Constrained Particle Swarm Optimization Algorithm Applied to Slope Stability

This article addresses the problem of the stability of a slope within the framework of the yield design theory (YDT), which is a rigorous method that avoids assumptions that could affect the obtained results. The case of a nonvertical slope with an angle α at the toe and β at the crest using multirotational rigid-block failure mechanisms is investigated. For fixed geometrical and mechanical parameters, the implementation of the upper-bound kinematic method permitted the definition of an optimization problem with five decision variables characterizing the failure mechanism. A constrained global optimization problem was then formulated and converted into an equivalent unconstrained one to get an upper-bound estimate of the studied slope-stability factor. A well-known particle swarm optimization (PSO) algorithm was then used for both its simplicity and efficiency in unconstrained optimization problems. The previous results for the stability factor have been improved and generalized.