基于半径参数周期性缓慢变化的双种群遗传算法  

Dual-population genetic algorithm based on periodic slow change in radius parameter

作　　者：刘伟[1] 涂井先[1] 

机构地区：[1]广东工业大学应用数学学院,广州510520

出　　处：《计算机应用研究》2012年第1期43-46,51,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60974077)

摘　　要：双种群遗传算法引进了主种群和辅助种群,通过控制半径参数的变化来控制辅助种群的变化从而实现种群多样性。但其存在半径参数变化过快导致辅助种群得不到控制的缺陷。针对该缺点,用余弦函数来刻画半径参数的变化,提出了基于半径参数周期性缓慢变化的双种群遗传算法。半径参数的这种变化规律既有利于维持种群多样性,又能增强局部搜索能力。为了估算半径参数的最大取值,给出了把个体与主种群中心的最大距离作为半径参数最大取值的方法。仿真实验表明,新算法优于当前一些较好的遗传算法。Dual-population genetic algorithm introduced main population and reserve pop-ulation,it dominated the reserve population through controlling the change of radius parameter which maintained the diversity of group.However,the radius parameter changed too fast to control the reserve population.For the sake of overcoming this shortcoming,described the change of radius parameter by means of cosin function and proposed a dual-population genetic algorithm based on periodic slow change in radius parameter.The regularity for change of radius parameter was not only advantageous to keep the diversity of group,but also enhanced the local search.In order to estimate the maximum value of radius parameter,took the distance between individual and the centre of main population view as the maximum value of radius parameter.The results obtained show that the improved genetic algorithm is more effective than some current optimization algorithms.

关 键 词：遗传算法 双种群 主种群 辅助种群 交叉繁殖 半径参数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP301.6[自动化与计算机技术—控制科学与工程]