一种具有学习能力的人工蜂群优化算法

An Artificial Bee Colony Optimization Algorithm with learning Ability

针对传统人工蜂群算法对高维多峰问题优化时常常易陷入局部最优解,导致算法早熟收敛,而对单峰问题优化时收敛速度不够快的不足.为了使算法的性能得到进一步的优化,提出了一种带有双重学习能力的人工蜂群改进算法(DLABC).DLABC算法中采蜜蜂对蜜源邻域进行局部搜索时,增加个体对其自身最优值的自我学习能力和对种群中的其他个体最优值的社会学习能力,使用随着迭代次数动态变化的学习权重因子来平衡种群的局部搜索和全局探测能力,防止算法早熟收敛和加快收敛速度.通过对标准函数仿真测试验证,和几个改进的人工蜂群算法比较,DLABC算法的优化性能有了较大程度的提高.

When using traditional artificial bee colony algorithm to optimize the multi-dimensional and multimodal problems,usually only the local optimal solutions are obtained,and the rate of convergence is slow when the classical algorithm is used to handle the single modal problems.In this paper,we propose an artificial bee colony algorithm with double learning ability,which are called DLABC.This algorithm improves the self-learning ability of its optimal value for the individual,and improves the social learning ability of optimal value for other individuals in the colony,while it is searching the bee source neighborhood locally.Learning weight factor is introduced to balance the local search and global search of colony,and it changes dynamic with the numbers of iteration,which can avoid premature convergence and accelerate the rate of convergence.Finally,the experimental evaluations show that DLABC algorithm is more efficient than other improved artificial bee colony algorithms.