基于距离收敛量和历史信息密度的多目标进化算法的研究

发布时间：2018-03-21 03:17

本文选题：多目标进化算法　切入点：距离收敛量　出处：《东北师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来,多目标优化问题已经成为人工智能领域中最重要的问题形式之一。多目标进化算法凭借其优秀的求解能力,已经成为求解多目标优化问题最重要的方法之一。多目标进化算法是一种基于群体演进的人工智能算法,该算法以生物进化过程中的优胜劣汰原理为基础,以此推演出一套随机的全局搜索策略。而在筛选个体的过程中,多目标进化算法大都利用非支配信息和密度信息评价个体。针对上述问题,本文提出了一种基于距离收敛量和历史信息密度的多目标进化算法,并同4个最新的多目标进化算法在2个基准问题集上进行了对比实验。本文的主要工作可以总结如下:(1)为了可以在个体支配等级相同的情况下有效的区分个体的收敛程度,本文提出的距离收敛量的概念,并在此基础上提出了一种新的配对选择操作;(2)为了可以在计算个体密度的过程中考虑整个搜索过程,本文提出的历史信息密度,并在此基础上提出了新的个体选择操作;(3)通过对比实验结果可以说明基于距离收敛量和历史信息密度的多目标进化算法比相对比的其他算法有更好的求解效果。
[Abstract]:In recent years, multi-objective optimization problem has become one of the most important problems in the field of artificial intelligence. Multiobjective evolutionary algorithm is an artificial intelligence algorithm based on population evolution, which is based on the principle of survival of the fittest in the process of biological evolution. A set of random global search strategies is derived. However, in the process of selecting individuals, the multi-objective evolutionary algorithms mostly evaluate individuals using non-dominated information and density information. In this paper, a multi-objective evolutionary algorithm based on distance convergence and historical information density is proposed. And compared with four new multi-objective evolutionary algorithms on two benchmark problem sets, the main work of this paper can be summarized as follows: (1) in order to effectively distinguish the degree of convergence of individuals under the same dominance level, In this paper, the concept of distance convergence is proposed, and a new pairing selection operation is proposed. In order to consider the whole searching process in the process of calculating individual density, the historical information density is proposed in this paper. On this basis, a new individual selection operation is proposed. By comparing the experimental results, it can be shown that the multi-objective evolutionary algorithm based on distance convergence and historical information density is more effective than other algorithms.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】