Transformer Fault Diagnosis Based on Reasoning Integration of Rough Set and Fuzzy Set and Bayesian Optimal Classifier

In accordance with intelligent complementary strategies, a new transformer fault diagnosis method is proposed based on rough set (RS) and fuzzy set (FS) and Bayesian optimal classifier in this paper. Through RS reduction, the diagnostic decision table is greatly simplified and fault symptoms information is compressed, dramatically, and the minimal decision rules can be obtained. In the light of the minimal decision rules, the complexity of Bayesian reasoning and difficulties of fault symptom acquisition are dramatically decreased. Moreover, probability reasoning may be realized applying Bayesian optimal classifier, it can be used to describe the characteristics of fault information and investigate the fault reasons of transformer. In the end, a practical application in transformer fault diagnosis indicates that the proposed method is very effective and intelligent and ubiquitous. Key-Words: Rough set; Fuzzy set; Bayesian optimal classifier; Fault diagnosis; Information entropy, Intelligent complementary; Transformer