoring processes with highly correlated data. Traditional PCA-based methods, nevertheless, often are relatively inefficient at detecting incipient anomalies. Here, we propose a statistical approach that exploits the advantages of PCA and those of multivariate memory monitoring schemes, like the multivariate cumulative sum (MCUSUM) and multivariate exponentially weighted moving average (MEWMA) monitoring schemes to better detect incipient anomalies. Memory monitoring charts are sensitive to incipient anomalies in process mean, which significantly improve the performance of PCA method and enlarge its profitability, and to utilize these improvements in various applications. The performance of PCA-based MEWMA and MCUSUM control techniques are demonstrated and compared with traditional PCA-based monitoring methods. Using practical data gathered from a heating air-flow system, we demonstrate the greater sensitivity and efficiency of the developed method over the traditional PCA-based methods. Results indicate that the proposed techniques have potential for detecting incipient anomalies in multivariate data.

AB - Detecting anomalies is important for reliable operation of several engineering systems. Multivariate statistical monitoring charts are an efficient tool for checking the quality of a process by identifying abnormalities. Principal component analysis (PCA) was shown effective in monitoring processes with highly correlated data. Traditional PCA-based methods, nevertheless, often are relatively inefficient at detecting incipient anomalies. Here, we propose a statistical approach that exploits the advantages of PCA and those of multivariate memory monitoring schemes, like the multivariate cumulative sum (MCUSUM) and multivariate exponentially weighted moving average (MEWMA) monitoring schemes to better detect incipient anomalies. Memory monitoring charts are sensitive to incipient anomalies in process mean, which significantly improve the performance of PCA method and enlarge its profitability, and to utilize these improvements in various applications. The performance of PCA-based MEWMA and MCUSUM control techniques are demonstrated and compared with traditional PCA-based monitoring methods. Using practical data gathered from a heating air-flow system, we demonstrate the greater sensitivity and efficiency of the developed method over the traditional PCA-based methods. Results indicate that the proposed techniques have potential for detecting incipient anomalies in multivariate data.

UR - http://www.scopus.com/inward/record.url?scp=84981734323&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84981734323&partnerID=8YFLogxK

U2 - 10.1016/j.applthermaleng.2016.08.047

DO - 10.1016/j.applthermaleng.2016.08.047

M3 - Article

VL - 109

SP - 65

EP - 74

JO - Applied Thermal Engineering

JF - Applied Thermal Engineering

SN - 1359-4311

ER -