项目名称: 面向自治的协同进化算法
项目编号: No.61202112
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: 陈昊
作者单位: 南昌航空大学
项目金额: 23万元
中文摘要: 协同进化算法源于对自然界协同进化现象的模拟,具有处理各种复杂优化问题的潜力,但无法在进化过程中根据问题复杂性调整算法中的协同进化关系,对协同进化机制自治性、动态性、自组织性等特征的刻画与利用不足。本项目拟根据面向自治的计算思想对协同进化算法进行研究,通过进化个体自治行为与协同演化实现系统级的复杂涌现行为,建立能够根据优化问题自治调整算法结构与复杂性的面向自治的协同进化算法框架与理论体系。研究协同进化算法的空间结构,构建进化个体在空间结构下的协同进化环境模型;研究进化个体状态变量的表达方法,建立冗余表示的个体基因表达机制;研究进化个体在不同生物水平上的自治行为,建立网络空间结构下的协同演化规则;研究协同进化关系下种群动态控制机制;分析选择压力在协同进化中的作用,研究协同进化系统的结构稳定性,建立协同进化算法的效能评价准则。此研究将为进化算法在复杂优化问题上的应用提供新的技术手段。
中文关键词: 协同进化算法;面向自治计算;演化规则;选择压力;种群控制
英文摘要: Coevolutionary algorithm (CoEA) inspired from the coevolution phenomena in nature. CoEA has been proved to have the potential to deal with complex optimization problems. However the current framework and theory system of CoEA are insufficient, and the simulation and utilization of the characteristics such as autonomy, dynamic and self-organizing of real coevlution mechanism are also inadequate. Moreover, CoEA is not yet competent to change the coevolution relationship according to the complexity of optimization problem among the evolution process. This project plans to research coevolutionary algorithm based on the thought of autonomy-oriented computing, establish the framework of autonomy-oriented coevlutionay algorithm which can achieve system-level emergence behavior by autonomous behavior and collaborative evolution of individuals, adjust the structure and complexity of the algorithm autonomously according to the complexity of optimization problem. This project intends to bring the network spatial structure into CoEA, researches the setting of living environment, expression of the state variables, and autonomous behavior of coevolutionary individuals, analyzes the coevlutionary mechanism on molecular level, individual level, population level and species level, studies the dynamic control mechanism of popula
英文关键词: Coevolutionary algorithm;Autonomy-oriented computing;Evolution rule;Selection pressure;Population control