eCite Digital Repository

Bayesian method for event-based alarm annunciation


Dalpatadu, P and Ahmed, S and Khan, FI, Bayesian method for event-based alarm annunciation, Proceedngs of the 9th IFAC Symposium on Fault Detection, Supervision andSafety for Technical Processes SAFEPROCESS 2015, 2-4 September 2015, Paris, France, pp. 832-837. ISSN 2405-8963 (2015) [Refereed Conference Paper]

Copyright Statement

Copyright 2015 IFAC

DOI: doi:10.1016/j.ifacol.2015.09.630


The concept of an event-based alarm system is introduced and a Bayesian method to annunciate event alarms is presented in this article. Unlike the current variable-based system, event alarms are allocated to abnormal events and are annunciated based on the risk of such events. The event-based alarm system design involves identification of abnormal events, identification of the sets of variables associated with each event and development of a model to predict the risk of each event using process measurements. Risk estimation involves the estimation of probabilities of individual events and their consequences. This article presents a method for estimating the probability of occurrence of abnormal events for the purpose of annunciation of an alarm. The methodology is demonstrated using the example of a simulated continuous stirred tank reactor.

Item Details

Item Type:Refereed Conference Paper
Keywords:abnormal event, alarm, Bayesian network, Bayesian networks, fault detection, plant management, risk perception, abnormal event, alarm, alarm annunciation, Bayesian methods, continuous stirred tank reactor, probability of occurrence, risk estimation
Research Division:Engineering
Research Group:Engineering practice and education
Research Field:Risk engineering
Objective Division:Expanding Knowledge
Objective Group:Expanding knowledge
Objective Field:Expanding knowledge in engineering
UTAS Author:Khan, FI (Professor Faisal Khan)
ID Code:120468
Year Published:2015
Deposited By:NC Maritime Engineering and Hydrodynamics
Deposited On:2017-08-25
Last Modified:2018-04-05

Repository Staff Only: item control page