Single Pole Ground Fault Location Method for LCC-MMC Hybrid DC Transmission Lines Based on Extremely Randomized Trees

Abstract

Aiming at the difficulty of locating single pole ground fault in LCC-MMC hybrid DC transmission lines, a single pole ground fault location method for transmission lines based on Variational model decomposition (VMD) - Decision Tree (DT) feature selection and Extremely randomized trees (Extra-Trees) is proposed. First, the collected two terminal fault waveforms are connected in series, sent to the VMD algorithm for decomposition, and then the decomposed modal components are connected in series with the original waveform. Then, the importance of features is calculated using the feature importance quantification method of DT algorithm, and the features with higher importance are constructed into a new feature set. Finally, the Extra-Trees model is built and trained with the new feature set to make it have the ability of fault prediction. A ±800kV LCC-MMC hybrid DC transmission system is built to verify the proposed method. The simulation results show that the proposed method can accurately locate the single pole ground fault location.