Abstract: In this paper, the dynamic time warping （DTW） method is introduced so as to solve the problem that the two observation data cannot be compared due to their different time length. On the basis of the principle of dynamic time warping technology, this paper verifies the feasi-bility of identifying the conformity of seismic observation data by testing sample data. At the same time, using the actual fault observation data in western Yunnan, the similarity between the anomalous morphology before the 1996 Lijiang M_S7.0 earthquake and the current anomalous morphology is studied. The results show that: ① The DTW algorithm can be used for similarity matching of precursor data in seismological field; ② Time-inconsistent precursor observation data can be expressed by cumulative warping distance, the shorter the cumulative distance, the more consistent the curve shape; ③ The DTW algorithm can be used for automatic extraction of precursor data similarity of template-making, which can improve the current work efficiency in the case of only relying on manual tracking precursor curve changes; ④ Considering from the point of view of pattern recognition, the data of Xiaguan  cross-fault leveling are consistent with those before Lijiang M_S7.0 earthquake in 1996 and Wenchuan M_S8.0 earthquake in 2008.