基于克隆选择原理的自适应免疫性遗传算法研究  

Research on adaptive immune genetic algorithms based on clone selection principles

作　　者：赵峰[1] 苏宏升[1] 

机构地区：[1]兰州交通大学自动化与电气工程学院,甘肃兰州730070

出　　处：《计算机工程与设计》2008年第23期6060-6063,共4页Computer Engineering and Design

基　　金：甘肃省自然科学基金项目(3ZS061-A25-045)。

摘　　要：针对遗传算法中存在着收敛方向无法控制和没有记忆能力等缺陷,提出了具有免疫功能的克隆遗传算法。该算法把目标函数和制约条件作为抗原,保证所生成的抗体与问题直接相关联,使收敛方向得以控制;对抗原亲和力高的抗体进行克隆记忆,促使优良个体的发育成熟并能有效地遗传到下一代;同时,基于浓度的概念提出对抗体数量进行抑制,确保群体更新的多样性,避免早熟。通过理论分析和实验研究,证明该算法具有快的收敛速度和搜索能力,是一种有效的生物计算方法。In view of the defects that there is no way to control convergence orientation and also no memory capability in genetic algorithms （GA）, a clone genetic algorithm is proposed which can make genetic algorithm possess immune function. In this algorithm, aim function and restriction condition are served as antigens （Ags） so as to ensure the created antibodies （Abs） directly related to the problem-solving, thus, convergence orientation in GA is controlled ultimately, the antibodies with high affinities are cloned and recollected to promote the growth of the excellent individual and can effectively pass it down to next generation, meanwhile, the quantities of the antibodies are also restricted to ensure the diversities of the updated colonies and avoid pre-maturity. Theory analysis and example study indicate that the method owns fast convergence speeds and search capabilities, and is an effective biology algorithm method.

关 键 词：克隆选择 遗传算法 亲和力 浓度 生物计算 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]