自适应并行混沌免疫进化规划

Adaptive parallel chaos immune evolutionary programming

基于克隆选择原理，提出一种新的并行混沌免疫进化规划算法．在算法中，根据抗体抗原亲和度将抗体种群分为两个子群，相应的提出混沌克隆算子和超变异算子，混沌克隆算子在局部空间具有较强搜索能力，超变异算子在广阔空间具有大范围搜索能力，通过两个算子的并行操作使局部寻优和多样性保持相结合，从而提高算法的搜索效率．仿真表明，与传统进化规划（EP）和基于混沌变异的进化算法（EACM）相比较，并行免疫进化规划搜索效率高，能有效抑制早熟收敛现象，可用于解决复杂的机器学习问题．

Based on clonal selection theory, a novel adaptive parallel chaos immune evolutionary programming （PCIEP） is proposed. On the basis of antigen-antibody affinity, the original antibody population is divided into two subgroups. Correspondingly, two immune operators, chaotic clonal operator （CCO） and super mutation operator （SMO） are proposed. The former has strong local search ability in local space while the latter has better search ability in broad space. Thus, the combination of searching local optimum with maintaining population diversity can be actualized by concurrently operating CCO and SMO, which can enhance searching efficiency of the algorithm. Compared with the conventional evolutionary programming （EP） and evolutionary algorithm with chaotic mutation （EACM）, experimental results show that convergence. Therefore, it can be employed to solve complicated PCIEP is of high efficiency and can effectively prevent premature optimization problems.