【作者】 董列奎；

【导师】 李正良；

【作者基本信息】 重庆大学 ， 固体力学， 2003， 硕士

【摘要】 遗传算法是近年来在计算机科学领域和优化领域中受到广泛关注的一种拟生物进化理论的仿生学算法。在广泛阅读文献和调研的基础上，本文对遗传算法及其在结构优化设计应用中的相关内容进行了分析和综合，对结构优化设计的基本概念、主要特点进行了介绍，对遗传算法应用于结构优化设计时的数学建模、约束条件处理、初始种群的产生及其遗传算法控制参数的选择几个关键因素作了概要的分析与研究。本文针对结构优化设计的特点，结合有限元结构分析原理，编制了基于简单遗传算法的结构优化设计程序，用于解决离散变量的框架结构优化设计。经过实例验证，该程序用于结构优化设计是可行的和高效的。由于简单遗传算法仅擅长全局搜索，而局部搜索能力不足，要达到真正的最优解则要花费相当长的时间。针对简单遗传算法的缺点，本论文提出了对简单遗传算法进行改进—自适应遗传算法，并且编制了基于自适应遗传算法的结构优化程序，算例分析表明得出自适应遗传算法确实提高了遗传算法在应用于结构优化设计方面收敛性和计算速度。在应用遗传算法时，初始种群的产生一般是随机的，它往往需要的编码长度很大，导致需要很大的种群规模或者很长的进化过程才能有较好的优化效果，这将耗费大量的计算时间和费用。如何确定恰当的种群规模是一个直接影响遗传算法效率的问题。由于没有这方面相关的文献资料，本文还针对群体规模对基本遗传算法的优化计算影响的问题结合三个算例进行了尝试性的对比计算，确定了较佳的群体规模。更多还原

【Abstract】 Genetic Algorithm (GA) is one of the main directions in today’s compute science. Its application in structural catches more and more attention. Based on a great many literature, this paper analyzes and summarizes the genetic algorithm for the design of structural optimization. The basic conceptions and primary characteristics are introduced in this paper. This paper discuss some primary factor when genetic algorithm applied on optimal design, for example the mathematical model、the constrained problem、the initial population and the controls parameter of genetic algorithmAimed at the characteristic of structural optimization and finite-element analysis, this paper accomplishes the optimal program based on the simple genetic algorithm. This program is present for optimizing structural systems with discrete design variables. The result shows that the program is feasible and effective.Because simple genetic algorithm is good at the global search and short of the local search, searching the best optimized solution need to consume more time. Aimed at the shortage of GA, this paper improves the simple genetic algorithm and accomplishes the optimal program based on the adaptive genetic algorithm. The result shows that the adaptive genetic algorithm improves the convergence rate.Generally, the genetic algorithm randomly generates the initial population. The coding of the variables that describe the problem is always large. In order to find the best optimal result, it needs the larger population and the longer course of optimization. That will spend much time and money, so the appropriate population size is a factor that affects the efficiency of genetic algorithm. Because no literature discusses that how to confirm the appropriate population size, this paper discusses how the population size affects GA’s optimal course based on three examples. This paper confirms the better population size.更多还原