改进基于记忆的人工蜂群算法

基于记忆的人工蜂群算法(ABCM)通过记住成功使用的邻居和系数指导人工蜂群下一步的搜索,需消耗多次函数评价收敛到吸引子,且始终使用与上次相同的排斥系数,造成收敛速度不快、多样性不足,易陷入局部最优解.提出一种改进ABCM(IABCM),当使用吸引系数时,候选解只消耗一次函数评价收敛到吸引子,如果候选解好于当前解,则替换当前解,否则直接删除该记忆,这样可以利用尽量小的代价得到尽量大的收益.当使用排斥系数时,该系数的数值部分重新随机生成,以增加多样性和随机性,有利于算法跳出局部最优解.在22个不同类型函数上的实验表明,IABCM在收敛速度和精度方面明显优于ABCM.

An Improved Artificial Bee Colony Algorithm with Memory

Artificial bee colony algorithm with memory ( ABCM ) memorizes successful coefficients and neighbors to guide the further foraging of the artiflcial bees. ABCM consumes many function evaluations to converge to the attractors and use the same rejection coefficients as last time, which easily results in slow convergence, low population diversity and falling into the local minima. In the improved ABCM ( IABCM) , the candidates converge to the attractors consuming only one function evaluation, and the candidate will replace the current solution if the former is better than the latter. Otherwise, the memory will be deleted directly. By doing so, IABCM can get the most profit at a minimum cost. When the rejec-tion coefficients are used, the numeric parts will be regenerated randomly to enhance the diversity and randomness, which is beneficial to help the algorithm to escape the local minima. Experiments on 22 functions with different characteristics demonstrate that the IABCM is significantly better than ABC and ABCM in terms of solution quality and convergence speed.