项目名称: 演化计算原理及其动态多目标优化应用的几个关键问题研究
项目编号: No.61202313
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 汤可宗
作者单位: 景德镇陶瓷学院
项目金额: 26万元
中文摘要: 演化计算(EC)算法是受进化论启发而提出的一大类随机优化算法,该类算法将进化论"物竞天择,适者生存"的思想用于求解优化问题。目前,EC在静态优化领域中成功的应用为其应用于动态优化问题提供了一种新的途径。本项目将在申请者攻读博士学位期间的工作基础上,重点对EC应用于动态多目标优化的如下关键问题进行研究:1)进化过程中如何维护群体的多样性以防止算法早熟?并使得算法搜索到的近似Pareto最优解集在目标空间中具有良好的分布特性;2)研究一种重要的选择策略的构建方法,提高算法的搜索速度和收敛性;3)EC算法进化过程中,研究一种新的适应度评价方法;4)基于EC原理,研究一种重要的多目标图像分割模型,揭示在动态背景下,多目标优化与图像分割间的内在联系,从而提高真实环境中该模型在不同类型图像中的分割效果。本项目的研究不仅能对 EC的基本理论有所贡献,而且还能够进一步丰富EC在实际应用领域中的研究成果。
中文关键词: 粒子群优化;遗传算法;优化问题;图像分割;
英文摘要: Evolutionary Computation (EC) Algorithms are a kind of stochastic optimization algorithms following Darwin's principle of "Survival of the fittest".Evolutionary computation provides a new approach for dynamic optimization problmes because it has been well used in static optimizatin fields. In this project, studies are mainly focused on third aspects based on some achievements during working on doctoral degree. 1)Fistly,How to maintain the diverstiy of population in evolution ? and to improve the divesity for Pareto solution set in search space. 2)Secondly,on the base of analyzing all kinds of entropies, a novel testing method keeping diversity of the population is proposed. In additional, a novel selectin scheme is incoporated into the improved evolutonary algorithm to slove dynamic multiobjective optimzation problems.3)Thirdly, study on a new fitness evaluation method during evolution of EC algorithm; 4) On the base of evolutionary algorithm, we are going to study an important model in image segmentation based on multiobjective optimization.The objective of the study are to reveal the inherent connection between multiobjective optimization and image segmentation in dynamic background so as to improve the robustness of image segmentation for the model in real scene. Research on this project is not only contribu
英文关键词: particle swarm optimizaton;genetic algorithm;optimization problem;image segmentation;