A three‐dimensional limit equilibrium method for slope stability analysis integrating machine learning optimized intercolumn force function

Abstract

A calculation of the safety factor is the basis of slope stability analysis. The three‐dimensional (3D) limit equilibrium method is one of the most effective safety calculation methods, and ignoring the partial shear force between columns is the most common calculation assumption, which will lead to an unreasonable safety factor. By constraining the range of the intercolumn shear function and the position of the action point of the force between columns and integrating machine learning to optimize the net moment of the column, the simulated annealing algorithm is used to obtain the intercolumn shear function under the minimum net moment, and the safety factor is calculated by using boundary conditions. Finally, proving the accuracy of this method by calculating two examples of commonly used slopes. Example A verifies that the result of the proposed method is more conservative than that of other 3D calculation methods and that the coordinates of the action point of the intercolumn forces in the and directions have little influence on safety factor. Example B verifies the applicability of this method to various slope forms. This method fully considers the stress of soil under actual conditions, making the calculation results more reasonable.