Abstract: A novel particle swarm optimization（MSPSO） with main-sub structure is proposed and it implements optimization for objective function and system parameters such as inertia weight,acceleration coefficients,maximum velocity simultaneously.One particle in the main program is treated as a set of system parameters in the subprogram.Under these control parameters,objective function in the subprogram is optimized with Particle Swarm Optimization（PSO） and global best value of subprogram is returned to main program and is viewed as a fitness value in the main program.At the same time,the fitness function of main program is also optimized with PSO.Experimental results show that the performance of MSPSO is superior to that of PSO.MSPSO guides also parameters selection for PSO.

Key words: particle swarm, parameters selection, optimization algorithm, global optimization

摘要： 提出了一种具有主从结构的粒子群优化算法,该算法实现了惯性权重、加速因子、最大速度等系统参数与目标函数的同步优化。将主程序的一个粒子作为子程序的一组系统参数,在该组控制参数下使用基本的粒子群算法对子程序的目标函数进行优化,并把子程序优化所得的全局最优值返回主程序作为主程序的一个适应值,同时使用基本的粒子群算法对主程序的适应度函数进行优化。实验结果表明,该算法的优化性能较基本的粒子群算法有了显著提高。该方法对于粒子群算法的参数选择具有指导意义。

关键词: 粒子群, 参数选择, 优化算法, 全局优化