Decision Strategy for Fault Troubleshooting Using Bayesian Influence Diagram

Abstract:

Using Bayesian Network is currently an effective solution to automotive fault diagnosis. However, Bayesian Networks can only be used to reason and calculate probability of component failure. During a fault troubleshooting process, apart from fault probability, diagnostic engineers also need consider the utilities of repair actions to make a sensible repair decision. The paper extends a Bayesian Network to a Bayesian influence diagram, which integrates the influences of both probability and utility. An automobile engine start-up failure is used as a case study to establish troubleshooting decision influence diagram. The diagram combines failure causes, decision actions with their utilities and is able to reason and calculate the expected utilities of each action. Troubleshooters choose faulty component with the highest utility to repair. The method ensures the most sensible repair action is selected in each troubleshooting step.

• Periodical: Applied Mechanics and Materials (Volumes 385-386)
• Pages: 541-545
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.385-386.541
• Online since: August 2013
• Authors: Yu Sha Wang, Ying Ping Huang, Ren Jie Zhang
• Keywords: Bayesian Influence Diagram, Bayesian Network (BN), Decision-Making, Fault Troubleshooting

References

[1] Zhu Yongli, Member, IEEE, Huo Limin, and Lu Jinling, Bayesian Networks-Based Approach for Power Systems Fault Diagnosis, IEEE TRANSACTIONS ON POWER DELIVERY, vol. 21, NO. 2, APRIL 2006: 634-639.

DOI: 10.1109/tpwrd.2005.858774

[2] Z. G. Tian, X. Y. Meng, and H. F. Zhang, Application of Bayesian Theory in Fault Diagnosis of Turbo-generators, 2005 IEEE/PES Transmission and Distribution Conference & Exhibition: Asia and Pacific Dalian, China.

DOI: 10.1109/tdc.2005.1546963

[3] Lu Di, Wang Jie, The application of improved BP neural network in the engine fault diagnosis, Proceedings of the 31st Chinese Control Conference July 25-27, Hefei, China.

[4] Huang, Y., McMurran, R. and Dhadyalla, G. (2008), Probability based vehicle fault diagnosis: Bayesian network method. Journal of Intelligent Manufacture, 19: 301-311.

DOI: 10.1007/s10845-008-0083-7

[5] M. Crowley, Evaluating Influence Diagrams, www. cs. ubs. ca/~crowley/papers/aiproj. pdf, (2004).

[6] Hongxing Liu, Weihua Li, An application of influence diagram in multi-objective decision, Consumer Electronics, Communication and Netwoks(CECNet), 2012 2nd International Conference on.

DOI: 10.1109/cecnet.2012.6201684

[7] Zhiqing Cai, Shudong Sun, Shubin si, An Influence Diagram for Maintenance Strategy Decision Making, Intelligent Systems and Application, 2009. ISA 2009. International Workshop on.

DOI: 10.1109/iwisa.2009.5072721

[8] Li Gao, Feng Tian, Xiaodan Zhang, Automobile Engine Fault Diagnosis Based on Bayesian Network Model. Computer measurement and control, 2009. 17(5): 830-832.

[9] Zhixiong Lu, Diagram of automobile engine fault diagnosis, Nanjing: Jiangsu science and technology publishing house, 2001. 8.