Abstract：This paper proposes a DC distribution network fault location method combining current integral variation trend and temporal convolutional network (TCN)-support vector machine (SVM), to distinguish and locate DC distribution network faults, and lay the foundation for DC distribution network protection. Firstly, the integral sequence of fault current is calculated, and the integral sequence is decomposed by variational mode decomposition (VMD) algorithm. The eigenvalues of the decomposed high frequency intrinsic mode function are used as the input eigenvectors of the combination model of TCN and SVM, and the fault lines are located and the fault types are determined. The simulation results show that the scheme can not only locate the fault line quickly and identify different faults accurately, but also has good adaptability and certain anti-interference ability.
祝光思涵, 洪翠. 基于电流积分与时序卷积网络-支持向量机的直流配电网故障定位[J]. 电气技术, 2025, 26(2): 1-13.
ZHU Guangsihan, HONG Cui. Fault location of DC distribution network based on current integration and temporal convolutional network-support vector machine. Electrical Engineering, 2025, 26(2): 1-13.
2025, Vol. 26 
