Fault Location of Ground Faults in Distribution Network Based on Empirical Wavelet Transform and Convolutional Neural Networks

Abstract

Fault location of transmission lines is an important guarantee for the stable operation of the distribution network. Aiming at the problem that the fault characteristics are not obvious when a single-phase-to-ground fault occurs in the distribution network, a new fault location method that combines empirical wavelet transform (EWT) and convolutional neural networks (CNN) is proposed. Firstly, The EWT is applied to the zero-sequence currents of different lines to obtain intrinsic mode function (IMF) components with different frequency domain characteristics. Then, the modal component with more fault features is extracted by the magnitude of the kurtosis value, and it is converted into a two-dimensional image by pseudo-color coding. Finally, the time-frequency images are used as the input of the CNN model for iterative training to construct a fault localization model. The experimental results show that the proposed approach in this paper can accurately realize the fault location in distribution network transmission lines under different fault conditions and environmental noise.