离散变量结构形状优化设计的混合遗传算法  

An Hybrid Genetic Algorithm for Shape Optimization of Structure with Discrete Variables

作　　者：朱朝艳[1] 张延年[2] 郭鹏飞[1] 王学志[1] 

机构地区：[1]辽宁工业大学土木建筑学院,锦州121001 [2]沈阳建筑大学土木工程学院,沈阳110168

出　　处：《武汉理工大学学报（交通科学与工程版）》2011年第1期156-159,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：国家自然科学基金项目(批准号:50709013);辽宁省教育厅科学技术研究项目(批准号:2006B052)资助

摘　　要：通过在遗传算法中嵌入拟满应力算子来增强其局部寻优能力,并将拟满应力遗传算法应用于离散变量结构形状优化设计问题.将形状设计变量和截面设计变量统一编码表示在同一染色体中,既解决了两类变量间耦合上的困难,又避免了将两类变量分开考虑只能求得局部最优解的问题.对混合离散变量问题,对形状变量提出二级搜索策略,即首先在整个搜索区间进行粗粒度搜索,进化后期再将搜索范围缩小至最优解附近,重新编码进行局部细致搜索,从而解决了求解的精度和求解效率间的矛盾.The imitative full-stress generic algorithm can increase the local searching capability by inserting imitative full-stress operator into generic algorithm,and put the imitative full-stress generic algorithm to the shapes ptimization of structurewith discrete variables.This paper encodes the shape design variables and cross-section variables into one chromosome.This solves the problems of the two kinds of variables in the binary system and brings forward a two-step searching strategies,that is,coarse searching is carried out in the whole searching range firstly.Then the searching range is decreased to the adjacency of the optimum solution in the late stage of evolution.The re-encoding is done and the local fine searching is carried out,whereby the conflict between the precision and the efficiency of solution is solved.

关 键 词：离散变量 形状优化 拟满应力法 混合遗传算法 拟满应力遗传算法 

分 类 号：TU311.4[建筑科学—结构工程]