Hybrid Immune Algorithm with EDA for Multi-Objective Optimization

Abstract

PDF HTML阅读 XML下载 导出引用 引用提醒 基于免疫算法和EDA 的混合多目标优化算法 DOI: 10.3724/SP.J.1001.2013.04341 作者: 作者单位: 作者简介: 通讯作者: 中图分类号: 基金项目: 国家自然科学基金(61303119); 国家教育部博士点基金(20090203120016, 20100203120008); 博士后面上基金(20090461283, 20090451369, 200801426, 20080431228, 201104658); 陕西省自然科学基础研究计划(2011JQ8010, 2009JQ8015); 中央高校基本科研业务费专项资金(K5051203007, K5051203002) Hybrid Immune Algorithm with EDA for Multi-Objective Optimization Author: Affiliation: Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:在免疫多目标优化算法的基础上,引入了分布估计算法(EDA)对进化种群进行建模采样的思想,提出了一种求解复杂多目标优化问题的混合优化算法HIAEDA(hybrid immune algorithm with EDA for multi-objectiveoptimization).HIAEDA 的进化过程混合了两种后代产生策略:一种是基于交叉变异的克隆选择算子,用于在父代种群周围进行局部搜索的同时开辟新的搜索区域;另一种是基于EDA 的模型采样算子,用于学习多目标优化问题决策变量之间的相关性,提高算法求解复杂多目标优化问题的能力.在分析两种算子搜索行为的基础上,讨论了两者在功能上的互补性,并利用有限马尔可夫链的性质证明了HIAEDA 算法的收敛性.对测试函数和实际工程问题的仿真实验结果表明,HIAEDA 与NSGAII 算法和基于EDA 的进化多目标优化算法RM-MEDA 相比,在收敛性和多样性方面均表现出明显优势,尤其是对于决策变量之间存在非线性关联的复杂多目标优化问题,优势更为突出. Abstract:The estimation of distribution algorithm (EDA) is a new type of evolutionary computation approach which reproduces offspring individuals by modeling and sampling the probability distribution of the evolving population. In this paper, the idea of EDA is introduced into the immune multi-objective optimization algorithm to form a hybrid algorithm termed as HIAEDA (hybrid immune algorithm with EDA for multi-objective optimization). It is proposed for solving complex multi-objective optimization problems (MOPs). In HIAEDA, two types of reproducing strategies are combined. One is a recombination and mutation based immune clonal selection operator. It performs a local search around the parent population and develops new searching areas. The other is a EDA based modeling and sampling operator. It learns the variable linkages and promotes the algorithm's capability of solving complex problems. By analyzing the searching behavior of the two operators, the paper comes to the conclusion that their functions are complementary to each other. The convergence of HIAEDA is proved using the theory of the finite Markov chain. Experimental results on benchmarking and real problems show that HIAEDA outperforms the outstanding NSGAII and the EDA based RM-MEDA in terms of both convergence and diversity, especially when solving complex MOPs with nonlinear relationship between decision variables. 参考文献 相似文献 引证文献