Research on Transformer Fault Diagnosis Method based on GWO Optimized Hybrid Kernel Extreme Learning Machine

Abstract

The intelligent fault diagnosis of power transformer is the main link to promote the development of smart grids, but the traditional single intelligent diagnosis algorithms cannot process the huge amount of the incomplete fault information of the transformers effectively, resulting in low accuracy of fault diagnosis. Therefore, combining the dissolved gas analysis (DGA) technology, a transformer fault diagnosis method based on Gray Wolf Optimization algorithm (GWO) optimized hybrid kernel extreme learning machine is proposed in this paper. Firstly, based on Mercer's theorem, a hybrid kernel extreme learning machine model is constructed by combining the local radial basis kernel function and the global polynomial kernel function. Secondly, the parameters of hybrid kernel function can be optimized by the GWO algorithm. Meanwhile, the Logistic chaotic map is used to generate the initial population parameters of the GWO algorithm, which makes the distribution of initial population parameters as evenly as possible to avoid adverse effect of convergence speed and the optimization results. The results show that the presented algorithm in this paper improves the accuracy of transformers fault diagnosis compared with the BP neural network and the extreme learning machine algorithm, which has the strong learning ability and generalization performance.