Performance Assessment of Anfis Applied to Fault Diagnosis of Power Transformer

Continuous monitoring of Power transformer is very much essential during its operation. Incipient faults inside the tank and winding insulation needs careful attention. Traditional ratio methods and Duval triangle can be employed to diagnose the incipient faults. Many times correct diagnosis due to the borderline problems and the existence of multiple faults may not be possible. Artificial intelligence (AI) techniques could be the best solution to handle the non linearity and complexity in the input data. In the proposed work, adaptive neuro fuzzy inference system (ANFIS), is utilized to deal with 9 incipient fault conditions including healthy condition of power transformer with sufficient DGA transformer oil samples. Comparison of the diagnosis performance of both the methods of ANFIS and the feasibility pertaining to the problem is presented. Diagnosis error in classifying the oil samples and the network structure are the main considerations of the present study.