Cover Image

Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process

This report summarizes the findings of the implementation of Classical Scaling (CMDS) within the framework of Multivariate Statistical Process Monitoring system. The whole set of the project can be also perceived as an extension version of the previous CMDS monitoring application which has been repo...

Full description

Saved in:
Bibliographic Details
Main Authors: Yunus, M. Y. M., Kanthasamy, Ramesh, Fatin Syazwana, Hashim
Format: Research Report
Language:English
Published: 2015
Subjects:
QA Mathematics
Online Access:http://umpir.ump.edu.my/id/eprint/36457/1/Development%20of%20multivariate%20statistical%20process%20monitoring%20system%20using%20multidimensional%20scalling%20technique%20for%20continuous%20and%20batch-based%20process.wm.pdf
http://umpir.ump.edu.my/id/eprint/36457/
Tags: Add Tag
No Tags, Be the first to tag this record!
id my.ump.umpir.36457
record_format eprints
spelling my.ump.umpir.364572023-03-07T03:39:01Z http://umpir.ump.edu.my/id/eprint/36457/ Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process Yunus, M. Y. M. Kanthasamy, Ramesh Fatin Syazwana, Hashim QA Mathematics This report summarizes the findings of the implementation of Classical Scaling (CMDS) within the framework of Multivariate Statistical Process Monitoring system. The whole set of the project can be also perceived as an extension version of the previous CMDS monitoring application which has been reported in Yunus, (2012). Thus, the primary goal of this project is mainly to enhance the CMDS monitoring performance of the initial works particularly in three distinguished directions - the introduction of general dissimilarity approach, new formulation of monitoring statistics and lastly fault diagnosis advancement. In the first, an upgraded CMDS framework which utilizes the general dissimilarity approach has been successfully developed. The main advantage of this approach is that it can adopt various forms of data inclusive of qualitative and quantitative data for monitoring. The overall result demonstrates that it has shown almost equal performance, in terms of fault detection, to the conventional Principal Component Analysis (PCA) based on a number of cases of a simulated Continuous Stirred Tank Reactor (CSTR) system. This verifies that the new CMDS algorithm is truly reliable. Next, a new statistic has also been introduced particularly in addressing the issue of high false alarm that suffered by the former statistic. A number of cases from the Tennessee Eastman Process have been chosen for demonstration. The overall results suggest that the false alarm rates have been reduced greatly which signify that the noise of the new statistics is somewhat stable compared to the original. Lastly, a new set of procedures for fault diagnosis improvement is proposed particularly by adopting the Analytical Hierarchical Process (AHP) technique. The results able to prove that the embedded AHP procedures can rather systematically help the diagnosis procedures, by means of fault prioritization, which traditionally heavily relied on the contribution plot alone. This new approach has been demonstrated based on six main abnormal cases of the simulated CSTR system. 2015-05 Research Report NonPeerReviewed pdf en http://umpir.ump.edu.my/id/eprint/36457/1/Development%20of%20multivariate%20statistical%20process%20monitoring%20system%20using%20multidimensional%20scalling%20technique%20for%20continuous%20and%20batch-based%20process.wm.pdf Yunus, M. Y. M. and Kanthasamy, Ramesh and Fatin Syazwana, Hashim (2015) Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process. , [Research Report: Research Report] (Unpublished)
institution Universiti Malaysia Pahang
building UMP Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Malaysia Pahang
content_source UMP Institutional Repository
url_provider http://umpir.ump.edu.my/
language English
topic QA Mathematics
spellingShingle QA Mathematics
Yunus, M. Y. M.
Kanthasamy, Ramesh
Fatin Syazwana, Hashim
Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
description This report summarizes the findings of the implementation of Classical Scaling (CMDS) within the framework of Multivariate Statistical Process Monitoring system. The whole set of the project can be also perceived as an extension version of the previous CMDS monitoring application which has been reported in Yunus, (2012). Thus, the primary goal of this project is mainly to enhance the CMDS monitoring performance of the initial works particularly in three distinguished directions - the introduction of general dissimilarity approach, new formulation of monitoring statistics and lastly fault diagnosis advancement. In the first, an upgraded CMDS framework which utilizes the general dissimilarity approach has been successfully developed. The main advantage of this approach is that it can adopt various forms of data inclusive of qualitative and quantitative data for monitoring. The overall result demonstrates that it has shown almost equal performance, in terms of fault detection, to the conventional Principal Component Analysis (PCA) based on a number of cases of a simulated Continuous Stirred Tank Reactor (CSTR) system. This verifies that the new CMDS algorithm is truly reliable. Next, a new statistic has also been introduced particularly in addressing the issue of high false alarm that suffered by the former statistic. A number of cases from the Tennessee Eastman Process have been chosen for demonstration. The overall results suggest that the false alarm rates have been reduced greatly which signify that the noise of the new statistics is somewhat stable compared to the original. Lastly, a new set of procedures for fault diagnosis improvement is proposed particularly by adopting the Analytical Hierarchical Process (AHP) technique. The results able to prove that the embedded AHP procedures can rather systematically help the diagnosis procedures, by means of fault prioritization, which traditionally heavily relied on the contribution plot alone. This new approach has been demonstrated based on six main abnormal cases of the simulated CSTR system.
format Research Report
author Yunus, M. Y. M.
Kanthasamy, Ramesh
Fatin Syazwana, Hashim
author_facet Yunus, M. Y. M.
Kanthasamy, Ramesh
Fatin Syazwana, Hashim
author_sort Yunus, M. Y. M.
title Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
title_short Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
title_full Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
title_fullStr Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
title_full_unstemmed Development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
title_sort development of multivariate statistical process monitoring system using multidimensional scalling technique for continuous and batch-based process
publishDate 2015
url http://umpir.ump.edu.my/id/eprint/36457/1/Development%20of%20multivariate%20statistical%20process%20monitoring%20system%20using%20multidimensional%20scalling%20technique%20for%20continuous%20and%20batch-based%20process.wm.pdf
http://umpir.ump.edu.my/id/eprint/36457/
_version_ 1759689598465736704
score 13.211869

Similar Items