The Research Of Transformer Fault Diagnosis Based On Data Mining

It is very important to find incipient faults as early as possible. With the development of domestic power industry, the data of transformer condition proliferate, and new methods in virtue of intelligent technique are called for transformer fault diagnosis. In this paper, in-depth research has been done based on data mining techniques. New scattering method and normalizing method is presented. Design and build the models of Naive Bayesian Classifier and SVM classifier, and then analyse their applications in transformer fault diagnoses. For the incomplete data, design and build the models of Rough Set and SVM Regression, and by analyzing, present to estimate the missed datum by SVM Regression. The experiments on actual samples show that combining SVM Regression with Naive Bayesian Classifier can get admirable veracity, even if there are incomplete data.