An ANFIS-based transformer insulation fault diagnosis method using emotional learning

To tackle the flaws in transformer fault diagnosis such as long computing time, weak generalized ability and fuzzy knowledge acquisition difficulty, a self-adaptive neuro-fuzzy inference system (ANFIS) is proposed based on emotional learning in this paper. The method can automatically adapt itself to the change of input information characteristics, and compensate for the flaws of the imperfectness of the 3-ratio-code. In addition, Due to applying emotional learning, the structure complexity and learning time of the networks are dramatically reduced, and the forecast accuracy is also improved. Finally, a practical example in transformer fault diagnosis indicates the availability of the method.