Fault Diagnostic Method of Transformers based on Hybrid ANN and Expert System

Abstract

Transformers are major equipments power system which affects not only the electrical energy availability of the area supplied, but also the economical operation of utilities. The most important concern about incipient faults in power transformer is that they may reduce the electrical and mechanical stress of the insulation. This proposed hybrid system utilizes dissolved gas in oil analysis techniques to diagnose for fault condition of power transformers. The Roger’s Four Ratio Method, NTT Flag point method, generation rate ratio method and Total dissolve combustible Gas method used of its in expert system. Both the Roger’s four -ratio and the NTT flag point method are implemented. These two methods are fuzzefied. Which is increasing the accuracy of both methods, as the generation rate ratio method is use to analyzed the trend of fault gases. All the four methods are combining into expert system, to achieve greater accuracy in diagnose of transformers. The topology and training data of its artificial neural network (ANN) are accurately selected to extract known as well as unknown diagnosis correlations. Both the ANN and expert system are used together into the hybrid system to take the advantages of the inherent positive features of each method to achieve greater accuracy in diagnosis of transformers. Test result show that the hybrid system has better performance then Roger’s ratio, ANN and expert system method used indivisibly.