A Bayesian Information System for Predicting Stator Faults in Induction Machines

The approach adopted in this paper focuses on the faults prediction in asynchronous machines. The main goal is to explore interesting information regarding the diagnosis and prediction of electrical machines failures by the use of a Bayesian graphical model. The Bayesian forecasting model developed in this paper provides a posteriori probability for faults in each hierarchical level related to the breakdowns process. It has the advantage that it can give needed information’s for maintenance planning. A real industrial case study is presented in which the maintenance staff expertise has been used to identify the structure of the Bayesian network and completed by the parameters definition of the Bayesian network using historical file data of an induction motor. The robustness of the proposed methodology has also been tested. The results showed that the Bayesian network can be used for safety, reliability and planning applications.