混合粒子群算法搜索土坡危险滑动面

SEARCH FOR THE CRITICAL SLIP SURFACE USING MIXED VERSION OF PARTICLE SWARM OPTIMIZATION ALGORITHMS

基本粒子群优化算法存在着对惯性因子敏感、计算量大等缺点,通过借鉴和声搜索算法产生新解的策略和不连续飞行假定,构成了混合粒子群算法。首先,当粒子飞行超越边界时,采用和声搜索算法产生新解;此外还引入了不连续飞行假定,即在每次迭代步中,随机选择一些个体更新速度、位置向量,以利于减少计算量。随机给定10组参数,分别利用基本粒子群优化算法和混合粒子群优化算法对某复杂土坡的最危险滑动面进行了搜索。比较发现,混合粒子群算法能在较短的计算时间内得到更好的结果。

Such disadvantages as sensitivity to ：inertia coefficient, time consuming etc. are always encountered in the application of particle swarm optimization algorithm. The procedure used in harmony search algorithm and the assumption of discontinuous flying are imported into the particle swarm optimization algorithm. Firstly, when the boundaries are exceeded, the harmony search algorithm is implemented to obtain a new trial solution; and secondly, random number of particles instead of static number are allowed to update their velocities and positions in order to decrease the time consumed by the algorithm. The particle swarm optimization algorithm and the mixed version of particle swarm optimization algorithm are applied to the determination of critical slip surface of given complicated soil slopes. The comparative study shows that the mixed version of particle swarm optimization algorithm tends to find better solution spending httle time than particle swarm optimization algorithm.