遗传算法在桁架结构优化设计中的应用  

Application of genetic algorithms in optimums design of truss structures

作　　者：孙仁范 牟在根[2] 颜谋[2] 陈云周[2] 

机构地区：[1]中国建筑技术集团有限公司,北京100013 [2]北京科技大学土木工程系,北京100083

出　　处：《建筑结构学报》2004年第3期75-79,共5页Journal of Building Structures

基　　金：国家自然科学基金(50078004)

摘　　要：本文探讨了遗传算法在桁架结构优化设计中的应用,通过对经典遗传算法进行改进,建立了桁架结构的优化设计模型,其中采用精华模型保证了上一代种群中最优个体进入到当前种群中进化,而且采用非均匀变异算子,使得进化前期变异明显,后期趋于稳定,以利于收敛到最优解附近。另外针对遗传算法容易收敛到局部最优点的缺陷,提出了通过改变各变量相应初始界限值,进行两次遗传算法操作的方法进行运算,提高了工作效率,且结果更为精确。最后通过十杆平面桁架的数值例题来验证遗传算法应用的可行性和有效性。This paper investigated the application of Genetic Algorithms (GA's) in the optimum design of truss structures. With GA's characteristic of running side by side, the overall optimization and feasible operation, the optimum design model of truss structures was established. Elite models were used to assure that the best units of the previous generation had access to the evolution of current generation. Using of non-uniformity mutation brought the obvious mutation at the prophase and stable mutation in the anaphase, this benefited the convergence of units to the brest result. In addition, to avoid GA's limitation of liability to converging to the part-optimum, the limit value of each variable was changed respectively and the GA's calculation was performed two times, so the program could work more efficiently and came out with more precise results. Finally, a numerical example of the ten bar truss was presented to demonstrate reliability and feasibility of GA's application.

关 键 词：遗传算法 桁架结构 优化设计 惩罚函数 

分 类 号：TU323.4[建筑科学—结构工程]