ANFIS based identification and location of paper insulation faults of an oil immersed transformer

Abstract

In this paper, a softcomputing technique namely, Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used to identify and locate the incipient faults developing in an oil immersed power transformer. Dissolved gas analysis (DGA) of the transformer insulating oil helps in effective condition motoring of a transformer. A number of interpretation standards have been developed to identify the fault type based on the outcome of the DGA. The proposed ANFIS model is based upon an implementation of the DGA interpretation standard IEC-599. The ANFIS model has the ability to both identify the incipient fault type in the power transformer as well as locate the fault. The model's fault diagnosis and fault locating capability has been tested using reported fault cases in various literatures. Results of the tests presented in this paper clearly indicate an encouraging trend towards a more reliable model.