Experience Feedback and Probabilistic Graphical Model for Failure Causes Isolation

In this article, a probabilistic graph of Bayesian network type is proposed for the isolation of system failure causes. This graph, merge various knowledge on monitored system such as dysfunctional analysis, expert judgment and feedback data. Each new event is considered to adjust the probabilities of the network and effectively guide the new search for the cause of failure. The goal is to improve the isolation process for an efficient maintenance and higher system availability. A case study is used to illustrate the benefits of the proposed approach in isolating monitored system failure causes.