自适应多种群粒子群算法及其应用分析

Analysis of Adaptive Multi Swarm Particle Swarm Optimization Algorithm and Its Applications

针对粒子群算法易陷入局部最优和全局搜索能力差等问题,将整个粒子群按照一定比例分为三个子种群,并根据每个子群中粒子的性质,按照不同的位置更新方式分别对粒子进化,以保证种群的多样性,平衡算法的全局和局部搜索能力。并且为了平衡算法局部和全局搜索的矛盾,自适应地更新算法的学习因子,并引入了莱维飞行和混沌搜索策略。新算法在4个测试函数上与其它3种算法进行了比较,结果表明新算法具有更好的性能。并将算法用于求解线性超定方程组,结果表明算法求出的解非常理想。

Since the standard particle swarm optimization(PSO)algorithm is easy to fall into local optima and its global search ability is poor.In this paper,the entire particle swarm is divided into three subgroups according to a certain percentage.And depend‐ing on the nature of particles in each subgroup,particles are updated according to the different evolve strategies.This ensure the di‐versity of the population and the balance of global and local search ability of the algorithm.In order to balance local and global search capabilities,learning factors are updated adaptively.And levy flight and chaotic searching is introduced.New algorithms on four test functions are compared with the other three kinds of algorithms,the results show that the new algorithm has better perfor‐mance.Algorithm has also been used to solve the overdetermined linear equations,the results obtained show that the algorithm’s so‐lutions are very ideal.