Synthetic Fault Diagnosis Of Power Transformer Based On Bayesian Network

The transformer is one of the important equipment of the power system, whoserunning state is vital to the safe operation of the entire power system. Therefore, timelyand accurate diagnosis of the potential failure of the transformer has a practicalsignificance.Transformer fault diagnosis technology is far from perfect, so looking for away to deal with incomplete information and to improve the accuracy of transformer faultdiagnosis is the key to solve the transformer fault diagnosis problem. Therefore, theresearch topics of this dissertation focus on the above problems, and the details of it are asfollows:By studying the theory of the transformer fault diagnosis based on Bayesian networks,the dissertation propose Fuzzy-Rough Sets combining with simple Bayesian network,Which solve the problems of the properties of redundant and incomplete data in the fieldof fault diagnosis.The paper studies the structure and parameters of the learning process of Bayesiannetwork theory, and proposes to apply Bayesian network classifier to the transformer faultdiagnosis field. Through analysis, the dissertation proves that NBC and TAN models havethe ability to deal with incomplete information, and still has a high diagnostic capability inthe absence of information a little while. Moreover, the dissertation demonstrates theaccuracy and efficiency of the Bayesian Network Classifier for Transformer FaultDiagnosis with actual examples.To solve the problems of attribute redundancy and data integrity, the dissertationpresents a Bayesian network classifier combining with Fuzzy-Rough Sets and build themodel of NB-FR classify. The simulation results show that the program has overcome thedeficiencies of the traditional method on these issues, and effectively improves thetransformer fault diagnosis.