一种混合自适应多目标Memetic算法  

A Hybrid Adaptive Multi-objective Memetic Algorithm

作　　者：郭秀萍[1] 杨根科[1] 吴智铭[1] 

机构地区：[1]上海交通大学自动化系,上海200240

出　　处：《控制与决策》2006年第11期1234-1238,共5页Control and Decision

基　　金：国家自然科学基金项目(60174009)

摘　　要：M em etic算法是求解多目标优化问题最有效的方法之一,融合了局部搜索和进化计算,具有较高的全局搜索能力.混合自适应多目标M em etic算法(HAM A)用基于模拟退火的加权法进行局部搜索,采用Pareto法实现交叉和变异,通过扰动增强算法的exp loration能力,且进化过程可根据改善率自适应调整,以提高搜索效率并改善算法的鲁棒性.算例测试说明HAM A能产生更接近Pareto前沿且多样性更好的近似集.Memetic algorithm is one of the most efficient methods for multi-objective optimization problems, incorporating local search into evolutionary computation and having high global search ability. Hybrid adaptive memetic algorithm （HAMA） uses a simulated annealing-based weighted-sum method to perform local search, uses Pareto-based approach to implement crossover and mutation, and employs perturbation to enhance the exploration capability of the algorithm. The evolution is made self-adjusted according to optimization ratio for better efficiency and robustness of the algorithm. A testing example shows that HAMA can generate near-Pareto optimal and wellextended approximation set.

关 键 词：混合 自适应 多目标优化 MEMETIC算法 多目标0/1背包问题 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]