机构地区: 武汉理工大学计算机科学与技术学院
出 处: 《计算机工程与应用》 2006年第13期48-50,105,共4页
摘 要: 论文提出了一种新的PSO算法——改进的速度变异粒子群算法(iPSOVMO)。其变异策略是:在每次迭代循环中,对具有m个粒子的粒子群的每一维d上的速度的绝对值|v1,d|,|v2,d|,…,|vm,d|最小的速度vTd,d以一定的概率进行变异:使vTd,d随机而均匀地分布于[-vmax,vmax]上。对四个多峰的测试函数所做的对比实验表明,无论是全局版还是局部版,iPSOVMO都大大优于原始的PSO和传统变异PSO,也优于速度变异PSO(PSOVMO)。 In this paper,we have presented a new PSO algorithm,improved velocity mutation particle swarm optimizer (iPSOVMO).The mutation strategy is that in each iteration loop,on every dimension d of particle swarm containing m particles,the smallest velocity vTd,d of the absolute value of speed among |v1,d|,|v2,d|,…,|vm,d| is mutated according to some probability,make vTd,d distribute on [-vmax,vmax] stochastically and evenly.Through contrast experiments on four multi-modal testing functions,it shows that,with global version and local version,iPSOVMO is superior to original PSO and traditional mutation PSO greatly,and also superior to velocity mutation PSO(PSOVMO).
领 域: [自动化与计算机技术] [自动化与计算机技术]