摘要
针对传统人工免疫算法中相似度、浓度以及抗体现有评价方式存在的缺陷,采用独特型网络动力学模型,通过改进亲和力计算方法,使之综合表达函数值和抗体相似程度的信息,以抗体的浓度作为适应值,提出了一种基于独特型网络动力学模型的人工免疫算法.仿真结果表明,这种算法对多模态函数优化是有效的,其搜索效率及收敛速度均优于常见的人工免疫网络算法Opt-aiNet.
To overcome the demerits in evaluating the similarity, the concentration and the antibody of the conventional artificial immune algorithm, this paper proposes an improved artificial immune algorithm based on the idiotypic-network dynamic model by defining the antibody concentration as the fitness. In the proposed algorithm, the information about the function value and the similarity of antibody can be comprehensively extracted by modifying the calculation method of affinity. Simulated results show that the improved algorithm is effective on the optimization of multi-mode function and is of better searching efficiency and higher convergence speed than the conventional OptaiNet algorithm.
出处
《华南理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第1期62-65,72,共5页
Journal of South China University of Technology(Natural Science Edition)
基金
广州市科技攻关引导项目(2003Z3-D0091)
关键词
人工免疫算法
动力学模型
独特型网络
优化
artificial immune algorithm
dynamic model
idiotypic network
optimization
