Fault Diagnosis of Transformer Winding Based on VMD-SVM

Abstract

Considering the nonlinear and non-stationary characteristics of the transformer vibration acceleration signal obtained from the surface of the transformer tank, the variational mode decomposition (VMD) theory is introduced. Simulation analysis shows that the VMD decomposition has obvious advantages over EMD when the needle frequency is similar to the signal. It effectively avoids two types of modal aliasing and over-decomposition, and accurately reflects the characteristics of the source signal. Aiming at the problem that the two core parameters of the support vector machine are difficult to determine, the Pareto particle swarm method is used to perform multi-objective parallel optimization of the two core parameters of the support vector machine to obtain the optimal parameters. The VMD-SVM fault diagnosis model is tested using the transformer instance fault data, and compared with the other two methods. The instance test results show that the VMD-SVM proposed in this paper has the highest diagnostic accuracy and realizes the latent fault of the power transformer winding. accurate diagnosis.