Abstract: The strength reduction method is widely used to slope stability analysis. Different criteria for slope instability have their own characteristics. The mutation in the displacement reduction coefficient curve is a simpler and more practical criterion for instability. Displacement mutation feature point is selected by cluster analysis. Displacement mutation of feature point is identified using self-organizing neural network change point analysis method. An improved criterion for feature point displacement mutation is proposed based on above-mentioned method. The fusion method of clustering self-organizing neural network change point analysis is more objective and mathematically based. For typical examples, the safety factor calculated by clustering-change point analysis method is closer to the judging value, compared with the safety factor calculated by the strength reduction method based on FLAC3D.