Research on Fault Identification Method of Transmission Line Based on CEEMDAN and Current Traveling Wave

Aiming at the problem of unsatisfactory identification of transmission line fault diagnosis based on transient current traveling wave, a time series denoising method based on the combination of CEEMDAN and Pearson's correlation coefficient with a feature extraction method is proposed to realize the accurate diagnosis of transmission line fault types. First, the original transient current traveling wave signals collected by the distributed fault monitoring system of the transmission line are decomposed by CEEMDAN to derive the IMF components, and then by calculating the correlation coefficients of each IMF component, the IMF components with lower correlation are discarded and the original signals are reconstructed. Finally, the XGBoost machine learning model is used to identify the reconstructed signal and realize the fine diagnosis of transmission line faults. Comparison experiments with different models show that the method proposed in this paper has better diagnostic effect.