摘要
针对原始人工蜂群算法存在收敛速度慢和易陷入局部最优的不足,提出了一种基于信息反馈和改进适应度评价的人工蜂群算法。首先,引入种群个体分量记忆机制对个体信息进行反馈以增强种群开发能力,加快算法收敛速度;其次,为避免因种群后期无法识别优秀个体导致的"早熟"现象,通过改进适应度函数增大不同个体间解的差异性;最后,采用最优蜜源引导机制改进淘汰更新函数以避免不良个体的产生。对标准函数的测试结果表明,改进后算法有较快的收敛速度和较高的收敛精度。
The artificial bee colony (ABC) algorithm converges slowly and easily gets stuck on local solutions; hence, an ABC algorithm based on information feedback and an improved fitness value evaluation is proposed. The algorithm first introduces a memory mechanism for individual components to feedback information to enhance its ca- pacity for population exploitation and to accelerate the convergence speed. Then, it adopts a new fitness function to increase the difference between individuals and to avoid premature convergence from failing to identify the best indi- vidual. Finally, the algorithm integrates an optimal nectar-source guidance mechanism into the knockout function to prevent the production of unexpected individuals. Experiments were conducted on standard functions and were compared with those with several typical improved ABCs. The results show that the improved algorithm accelerates the convergence rate and improves the solution accuracy.
出处
《智能系统学报》
CSCD
北大核心
2016年第2期172-179,共8页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(61573167)
高等学校博士学科点专项科研基金项目(20130093110011)
江苏省自然科学基金项目(BK20141114)
关键词
人工蜂群算法
群体智能
进化算法
函数优化
信息反馈
artificial bee colony algorithm
swarm intelligence
evolutionary algorithm
function optimization
in-formation feedback