摘要
给出粗粒度并行遗传算法对于子种群间迁移策略的一种改进,即每隔一定的进化代数,各子种群与公共池交换最佳个体和代表个体.改进后的迁移算子淡化了子种群间交换个体时的拓扑结构,提高了各子种群的多样性.对复杂非线性函数求极值的仿真结果表明,改进迁移算子后的粗粒度并行遗传算法相对于固定拓扑结构的粗粒度并行遗传算法,得到最优解的进化代数提前,并且最优解的质量有所提高.
An improved migrate strategy of coarse-grained parallel genetic algorithm is pro- posed. The sub-populations exchange the best individual and representation individual with sharing pool every specified interval of generations. Diversity of sub-populations is enforced and topology structure is ignored with the improved migrate operator. Emulation experiment results in complicated nonlinear functions show that it has better quality of solution and reduced interval of generations compared with the coarse-grained parallel genetic algorithm using fixed topology.
出处
《福建师范大学学报(自然科学版)》
CAS
CSCD
北大核心
2013年第1期42-47,共6页
Journal of Fujian Normal University:Natural Science Edition
基金
福建省教育厅资助项目(JB11036)
关键词
粗粒度并行遗传算法
迁移算子
种群多样性
代表个体
coarse grained parallel genetic algorithm
migrate operator
population diversity
representation individual