Abstract: When a single-phase earth fault occurs in the distribution network with neutral earthing via arc suppression coil,the zero-sequence current of each line is rich in transient content. On the basis of a theoretical analysis of the zero sequence current characteristics of single-phase ground faults,a clustering line selection method is proposed through zerosequence current waveform similarity measurement based on Dynamic Pattern Matching distance（DPM）. Firstly, the transient zero-sequence current of the line after single-phase grounding is dynamically segmented with equal segment points and extreme points as the end points,the segmented sequence configurational patterns are expressed symbolically according to each segment average and variation trend,and the pattern matching distance is defined by the similarities and differences among patterns. Secondly, the dynamic pattern matching distance matrix among zero-sequence current waveforms is obtained using the Dynamic Time Warping（DTW）principle.Finally,the single-phase grounded line selection is realized by using the fuzzy C-mean clustering algorithm. Many kinds of cases such as different fault initial angle, asynchronous sampling,information loss,two-point ground faults,arc faults are simulated. The results show that the proposed method has higher fault discrimination than DTW and effectively improves the accuracy and fault tolerance of fault line selection in the distribution network,and has certain practical significance for the safe operation of the distribution network.