Maintenance management system based on Bayesian networks

Zhiqiang Cai, Shudong Sun, Shubin Si, Bernard Yannou

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

6 Scopus citations

Abstract

With the application of maintenance management system, the daily failure data of airplane components has been collected in great number. The challenge faced currently is how to predict the failure probability of certain components and carry out optimal maintenance strategy (when and how) to reduce the lifecycle cost of the airplane. In this paper, we have presented a proactive maintenance management framework as well as the entire maintenance work flow which guides the establishment of maintenance system in detail. Furthermore, we brought out the procedures of building prognosis Bayesian networks and decision Bayesian networks. The prognosis network which is based on discovered associations from failure data can provide reliable predictions on component failure probability; while the decision network could help to make maintenance decisions balancing among resource availability, repair effect, cost and product risk.

Original languageEnglish
Title of host publication2008 International Seminar on Business and Information Management, ISBIM 2008
PublisherIEEE Computer Society
Pages42-45
Number of pages4
ISBN (Print)9780769535609
DOIs
StatePublished - 2008
Event2008 International Seminar on Business and Information Management, ISBIM 2008 - Wuhan, China
Duration: 19 Dec 200819 Dec 2008

Publication series

Name2008 International Seminar on Business and Information Management, ISBIM 2008
Volume2

Conference

Conference2008 International Seminar on Business and Information Management, ISBIM 2008
Country/TerritoryChina
CityWuhan
Period19/12/0819/12/08

Keywords

  • Bayesian networks
  • Decision making
  • Failure prognosis
  • Proactive maintenance

Fingerprint

Dive into the research topics of 'Maintenance management system based on Bayesian networks'. Together they form a unique fingerprint.

Cite this