Abstract: In response to the issues of susceptibility to noise interference and relatively low line selection accuracy during single-phase grounding faults in the low-current grounding systems of distribution networks, this paper proposes a multi-criteria fusion-based fault line selection method. This approach integrates fast fourier transform (FFT) and variational mode decomposition (VMD) techniques to extract three characteristic components of the zero-sequence current from each line, thereby constructing multi-dimensional input features. On this basis, the random forest classifier optimized by the particle swarm optimization algorithm (PSO-RF) is proposed in this study. The model is trained using the faulty line as the output label, enabling accurate identification of fault lines in new fault data. A single-phase grounding fault simulation model was established, using MATLAB/Simulink software for simulation experiments to validate the effectiveness of the proposed method, and comparisons are made with existing algorithms. The results demonstrate that the proposed method exhibits significant superiority in fault line selection accuracy, showing promising potential for engineering applications.

    Key words: fault line selection; low-current grounding; particle swarm optimization; random forest classifier; multi-criteria fusion