摘要
人工蜂群(artificial bee colony,ABC)算法在解决多峰函数优化问题时经常会陷入局部最优,使得算法过早停滞,而在解决单峰问题时往往出现收敛速度过慢的问题。针对上述不足,为了进一步提高算法的优化性能,提出了一种基于交叉突变的人工蜂群(intersect mutation ABC,IMABC)算法。IMABC算法将整个蜂群依据其适应度值优劣进行划分,引入种群划分参数,对不同种群中的个体运用交叉突变算子,有效地平衡了种群的局部开采与全局探测能力,避免早熟收敛和提高收敛速度。从对基本函数的测试上可以看出,IMABC相对于GABC、IABC、ABC/best等改进的ABC算法,优化能力有了较大的提高。最后,将IMABC用于优化K-means算法,验证了该方法具有一定的实用性。
It is easier to get trapped in the local optima in optimizing complex muhimodal functions while slow convergence speed in solving unimodal functions using ABC algorithm. In order to overcome these shortcomings and improve the optimiza- tion performance, this paper introduced an improved ABC algorithm based on intersect mutation strategies. According to the IMABC algorithm, the whole bee colony could be divided into two sub-populations by introducing the parameter M. An inter- sect mutation operator could be applied to the individuals belong to different sub-populations and got a balance between local exploitation and global exploration among the populations avoiding premature convergence and enhancing convergence speed. The experiments test the performance improvement comparing the other improved ABC algorithm, such as, GABC, IABC, ABC/best. At last, the IMABC algorithm can be applied to optimize K-means algorithm and a better experimental result vali- date its practicability and effectiveness.
出处
《计算机应用研究》
CSCD
北大核心
2014年第5期1336-1341,共6页
Application Research of Computers
基金
国家自然科学基金资助项目(61075049)
安徽省级自然科学研究基金资助项目(KJ2013A009
KJ2012B038)
安徽省优秀青年人才基金资助项目(2011SQRL018)
安徽大学青年科学研究基金资助项目(KJQN1015)
关键词
人工蜂群算法
交叉突变算子
差分进化
函数优化
K-均值
artificial bee colony algorithm
intersect mutation operator
differential evolution
function optimization
K-means
