克隆选择单变量边缘分布算法

张庆彬，吴惕华，刘波针对单变量边缘分布算法（UMDA）求解复杂优化问题的局限性，将人工免疫系统引入分布估计算法（EDAs）领域，提出了一种基于克隆选择原理的单变量边缘分布算法.该算法在进化过程中的每一代执行若干次克隆选择算法（CLONALG），利用克隆选择过程中的高频变异操作提高混合算法的局部搜索能力.通过对2种不同旅行商问题（TSP）的仿真实验表明，与UMDA、CLONALG以及UMDA和2 opt局部搜索算法的混合算法(UMDA2 opt)相比，克隆选择单变量边缘分布算法具有更高的优化性能.

Hybrid univariate marginal distribution algorithm based on clonal selection principle

Artificial immune system(AIS) was introduced to the estimation of distribution algorithms(EDAs) and a hybrid UMDA based on the clonal selection principle was proposed in order to improve the performance of univariate marginal distribution algorithm(UMDA) to solve difficult optimal problems.In each generation during the evolution,the clonal selection algorithm(CLONALG) was implemented.The local search capability was improved through hypermutation process.Simulation results on two different traveling salesman problems(TSP) show that the proposed algorithm has higher optimal performance than that of the UMDA and the CLONALG as well as the UMDA combining with 2-opt local search(UMDA2-opt),respectively.