Abstract: In recent years, the seismicity of the Changning area in the Sichuan Province has increased significantly. Seismogenic models and seismogenic structures on the background of structural loading coupled with human activities have gradually become the focus of research in the field. Using abundant and accurate hypocenters in the Changning area, we established a program for automatically extracting morphological fault features by using machine learning algorithms including supervised classification and clustering. The method provides a reliable, detailed model of seismogenic faults for relative researches. As a result, four earthquake clusters were identified by clustering analysis, and four fracture planes were fit based on the distribution of hypocenters. The fracture plane on the Changning anticline spreads in NW-SE direction along a high velocity body beneath the Shizitan anticline. The fracture plane is straight with steep dip angle, and inclines SE. The three fracture planes in the inner part of the Jianwu syncline are mainly distributed in small scale on the limbs of the syncline with strike of NW, NNE, and NNW, respectively. They are also distributed in the periphery of the high-velocity body at the core of the Jianwu syncline, and their spreading directions are consistent with the strike of nodal planes of three main focal mechanism solutions in this area. Among these fracture planes, the Xincheng fracture plane extends deep to about 20 km and dips ENE with dip angle 70°. Based also on the geological tectonic settings and velocity structure, the fracture planes mainly exist in weak tectonic zones, such as the nucleus of the anticline and the limbs of the syncline. In particular, the fragile zone around the high-velocity body is more likely to rupture and nucleate under the loading of tectonic stress and industrial mining, forming new seismogenic structures.