Optimal dissolved gas ratios selected by genetic algorithm for power transformer fault diagnosis based on support vector machine

Abstract

Dissolved gas analysis (DGA) of oil is used to detect the incipient fault of power transformers. This paper presents a new approach for transformer fault diagnosis based on selected gas ratios concentrated in oil and support vector machine (SVM). Firstly, based on IEC TC 10 database, the optimal dissolved gas ratios (ODGR) are obtained by genetic algorithm (GA) that is designed for simultaneous DGA ratios selection and SVM parameters optimization. Three traditional methods, namely, DGA data with SVM and back propagation neural network (BPNN), IEC criteria, and IEC three-key gas ratios with SVM and BPNN are employed for effectiveness comparison. The fault diagnosis results of IEC TC 10 database show that the proposed ODGR with SVM may be used as an alternative tool for transformer fault diagnosis. In addition, the robustness and generalization ability of ODGR is confirmed by the diagnosis accuracy of 87.18% of China DGA samples. The obtained results illustrate that it is preferable to apply the proposed ODGR to transformer fault diagnosis with the assistance of SVM.