Slope Stability Analysis Based On Improved Particle Swarm Algorithm

Landslide is a destructive geological disaster to mankind, environment, resources, etc. how economic, safe and reliable design slope or analysis the stability of slope is become more and more prominent.BP neural network have many advantage like nonlinear, self-learning ability, generalization ability, etc. so it can descript the complexity of the non-linear relationship between the slope stability and its influencing factors, but neural networks have defects too,it easy to fall into local optima, slow convergence, etc. In order to improve the performance of BP network, use particle swarm algorithm to improve neural network, particle swarm algorithm has global optimization capability, optimizing speed fast and other characteristics, it may lose its biodiversity because of its fast convergence in the late phase, for such shortcomings of particle swarm, we propose dual phase interaction of the Particle Swarm algorithm, optimization process is divided into two stages, the first stage is a standard particle swarm algorithm, the stage is intend to ensure the convergence of particle swarm, the second stage is the stage of interaction between the particles, this stage to increase the diversity of the particles, PSO can enhance the ability to escape from local optima To this way. Use improved PSO algorithm to optimize BP neural network, BP network can exert its advantages fully, in the end use improved particle swarm optimization BP neural network model of intelligent to assess the stability of slope.This paper introduces the research background of slope stability and the evaluation methods develop status of at home and abroad, we also introduce the formation conditions and contributing factors of landslide; Introduced the BP neural network composition, structure of the network, the network learning process, as well as its defect, its shortcomings were compensate with improved PSO, introduced the idea of PSO algorithm, the basic principles of optimization processes and defects of it, A dual stage interaction patterns to improve PSO algorithm, Let the performance of the algorithm better than before, take the particle swarm algorithm to make the BP network better play to their strengths; describes the main affect factors of the xiabi Highway K85 Slope and take improved Particle Swarm Optimization neural network model and BP neural network model to predict its stability, the results show that improved PSO-BP neural network model has better applicability than the BP neural network model, it is a new method of slope evaluation.