基于精英策略和Levy飞行的粒子群算法

Particle swarm optimization algorithm based on elite strategy and levy flight

为了提高粒子群算法的搜索效率,避免粒子过早收敛,将精英策略思想和自适应动态Levy飞行步长引入到粒子群算法中,提出了一种新的算法(ELPSO).首先依据精英策略(ES)的搜索机制来扩大粒子的搜索空间,利用精英粒子良好的反向搜索能力,使粒子避免早熟收敛.此外,使用自适应动态Levy飞行更新该算法对在进化中因发生早熟而无法进化到更好位置的粒子.利用改进的新算法(ELPSO)测试6个标准测试函数,并与标准的PSO算法和权重线性递减的粒子群算法(RWPSO)进行比较.结果表明,ELPSO算法的收敛速度和收敛精度都优于标准PSO算法和RWPSO算法.

In order to improve the search efficiency of particle swarm optimization,avoid premature convergence of particles.A new algorithm(ELPSO)is proposed by introducing elitist strategy and Levy flight into particle swarm optimization(PSO).First,according to the search mechanism of the elite strategy(ES),the search space of particles is expanded.Because the elite particles have good reverse search ability,the particles can avoid precocious convergence.In addition,the algorithm updates the location of particles using Levy to update particles that are precocious in evolution and can not evolve to better positions.The improved new algorithm(ELPSO)is used to test six standard test functions.The simulation results show that the performance of the ELPSO algorithm is significantly better than the basic particle swarm optimization algorithm.