一种基于免疫的抗体优化策略及应用  

An policy of antibody optimization based on immune and its application

作　　者：王刚[1,2] 周激流[2] 康钦建[3] 张卫华[2] 何新国[2] 雷刚[2] 

机构地区：[1]四川警察学院,四川省公安厅警官培训基地,泸州646000 [2]四川大学计算机学院,成都610065 [3]四川大学电子信息学院,成都610065

出　　处：《激光杂志》2008年第3期84-86,共3页Laser Journal

基　　金：教育部博士点基金(20020610013)

摘　　要：受自然免疫系统相关机理的启发,本文提出了一种引入进化梯度的免疫遗传算法(EG-IGA)并应用于抗体的优化。该算法基本思想是将生成的多样性抗体,分成若干个小生境,并排挤掉同一个小生境中的较差抗体,保留优良抗体;然后执行GA交叉、变异等操作:再利用进化梯度为优良抗体标记进化方向,按标记的进化方向继续局部寻优,以较快的速度寻找最优的抗体种群。最后将EG-IGA算法和NGA算法分别运用在同一个多峰值函数上,实验结果表明,本文所提出的EG-IGA算法具有提高解的精度及收敛速度、找到更多最优解等特点。Inspired by the relevant mechanism of biology immune system, we presented Evolution Grads Included Immune Genetic Algorithm（EG-IGA） which idea was that many different antibodies divided into certain Niche , threw away some bad antibodies, and preserved better or best antibodies according to their affinity; and then implement the operation of crossover and mutation by using GA; and finally , evolutionary grads was used to improve the ability of finding the local best antibody and the precision was improved, It was conductive to generate antibodies which had much more affinity by the use of the niche algorithm based on crowding mechanism. In conclusion, experiment result showed that the EG-IGA had more superior than NGA in precision and convergence rate.

关 键 词：免疫原理 进化梯度 抗体优化 函数优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]