Abstract: Spatial distribution of faults is important to earthquake prediction. Spatial distribution characteristics of microearthquakes can help to detect faults. We use DBSCAN clustering method to analyze catalog of microseisms from auto-picking algorithm in east of Sichuan-Yunan area and then separate microseism sequences of different seismic zones. For every seismic zones in this area, we further employ waveform information of earthquakes and do second clustering. By second clustering, we separate different seismic sequences belong to different faults, which depicts fine fault structure in this area.