Synthesized Diagnosis On Transformer Faults Based On Bayes Networks

The normal and stable operation of power transformer is the basic guarantee of reliable power supply for the whole power system.But with higher power transformer voltage grade,the structure is more complex,the fault is also becoming more complicated and diverse,so it is difficult to complete and correct the fault samples collected,it will greatly increase the difficulty of transformer fault diagnosis.Therefore,based on the analysis of the fault diagnosis of power transformer after all kinds of methods,selection based on the improved three ratio method,using Bayesian network to deal with the advantage of simplicity,causality with a simple Bayesian networks to build a network model of fault diagnosis,and use the probability calculation to quantify the relationship between variables,to make it to judge uncertainty factors in power transformer operation caused by fault has obvious advantages;In order to verify the simple Bayesian network combined with rough set theory of diagnosis method's ability of dealing with incomplete information,this thesis collected a large number of transformer fault data samples,using Matlab to simulate,and the results of simulation is better than the method that only use the simple Bayesian network,so the simulation results of validation of the presented method can effectively overcome the lack of information or incomplete caused difficulties for transformer fault diagnosis,effectively improved the accuracy of fault diagnosis.