一种基于自适应选择策略的改进型MOEA/D算法

摘要

针对MOEA/D单纯使用邻域更新作为选择策略而造成的个体解的重复更新、缺乏全局适配性等问题,提出了一种兼及全局替换和局部更新策略的新算法,即基于自适应选择策略的改进型MOEA/D(MOEA/D-AS).算法首先设计了一种新的基于最佳二分图匹配的选择策略(KMS),利用子问题和个体解的匹配关系,从全局角度实现精英个体集的最优选择;然后利用种群的进化信息构造一种匹配紊乱判断机制;最后利用紊乱判断机制,在综合分析邻域更新策略和KMS各自优势的基础上,使算法自适应地选择最合适的选择策略,以提高鲁棒性和优化效率.选取LZ09,DTLZ,CEC09等作为标准测试函数,将改进后的算法MOEA/D-AS与经典MOEA/D系列算法进行对比实验,并以Spread和IGD为性能评估指标.实验结果表明新算法具有更好的收敛性和分布性,验证了自适应选择策略能够有效地指导精英解的选择过程.%Using simple and efficient neighbouring updating as the main selection strategy,MOEA/D focuses more on local replacement.In this case,it ignores the individual's global fitness and makes algorithm get into local minimum easily.In order to overcome the flaw of MOEA/D in local replacement,an improved MOEA/D which takes the global replacement and local updating into account based on self-adaptive selection strategy(MOEA/D-AS) was proposed.Firstly,a new selection strategy based on perfect matching of bipartite graph(KMS) is applied to select the elite solutions from a global perspective by using matching relationship between sub-problems and indivdual soltuions.Then,using the evolutionary information of the population,a matching disorder judgment mechanism was proposed.At last,on the basis of comprehensive analysis of the advantages of neighbourhood update strategy and KMS,the new algorithm can adaptively choose the most suitable selection strategy by using the disorder judgment mechanism to improve robustness and efficiency.The proposed algorithm was compared with some state-of-the-art versions of MOEA/D on LZ09,DTLZ and CEC09 Benchmarks.The values of Spread and IGD show that MOEA/D-AS has certain advantages than other algorithms in terms of convergence and distribution,and suggest that the self-adaptive selection strategy can effectively guide the selection process of elite solution.