Reduced Gaussian process regression for fault detection of chemical processes

Radhia Fezai; Majdi Mansouri; Nasreddine Bouguila; Hazem Nounou; Mohamed Nounou

doi:10.1109/IINTEC48298.2019.9112136

Reduced Gaussian process regression for fault detection of chemical processes

Radhia Fezai, Majdi Mansouri, Nasreddine Bouguila, Hazem Nounou, Mohamed Nounou

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Citations (Scopus)

Abstract

In this paper, a reduced Gaussian process regression (RGPR)-based generalized likelihood ratio test (GLRT) is proposed for fault detection in industrial systems. In contrast to the classical GPR technique, the RGPR model can handle domains with a large number of activities that require very large training sets. The developed RGPR-based GLRT method aims first to build a RGPR model, then, it consists to apply GLRT to the monitored residuals obtained from PGPR for fault detection purposes. The fault detection performance of the developed RGPR-based GLRT method is illustrated through the Tennessee Eastman process. The simulation results show that the RGPR-based GLRT method outperforms the conventional GPR-based GLRT technique in terms of miss detection rate and CPU-time.

Original language	English
Title of host publication	2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	186-191
Number of pages	6
ISBN (Electronic)	9781728151847
DOIs	https://doi.org/10.1109/IINTEC48298.2019.9112136
Publication status	Published - Dec 2019
Externally published	Yes
Event	2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Gammarth, Tunisia Duration: 20 Dec 2019 → 22 Dec 2019

Publication series

Name	2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings

Conference

Conference	2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019
Country/Territory	Tunisia
City	Gammarth
Period	20/12/19 → 22/12/19

Keywords

Gaussian process regression (GPR)
Kmeans
chemical processes
fault detection
generalized likelihood ratio test (GLRT)
reduced Gaussian process regression (RGPR)

Access to Document

10.1109/IINTEC48298.2019.9112136

Cite this

Fezai, R., Mansouri, M., Bouguila, N., Nounou, H., & Nounou, M. (2019). Reduced Gaussian process regression for fault detection of chemical processes. In 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings (pp. 186-191). Article 9112136 (2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IINTEC48298.2019.9112136

Fezai, Radhia ; Mansouri, Majdi ; Bouguila, Nasreddine et al. / Reduced Gaussian process regression for fault detection of chemical processes. 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 186-191 (2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings).

@inproceedings{c93d070fc2604989b114c18dba4c2aa4,

title = "Reduced Gaussian process regression for fault detection of chemical processes",

abstract = "In this paper, a reduced Gaussian process regression (RGPR)-based generalized likelihood ratio test (GLRT) is proposed for fault detection in industrial systems. In contrast to the classical GPR technique, the RGPR model can handle domains with a large number of activities that require very large training sets. The developed RGPR-based GLRT method aims first to build a RGPR model, then, it consists to apply GLRT to the monitored residuals obtained from PGPR for fault detection purposes. The fault detection performance of the developed RGPR-based GLRT method is illustrated through the Tennessee Eastman process. The simulation results show that the RGPR-based GLRT method outperforms the conventional GPR-based GLRT technique in terms of miss detection rate and CPU-time.",

keywords = "Gaussian process regression (GPR), Kmeans, chemical processes, fault detection, generalized likelihood ratio test (GLRT), reduced Gaussian process regression (RGPR)",

author = "Radhia Fezai and Majdi Mansouri and Nasreddine Bouguila and Hazem Nounou and Mohamed Nounou",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 ; Conference date: 20-12-2019 Through 22-12-2019",

year = "2019",

month = dec,

doi = "10.1109/IINTEC48298.2019.9112136",

language = "English",

series = "2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "186--191",

booktitle = "2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings",

address = "United States",

}

Fezai, R, Mansouri, M, Bouguila, N, Nounou, H & Nounou, M 2019, Reduced Gaussian process regression for fault detection of chemical processes. in 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings., 9112136, 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings, Institute of Electrical and Electronics Engineers Inc., pp. 186-191, 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019, Gammarth, Tunisia, 20/12/19. https://doi.org/10.1109/IINTEC48298.2019.9112136

Reduced Gaussian process regression for fault detection of chemical processes. / Fezai, Radhia; Mansouri, Majdi; Bouguila, Nasreddine et al.
2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. p. 186-191 9112136 (2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Reduced Gaussian process regression for fault detection of chemical processes

AU - Fezai, Radhia

AU - Mansouri, Majdi

AU - Bouguila, Nasreddine

AU - Nounou, Hazem

AU - Nounou, Mohamed

PY - 2019/12

Y1 - 2019/12

N2 - In this paper, a reduced Gaussian process regression (RGPR)-based generalized likelihood ratio test (GLRT) is proposed for fault detection in industrial systems. In contrast to the classical GPR technique, the RGPR model can handle domains with a large number of activities that require very large training sets. The developed RGPR-based GLRT method aims first to build a RGPR model, then, it consists to apply GLRT to the monitored residuals obtained from PGPR for fault detection purposes. The fault detection performance of the developed RGPR-based GLRT method is illustrated through the Tennessee Eastman process. The simulation results show that the RGPR-based GLRT method outperforms the conventional GPR-based GLRT technique in terms of miss detection rate and CPU-time.

AB - In this paper, a reduced Gaussian process regression (RGPR)-based generalized likelihood ratio test (GLRT) is proposed for fault detection in industrial systems. In contrast to the classical GPR technique, the RGPR model can handle domains with a large number of activities that require very large training sets. The developed RGPR-based GLRT method aims first to build a RGPR model, then, it consists to apply GLRT to the monitored residuals obtained from PGPR for fault detection purposes. The fault detection performance of the developed RGPR-based GLRT method is illustrated through the Tennessee Eastman process. The simulation results show that the RGPR-based GLRT method outperforms the conventional GPR-based GLRT technique in terms of miss detection rate and CPU-time.

KW - Gaussian process regression (GPR)

KW - Kmeans

KW - chemical processes

KW - fault detection

KW - generalized likelihood ratio test (GLRT)

KW - reduced Gaussian process regression (RGPR)

UR - http://www.scopus.com/inward/record.url?scp=85087497089&partnerID=8YFLogxK

U2 - 10.1109/IINTEC48298.2019.9112136

DO - 10.1109/IINTEC48298.2019.9112136

M3 - Conference contribution

AN - SCOPUS:85087497089

T3 - 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings

SP - 186

EP - 191

BT - 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019

Y2 - 20 December 2019 through 22 December 2019

ER -

Fezai R, Mansouri M, Bouguila N, Nounou H , Nounou M. Reduced Gaussian process regression for fault detection of chemical processes. In 2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 186-191. 9112136. (2019 International Conference on Internet of Things, Embedded Systems and Communications, IINTEC 2019 - Proceedings). doi: 10.1109/IINTEC48298.2019.9112136

Reduced Gaussian process regression for fault detection of chemical processes

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this