Download the full-sized PDF of Advanced Alarm Monitoring Based on Alarm DataDownload the full-sized PDF



Permanent link (DOI):


Export to: EndNote  |  Zotero  |  Mendeley


This file is in the following communities:

Graduate Studies and Research, Faculty of


This file is in the following collections:

Theses and Dissertations

Advanced Alarm Monitoring Based on Alarm Data Open Access


Other title
Alarm monitoring
Alarm system
Process safety
Alarm floods
Causality inference
Mode-depedent alarms
Correlated alarms
Type of item
Degree grantor
University of Alberta
Author or creator
Hu, Wenkai
Supervisor and department
Chen, Tongwen (Department of Electrical and Computer Engineering)
Examining committee member and department
Chen, Tongwen (Department of Electrical and Computer Engineering)
Zhao, Qing (Department of Electrical and Computer Engineering)
Tavakoli, Mahdi (Department of Electrical and Computer Engineering)
Shah, Sirish L. (Department of Chemical and Materials Engineering)
Chen, Xiang (Department of Electrical & Computing Engineering, University of Windsor)
Department of Electrical and Computer Engineering
Control systems
Date accepted
Graduation date
2016-06:Fall 2016
Doctor of Philosophy
Degree level
Alarm systems are critical assets of modern industrial plants to assist operators in managing plant upsets and hazardous situations. A good alarm system must detect abnormalities and warn the operators promptly and yet at the same time not mislead, overload or distract the operators. However, in many actual cases, alarm systems often function poorly, and distract and overload operators due to nuisance alarms and alarm floods. Consequently, operators might be confounded by these problems, and be less alert to true alarms. To ensure process safety and improve efficiency of alarm systems, this thesis focuses on the development of advanced alarm management techniques, specifically using alarm data. Four problems have been solved. First, a new method is developed to detect and quantify correlated alarms in the presence of occurrence delays, which are identified as the main causes leading to erroneous conclusions from existing methods. The effectiveness of the method is ensured by a statistical test based on the distribution of occurrence delays and a signal conversion mechanism of generating continuous-valued pseudo alarm sequences. Second, in order to assist prediction of alarm floods and prevention of their negative consequences, an accelerated local alignment method is proposed to find similar alarm flood sequences, which are very likely caused by the same root cause. To improve the computational efficiency and accuracy, three novel strategies are incorporated, including the priority-based similarity scoring strategy, the set-based pre-matching mechanism, and the modified seeding-extending steps. Third, to identify abnormality propagation paths and detect root causes, a causality inference method based on binary-valued alarm data is developed as an alternative in the absence of process data. A modified transfer entropy (TE) and a direct transfer entropy (DTE) are formulated based on the two characteristics of alarm signals, namely, the random occurrence delays and the mutual independence of alarm occurrences. Lastly, a data-driven technique is proposed to discover association rules of mode-dependent alarms from historical Alarm & Event (A&E) logs. The proposed method can help process engineers in discovering consequential nuisance alarms, and configuring state-based alarming strategies. The effectiveness and applicability of the proposed methods are validated by case studies using real industrial alarm data. Using the proposed methods, interesting patterns can be extracted from the alarm data, and used to improve alarm monitoring by reducing nuisance alarms, addressing alarm floods, and tracking the propagation of abnormalities.
This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for the purpose of private, scholarly or scientific research. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
Citation for previous publication

File Details

Date Uploaded
Date Modified
Audit Status
Audits have not yet been run on this file.
File format: pdf (PDF/A)
Mime type: application/pdf
File size: 2593783
Last modified: 2016:11:16 14:58:52-07:00
Filename: Hu_Wenkai_201609_PhD.pdf
Original checksum: a1e647324b30122153c577826d9776ca
Activity of users you follow
User Activity Date