Transformer Fault Diagnosis Method Based On Bayesian Networks

In the view of the special status and the important role transformers play in the power system,in order to find all kinds of troubleshooting methods,this paper makes a in-depth study on the transformer common failures and its causes. If transformer failures happen,the incomplete or uncertain fault information will lead to greatly increase the difficulty of dealing with failure,sometimes even delay the repair time.By comparing the characteristics of currently the more popular fault diagnosis methods, Bayesian network shows great advantages in this area, and determined the transformer fault diagnosis method based on Bayesian network. And because of the uncertainty of TAN model and the subjectivity of causal model, this article identified the Bayesian network as the fault diagnosis method. Moreover, after a simple case analysis to prove that the model has strong reliability and good accuracy. In the analysis based on complete data, simply introduce the EM and SEM algorithms in the case of incomplete sample data.Given a basic fault diagnosis method based on Na?ve Bayesian network. Furthermore,combined with the project, simulate the network model on the basis of analysis of a certain scale samples. Through analysis of the results shows the Bayesian network for fault diagnosis in the area of outstanding capability and broad prospects.Finally, the paper discusses the application of Bayesian networks to fault diagnosis of focus and research.