摘要
基于免疫应答原理及小生境概念,采用实数编码策略,提出解决多模态函数优化的免疫算法。构建此算法的目的在于将其与遗传算法比较,分析二者的差异。算法设计的关键在于抗体评价规则及亲和突变算子,以及引入小生境技术、抗体浓度概念及免疫系统中群体多样性的机理,增强群体多样性。此算法具有自适应地调整进化群体规模、并行搜索最优解及强稳定性等特点,特别能搜索多个最优解(若存在)及大量局部最优解;同时其收敛性获证。事例仿真比较获该文算法的有效性,此暗示免疫算法的研究具有广阔前景。
An immune algorithm, applied to multi-modal function optimization, is proposed based on real decoding and niche and immune response principle of the immune system to be compared with GA. Its key is to design evaluating rule for antibodies and affinition mutation operator, and introduce niching technology to strength population diversity. It has such properties as determining population size automatically, parallel search optimum, strong robustness, and so forth. Besides, its convergence is proved. Simulation shows that the algorithm is better than the algorithm REGA, which hints that immune algorithms are a potential research area.
出处
《重庆大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第9期130-133,144,共5页
Journal of Chongqing University
关键词
免疫算法
免疫应答
小生境
全局收敛
immune algorithm
immune response
niche
global convergence
