Transformer Fault Diagnosis Based on DLH-GWO-SVM

Abstract

In the field of transformer fault diagnosis, there are many diagnostic methods. These diagnostic methods are either not high accuracy or are used too long. In this paper, a dimension learning-based hunting (DLH) search strategy optimization grey wolf algorithm support vector machine (DLH-GWO-SVM) model was proposed. In the process of initializing the grey wolf population, logistic chaotic mapping was used to improve the quality of the initial wolf position. In the Wolf screening process, the DLH search strategy was adopted to solve the problem that GWO was prone to fall into the local optimal solution, and the improved Grey wolf algorithm (IGWO) was formed. Finally, the transformer fault types were classified by support vector machine (SVM), and IGWO was used to optimize the penalty factor and kernel parameter (g) of the SVM. Through the experimental analysis, the accuracy of transformer fault diagnosis can reach 95.83% after the application of the DLH-GWO-SVM model.