基于遗传算法的行为结构分组创新  

INNOVATION OF BEHAVIOR STRUCTURE GROUPING BASED ON GENETIC ALGORITHM

作　　者：金熙哲[1] 郭为忠[1] 王玉新[2] 邹慧君[1] 

机构地区：[1]上海交通大学机械工程与动力学院 [2]同济大学机械工程学院,上海200092

出　　处：《机械工程学报》2007年第5期24-28,共5页Journal of Mechanical Engineering

基　　金：国家自然科学基金(50405017;50275092);上海市教委青年基金(01QN10)资助项目

摘　　要：针对执行系统概念设计过程前期阶段的计算机辅助实现是一个难题的现状,将遗传算法引入执行系统概念设计的行为结构分组创新。将行为集合分解为相互矛盾的行为和相容行为,相互矛盾的行为数决定基因串各段规模,相容行为数决定基因串段数,构成行为结构的基因串编码方法。基于行为分组评价规则,定量化形成目标函数并构造适应度函数,通过设定的基因操作,实现行为分组的计算机优化,达到行为结构的变型创新。算例分析表明所提出方法的有效性,有助于实现执行系统的计算机辅助概念设计。A GA-oriented computerized method is presented to group behaviors automatically during the innovation of behavior structures for computer-aided conceptual design of executive systems. The set of behaviors is separated into contradictory behaviors, the size of which determines the size of each section of a gene, and compatible ones, the size of which determines the number of sections of the gene. By quantifying the given evaluation criteria of behavior grouping, the fitness function is constructed. The behaviors are optimized to be grouped using the genetic algorithm with modified gene operations which leads to an optimal behavior structure. Examples are given to illustrate the presented method.

关 键 词：执行系统 计算机辅助概念设计 行为结构 行为分组 遗传算法 

分 类 号：TH122[机械工程—机械设计及理论]