基于双层进化的多种群并行人工蜂群算法  

Parallel evolution with multi-populations artificial bee colony algorithm based on dual evolution structure

作　　者：赵丹亚[1] 张月[1] 

机构地区：[1]首都经贸大学信息学院,北京100070

出　　处：《计算机工程与设计》2015年第1期178-183,共6页Computer Engineering and Design

基　　金：北京市教育委员会科学研究计划基金项目(SM201410038013)

摘　　要：针对人工蜂群算法(ABC)容易陷入早熟收敛等不足,引入文化算法双层进化结构和多种群并行进化思想,提出基于双层进化的多种群并行人工蜂群算法(PMABC)。将采蜜蜂群划分为具有不同搜索策略的子种群并行进化,平衡全局开发能力与局部搜索能力,避免算法过早陷入局部最优。采用双层进化结构,采蜜蜂群作为种群空间寻找可行解,追随蜂群作为信仰空间,记忆采蜜蜂群搜索的优质蜜源并继续搜索。将其搜索结果用于指导蜂群寻优,可加速算法收敛,提高收敛精度。通过6个经典的适应度测试函数仿真验证了该算法能够有效避免陷入局部最优,具有较快收敛速度和较高收敛精度。Considering the shortcomings of premature convergence and rough convergence precision in artificial bee colony （ABC）, the dual evolution structure and the idea of parallel evolution with multi-populations were introduced, a novel parallel evolution with the multi-populations artificial bee colony algorithm based on the dual evolution structure （PMABC） was pro- posed. The employed bee group was divided into a number of different sub-groups using different optimization strategies which e- volved in parallel. The global exploration and local exploitation ability were balanced, and the population diversity was improved, the prematurity was avoided. In the dual evolution structure, employed bee groups were taken as population space which evolved independently and in parallel, and on-look bee group was taken as belief space which continually evolved the current best solution contributed by population space. The evolved solution was then used to guide the evolutionary search to accelerate the conver- gence rapidity and to improve the convergence precision. Simulation results of six benchmark functions show that the PMABC algorithm can effectively avoid prematurity, accelerate the convergence rapidity and improve the convergence precision.

关 键 词：人工蜂群算法 函数优化 多种群 并行进化 双层进化结构 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]