遗传算法的改进及其在预应力空间网格结构中的应用  

Improvement and application of genetic algorithm in prestressed space latticework

作　　者：张春玉 王振清[1] 梁文彦[1] 

机构地区：[1]哈尔滨工程大学航天与建筑工程学院,黑龙江哈尔滨150001 [2]黑龙江科技学院建筑工程学院,黑龙江哈尔滨150027

出　　处：《哈尔滨工程大学学报》2010年第10期1317-1322,共6页Journal of Harbin Engineering University

基　　金：教育部博士点基金资助项目(20060217010);哈尔滨工程大学基础研究基金资助项目(HEUFT07005)

摘　　要：针对预应力空间网格结构设计变量的特点,提出了一种采用离散变量、基于遗传算法的优化设计方法.该方法将杆件截面和预应力置于同一水平变量空间,个体编码中同时体现两种变量的遗传基因,避免了割裂设计空间、分级优化而带来的误差影响.优化方法采取了适应度变换、最优保留及自适应遗传等改进措施,以改进标准遗传算法易于发生早熟现象、局部寻优能力差等缺点.Directed at the feature of structural design variable of prestressed space latticework,the paper puts forward a optimum design method based on discrete variable and genetic algorithm.The method puts cross section of member bar and prestress in a same level variable space,it′s individual codes externalize genetic gene of two variables at the meantime,keep out of error effect created by rending design space and cascade optimum.The optimum method adopts improved means of sufficiency transformation,best retention,self adapting inheritance and so on,to improve the defects of premature phenomenon and bad ability in finding the best.

关 键 词：遗传算法 网格结构 优化设计 

分 类 号：TU394[建筑科学—结构工程]