混合变异克隆选择多目标优化算法

摘要

In this paper, a hybrid mutation clonal selection multiobjective optimization algorithm was presented. In the algorithm, evolution was performed in three antibody groups, every group used different mutation operator, through the updated external memory antibody population to keep the Pareto optimal solution. The large-scale mutation operator could be utilized to quickly localize the global Pareto optimal space, the small-scale mutation operator could implement local accurate Pareto optimal solution search, and the uniform mutation operator could make the algorithm have the ability to escape from local optimal solution, which ensured the diversity of the algorithm. We compared the proposed method with NSGA-II, ε-MOEA and the multiobjective clonal selection algorithm ( MCSA) based on the signal group in solving five DTLZ problems, and the experimental results show that the new algorithm is effective.%研究多目标优化问题,针对提高算法的快速性,提出一种混合变异克隆选择多目标优化算法.进化在三个抗体群中进行,不同的抗体群采用不同的变异算子,并通过外部记忆抗体群的更新,来保留进化的最优抗体,避免算法进化后期出现退化现象.算法采用的三种变异算子:高频大尺度高斯变异算子带有振荡性质,能够对Pareto最优解区域进行勘探,单基因小尺度衰减的高斯变异算子能够使优化结果逼近Pareto最优解,均匀变异算子使算法具有局部逃逸能力,能够保证解的多样性.将算法和经典的NSGA -Ⅱ、ε- MOEA算法以及单一变异的多目标克隆选择算法(MCSA)进行性能比较,结果证明新算法具有较好的快速搜索性能和鲁棒性.