Diagnosis of DGA based on fuzzy and ANN methods

The accuracy of diagnosis with DGA (Dissolved Gas Analysis) is not satisfied though it is used widely in oil-immersed insulation. In this paper, the FART (Fuzzy Adaptive Resonance Theory) network is constructed to enhance the diagnostic accuracy of DGA method. Two input manners are discussed, one is the membership function of dissolved gases based on statistic method, another is the principal component analysis method. Finally, the practical examples had been given for checking the results of insulation diagnosis, it is shown that with the method introduced, the diagnosis will be more effective.