Research and Application of Intelligence Fault Diagnosis Based on Bayesian Networks

摘要

According to the uncertain factors and complex correlativity exist in the diagnosis of equipment faults and the disadvantages of uneasy update in fault tree method's, we put forward a model of intelligent fault diagnosis which based on Bayesian networks. The Bayesian networks is built by expertise, and take the data as the evidence which gathered by multi-sensors, and also utilize it to update belief of network, and make final decision. In the application of speed governor fault diagnose, via over 100 times factual diagnosis, the average error is 4.6%, it is proved that it is reliable to put Bayesian networks in fault diagnosis of speed governor.