Abstract: Due to factors such as arc suppression coils, transition resistors, and environmental noise, the fault signal characteristics of resonant grounding systems are weak, and the selection method based on a single criterion is difficult to ensure the reliability of the selection results. Therefore, this article proposes a fault line selection method for resonant grounding systems that integrates the Dynamic Time Bending Distance (DTW) algorithm and Hilbert envelope energy. Firstly, the DTW distance algorithm is used to quantitatively characterize the waveform similarity between the current sequences of each line. Secondly, to avoid potential blind spots in line selection using a single criterion, the Hilbert envelope energy is introduced to measure the amplitude of high-frequency components in transient zero sequence current signals. Thirdly, to enhance the data processing ability and efficiency of the proposed line selection method, an improved K-means clustering algorithm is used to classify the fault feature dataset. Finally, a 10kV distribution network simulation model is built in PSCAD/EMTDC to verify the feasibility and accuracy of the proposed method.