Fault Isolation Manuals (FIMs) are derived from a type of decision tree and play an important role in maintenance troubleshooting of large systems. However, there are some drawbacks to using decision trees for maintenance, such as requiring a static order of tests to reach a conclusion. One method to overcome these limitations is by converting FIMs to Bayesian networks. However, it has been shown that Bayesian networks derived from FIMs will not contain the entire set of fault and alarm relationships present in the system from which the FIM was developed. In this paper we analyze Bayesian networks that have been derived from FIMs and report on several measurements, such as accuracy, relative probability of target diagnoses, diagnosis rank, and KL-divergence. Based on our results, we found that even with incomplete information, the Bayesian networks derived from the FIMs were still able to perform reasonably well.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    An empirical evaluation of Bayesian networks derived from fault trees


    Contributors:


    Publication date :

    2013-03-01


    Size :

    5309643 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English




    Empirical Modeling Based on Neural Networks and Bayesian Learning

    Makeev, Andrew / Nikishkov, Yuri / Armanios, Erian | AIAA | 2004


    Object-Oriented Algorithm For Evaluation Of Fault Trees

    Patterson-Hine, F. A. / Koen, B. V. | NTRS | 1992



    Fault-tolerant ADCS for Small Satellites Using Bayesian Networks

    Yildirim, Cansu / Soken, Halil Ersin | IEEE | 2024