查询结果:   刘红平,黎福海.面向多目标优化问题的自适应差分进化算法[J].计算机应用与软件,2015,32(12):249 - 252,269.
中文标题
面向多目标优化问题的自适应差分进化算法
发表栏目
算法
摘要点击数
898
英文标题
SELF-ADAPTIVE DIFFERENTIAL EVOLUTION ALGORITHM FORMULTI-OBJECTIVE OPTIMISATION
作 者
刘红平 黎福海 Liu Hongping Li Fuhai
作者单位
长沙师范学院电子信息工程系 湖南 长沙 410100 湖南大学电气与信息工程学院 湖南 长沙 410082    
英文单位
Electronics Information Engineering Department, Changsha Normal College, Changsha 410100, Hunan, China Electrical and Information Engineering College, Hunan University, Changsha 410082, Hunan, China    
关键词
多目标优化 差分进化 自适应 缩放因子 优胜累积量
Keywords
Multi-objective optimisation Differential evolution Self-adaptive Scaling factor Accumulative superiority
基金项目
国家自然科学基金项目(51107036);湖南省科技厅科技计划项目(2012FJ3013)
作者资料
刘红平,副教授,主研领域:自动化与人工智能。黎福海,教授。 。
文章摘要
针对多目标优化得到一个最优解集和解之间难以比较的问题,对单目标优化中的自适应策略进行了改进,提出一种面向多目标优化问题的自适应差分进化算法,在已有方法自适应改变交叉率的基础上,设定缩放因子有三种不同的分布模型,通过统计一定代数内个体的优劣来自适应选择合适的模型并生成相应取值,从而控制了搜索长度,防止新个体陷入在最优解集的部分区域。该算法还提出利用第三方解集和优胜累积量的概念来处理最优解之间的比较问题。通过5个标准优化问题的测试结果以及与其他几种算法的对比研究表明,所提出的改进算法性能更好,其在IGD指标上减小了0.0031~0.0669,在IH指标上最多减小了0.0821。
Abstract
It is hard to compare an optimal solution set with solutions both derived from multi-objective optimisation, for this problem, we proposed a self-adaptive differential evolution algorithm for multi-objective optimisation, and based on existing algorithm which adaptively changes the crossover rate, we set that the scaling factors have three different distribution models, by counting the advantage and disadvantage of individuals within certain algebra the algorithm adaptively selects proper model and generates corresponding assigned value, so that it controls the search size as well as prevents the new individual falling into partial area of the optimal solution set. This algorithm also puts forward to use the solution set of the third party and the concept of accumulative superiority to deal with the problem of comparison between optimal solutions. It is demonstrated through five standard optimisation problems and the contrast study with some other algorithms that the improved algorithm proposed has better performance, where the IGD index values reduces 0.0031 to 0.0669, and the IH index value reduces up to 0.0821.
下载PDF全文