摘要
结合免疫系统的机理和进化规划原理,对免疫进化规划进行改进。即引入多样性函数和群体局部退化相结合的方法,对克隆细胞进行选择和更新,增强群体信息的多样性,克服近亲细胞过度繁殖而引起早熟收敛;利用双曲正切函数,无须区分亲和度界限,决定个体细胞的变异率,实现细胞群的自适应变异;选择亲和度高的一半细胞作为记忆细胞,利用其替换原始细胞群亲和度低的细胞。对各部分改进的原因和优点进行了分析,给出了算法的主要步骤,并对自适应免疫进化规划的收敛性进行了说明。最后用不同的测试函数进行仿真实验,结果表明了方法的有效性。
Four operators, such as clone, hyper-mutation, selection and memory of adaptive immune evolutionary programming, were improved based on combining immune system＇s mechanism and principle of evolutionary programming. Diversity function and local degeneration algorithm were used to select and renew clone cells, and the characters of diversity for cells population were increased. Premature integration was avoided Because of over inbreeding in cells population in conventional methods, Adaptive mutation of cells population was realized by using hyperbolic tangent function to determine mutation probability without considering limitation of affinity. Memory cells were composed by half population with high affinity, and the half populations with low affinity in original generation were replaced. The analysis of reasons and merits for improved part were proposed. The main steps of the algorithm were given and the convergence of adaptive immune evolutionary programming were formulated. In simulation experiments, different functions were optimized by the presented algorithm, and the results show that the algorithm is valid.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第5期1147-1150,共4页
Journal of System Simulation
基金
教育部优秀青年教师计划资助项目(2001年度)
安徽省自然科学基金资助项目(2006年度)
关键词
免疫系统
进化规划
优化
局部退化
immune system
evolutionary programming
optimization
local degeneration
