Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)

Processes exhibit complex behavior in chemical industries which makes the development of reliable theoretical models a very difficult and time consuming task. The resulting models are also often complex which poses additional problem for robust on-line process fault detection, diagnosis and control...

Full description

Bibliographic Details
Main Authors: Ibrahim, Kamarul 'Asri, Ahmad, Arshad, Ali, Mohamad Wijayanuddin, Mak, Weng Yee
Format: Monograph
Language:English
Published: Faculty of Chemical and Natural Resources Engineering 2006
Subjects:
Online Access:http://eprints.utm.my/2795/
http://eprints.utm.my/2795/1/74215.pdf
_version_ 1848890434065006592
author Ibrahim, Kamarul 'Asri
Ahmad, Arshad
Ali, Mohamad Wijayanuddin
Mak, Weng Yee
author_facet Ibrahim, Kamarul 'Asri
Ahmad, Arshad
Ali, Mohamad Wijayanuddin
Mak, Weng Yee
author_sort Ibrahim, Kamarul 'Asri
building UTeM Institutional Repository
collection Online Access
description Processes exhibit complex behavior in chemical industries which makes the development of reliable theoretical models a very difficult and time consuming task. The resulting models are also often complex which poses additional problem for robust on-line process fault detection, diagnosis and control of these processes. Efficient process fault detection and diagnosis in processes is important to reduce the cost of producing products with undesired specifications. Multivariate Statistical Process Control (MSPC) uses historical data of processes to develop useful process fault detection, diagnosis and control tools. Thus, the availability of theoretical models is not an important factor in the implementation of MSPC on processes. The present fault detection and diagnosis (FDD) method based on MSPC uses statistical control charts and contribution plots. These charts are efficient in fault detection but ambiguous in diagnosis of fault cause of detected faults due to the absence of control limits in the contribution plots. In this research work, an FDD algorithm is developed using MSPC and correlation coefficients between process variables. Normal Correlation (NC), Modified Principal Component Analysis (PCA) and Partial Correlation Analysis (PCorrA) are used to develop the correlation coefficients between selected key process variables and quality variables of interest. Shewhart Control Chart (SCC) and Range Control Chart (RCC) are used with the developed correlation coefficients for FDD. The developed FDD algorithm was implemented on a simulated distillation column which is a single equipment process. Results showed that the developed FDD algorithm successfully detect and diagnosed the pre-designed faults. The implementation of the developed FDD algorithm on a chemical plant can reduce the operational cost due to early detection and diagnosis of faults in the process and improving the performance of the plant.
first_indexed 2025-11-15T20:42:00Z
format Monograph
id utm-2795
institution Universiti Teknologi Malaysia
institution_category Local University
language English
last_indexed 2025-11-15T20:42:00Z
publishDate 2006
publisher Faculty of Chemical and Natural Resources Engineering
recordtype eprints
repository_type Digital Repository
spelling utm-27952010-06-01T03:05:13Z http://eprints.utm.my/2795/ Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC) Ibrahim, Kamarul 'Asri Ahmad, Arshad Ali, Mohamad Wijayanuddin Mak, Weng Yee TP Chemical technology Processes exhibit complex behavior in chemical industries which makes the development of reliable theoretical models a very difficult and time consuming task. The resulting models are also often complex which poses additional problem for robust on-line process fault detection, diagnosis and control of these processes. Efficient process fault detection and diagnosis in processes is important to reduce the cost of producing products with undesired specifications. Multivariate Statistical Process Control (MSPC) uses historical data of processes to develop useful process fault detection, diagnosis and control tools. Thus, the availability of theoretical models is not an important factor in the implementation of MSPC on processes. The present fault detection and diagnosis (FDD) method based on MSPC uses statistical control charts and contribution plots. These charts are efficient in fault detection but ambiguous in diagnosis of fault cause of detected faults due to the absence of control limits in the contribution plots. In this research work, an FDD algorithm is developed using MSPC and correlation coefficients between process variables. Normal Correlation (NC), Modified Principal Component Analysis (PCA) and Partial Correlation Analysis (PCorrA) are used to develop the correlation coefficients between selected key process variables and quality variables of interest. Shewhart Control Chart (SCC) and Range Control Chart (RCC) are used with the developed correlation coefficients for FDD. The developed FDD algorithm was implemented on a simulated distillation column which is a single equipment process. Results showed that the developed FDD algorithm successfully detect and diagnosed the pre-designed faults. The implementation of the developed FDD algorithm on a chemical plant can reduce the operational cost due to early detection and diagnosis of faults in the process and improving the performance of the plant. Faculty of Chemical and Natural Resources Engineering 2006-12-11 Monograph NonPeerReviewed application/pdf en http://eprints.utm.my/2795/1/74215.pdf Ibrahim, Kamarul 'Asri and Ahmad, Arshad and Ali, Mohamad Wijayanuddin and Mak, Weng Yee (2006) Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC). Project Report. Faculty of Chemical and Natural Resources Engineering, Skudai, Johor. (Unpublished)
spellingShingle TP Chemical technology
Ibrahim, Kamarul 'Asri
Ahmad, Arshad
Ali, Mohamad Wijayanuddin
Mak, Weng Yee
Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title_full Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title_fullStr Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title_full_unstemmed Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title_short Development of fault detection, diagnosis and control system identification using multivariate statistical process control (MSPC)
title_sort development of fault detection, diagnosis and control system identification using multivariate statistical process control (mspc)
topic TP Chemical technology
url http://eprints.utm.my/2795/
http://eprints.utm.my/2795/1/74215.pdf