一种多粒子群的协同优化算法

A Strategy of Particle Swarms Cooperative Optimizer

提出了一种多个粒子群优化的协同方法(MPSCO)。该算法用多个粒子群同时进化搜索,各粒子群定期交流信息,每经过若干代后相互交换各自的全局最优解,这样就使各粒子群在搜索过程中借鉴了其他粒子群的搜索成果,使搜索更具全局性。通过对Griewank函数等4个标准函数做优化测试,MPSCO的性能明显优于基本PSO。

A Multi Particle Swarms Cooperative Optimizer（MPSCO） algorithm is proposed in the paper. Particles swarms search the problem space in parallel, each changing its best position so far into the best position so far of its neighbor swarm after every certain generations. By experiments on four classic functions, it shows that this cooperation strategy is much more effective than classic PSO with one particle swarm.