基于小生境遗传算法的单层网壳结构优化设计研究  

Study of optimum design of single layer dome structures based on niche genetic algorithm

作　　者：牟在根[1] 梁杰[1] 隋军[2] 宁平华[2] 颜谋[1] 

机构地区：[1]北京科技大学土木工程系,北京100083 [2]广州市市政工程设计研究院,广东广州510060

出　　处：《建筑结构学报》2006年第2期115-119,共5页Journal of Building Structures

基　　金：国家自然科学基金资助项目(50078004)

摘　　要：本文提出了一种改进的遗传算法,将小生境技术与模糊控制思想引入其中,从而达到保护种群的多样性的目的,同时使每代最优解得以保存。简单遗传算法本身存在着过早成熟收敛,后期收敛速度放慢等缺陷,其在加入小生境技术后虽可保持种群群体的多样性,使其优化过程不致过早收敛,但是由于小生境技术本身的特点,又不可避免地会产生部分个体的早熟以及陷入局部最优,本文于是在小生境遗传算法基础上加入模糊控制思想,对种群的杂交概率Pc和变异概率Pm进行模糊控制,从而形成了一种新型的模糊控制小生境遗传算法。最后通过单层网壳结构的数值实例证明了该方法的有效性和可行性,进一步为空间网格结构的优化设计奠定了基础。This paper first illustrates an improved genetic algorithm（GA）. The theory of niche technique and fuzzy control are introduced into the CA, for the purpose of enhancing the population diversity and maintaining the best parts of each generation. The simple GA itself has the problem of converging prematurely, and then reducing convergence speed in the anaphase. In order to avoid premature convergence and occurrence of minimal deceptive problems, which is caused by the niche technique, fuzzy control is presented for the controlling of the crossover probability, Pc and mutation probability, Pm. The new type of algorithm is fuzzy controlled niche genetic algorithm （FNGA）. Finally, a numerical example of single layer dome is presented to demonstrate reliability and feasibility of the new algorithm. It further establishes the base of the optimum design of space frame structures.

关 键 词：遗传算法 小生境技术 模糊控制 单层网壳 优化设计 

分 类 号：TU311.4[建筑科学—结构工程] TU356