Identification of Single-phase Grounding Fault Type in Distribution Network Based on VMD and SVM

Abstract

This paper put forward a method for the identification in single-phase grounding faults to effectively classify and identify resistance grounding faults and arc grounding faults in distribution networks, which is consist of Variational Mode Decomposition (VMD) and Support Vector Machine (SVM). Firstly, a simulation model describing the 10kV distribution network is established by PSCAD/EMTDC, and four types of single-phase grounding faults are constructed. Secondly, the different eigenmode functions are obtained through VMD decomposition of the collected fault signals of the zero-sequence current, and the permutation entropy that contributes the most is found by extract the typical characteristic of the fault signal. Thereafter, the eigenvector is constructed and imported into the SVM for identifying the single-phase grounding fault type. Finally, the simulation results in MATLAB indicate that the proposed method has more satisfactory practicability and can achieve nice accuracy compared with traditional intelligent algorithms.