摘要
为定量解决非支配解排序问题,并兼顾多目标粒子群优化算法(multi-objective particle swarm optimization,MOPSO)的收敛性和多样性,提出了一种基于Pareto云隶属度的MOPSO算法。利用Logistic混沌映射优化种群的初始空间分布并融合布谷鸟搜索(cuckoo search,CS)指导粒子跳出局部陷阱,以增强算法的全局寻优能力。首次提出云向量评价Pareto最优解集方法,采用云隶属度评价准则对粒子适应度值进行量化评价。依据云隶属度选取个体最优和群体最优,平衡全局开发与开采,进而实现外部档案维护。测试函数集ZDT的实验结果表明,改进算法在收敛性和多样性方面较MOPSO和NSGA-Ⅱ有一定优势。
An improved MOPSO algorithm based on Pareto cloud membership is proposed to cope the problem of Pareto-Sort,as well as convergence and variety of MOPSO.First,Logistic map is adopted to initialize the population combined with cuckoo search to enhance the ability on global searching.Meanwhile,cloud membership is used to measure global best and personal best solution in order to maintain the external archive.Finally,experiment results on ZDT test function set show that the proposed al-gorithm outperforms on both convergence and variety than MOPSO and NSGA-Ⅱ.
出处
《中国科技论文》
CAS
北大核心
2015年第14期1610-1613,共4页
China Sciencepaper
基金
国家自然科学基金资助项目(61309008)
陕西省自然科学基金资助项目(2014JQ8049)
关键词
云隶属度
多目标粒子群优化算法
布谷鸟搜索
混沌映射
云相似度
cloud membership
multi-objective particle swarm optimization
cuckoo search
chaos map
cloud similarity
