基于Logistic模型的改进人工蜂群算法  

Logistic model-based improved artificial bee colony algorithm

作　　者：魏焕新[1] 胡招娣[1] 唐明珠[2] 

机构地区：[1]湖南机电职业技术学院信息学院,湖南长沙550004 [2]长沙理工大学能源与动力工程学院,湖南长沙410114

出　　处：《兰州理工大学学报》2017年第2期104-109,共6页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(61403046);湖南省科学技术基金(S2014F1023)

摘　　要：针对基本人工蜂群(ABC)算法存在着收敛速度慢、易陷入局部最优、求解精度低等缺点,提出一种基于Logistic模型的自适应人工蜂群(A-ABC)算法.首先,利用反向学习策略初始化种群个体以保证群体的多样性,为算法进行全局搜索奠定基础;其次,通过引入参数λ对ABC算法的搜索方程进行改进以产生新的候选个体,在进化过程中,λ的大小基于Logistic模型自适应调节,以协调算法的探索能力和开发能力;引入基于排序的选择概率以避免算法出现早熟收敛.对几个典型的Benchmark函数进行了测试,实验结果表明,与基本ABC算法相比,A-ABC算法具有更高的求解精度和更快的收敛速度.Aimed at the disadvantage in basic artificial bee colony （ABC） algorithm, such as slow to con- verging, easily falling into local optima, and low precision of solution, an adaptive artificial bee colony （A- ABC） algorithm was proposed based on Logistic model. First, the opposition-based learning strategy was used to initiate individuals in population, guarantying the diversity of the population and laying a basis for global searching with the algorithm. Then, by means of introducing a parameter 2, the searching equation in ABC algorithm was improved to generate new candidate individuals. The parameter 2 would be adap- tively adjusted based on the l.ogistic model during evolution process to coordinate the exploration and ex- ploitation ability of the algorithm. The selection probability was introduced based on ordering to avoid pre- mature convergence of searching. The experimental results of several typical benchmark functions showed that the A-ABC algorithm would have a higher solution precision and faster convergence speed compared to the basic ABC algorithm.

关 键 词：人工蜂群算法 LOGISTIC模型 自适应 反向学习 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]