Research And Implementation Of Fault Integrated Diagnosis Expert System For Power Transformers

Power transformer is important electrical equipment in the electrical system, and the running status of the transformer affects directly the safety of the electrical system. Therefore, it is vital for electrical department to find out the potential faults of the transformer as soon as possible. With aids of the expert system and the fuzzy theory, the intelligent transformer fault diagnosis is achievable. And the intelligent realization accelerates the engineering application of the research results of the transformer fault diagnosis.In general, the IEC-three ratio method is an effect method to be used in the transformer fault diagnosis based on the dissolved gases analysis (DGA) results. In this thesis, a novel transformer fault diagnosis method is presented by combining the IEC three-ratio method and the artificial intelligence techniques. The proposed method constructs a transformer fault integrated diagnosis expert system consisting of the IEC three-ratio method based on production rule, fuzzy diagnosis, and visual technique. In this expert system, the IEC three-ratio method based on production rule is used when the gas ratio is far from the three-ratio bounds, while the fuzzy diagnosis is utilized to increase the probability of correct diagnosis when the gas ratio lies in the three-ratio bounds. Furthermore, in this expert system, we present the visualized cubic schema method and visualized Duval triangular schema method to make the subscriber have a clean and intuitionistic view of the position of the transformer faults. In fact, the two visualized method can decrease the computational load of the fuzzy diagnosis, and the visualized Duval triangular schema method can always give an assistant diagnosis result.Simulation results are also given to validate the proposed integrated diagnosis expert system. In particular, the proposed integrated diagnosis expert system can decrease effectively the probability of wrong diagnosis and avoid the problem of having no diagnosis results. Further, we can use the three methods in the expert system to check some diagnosis results simultaneously, which will result in an increased probability of correct diagnosis.