基于遗传算法的自由曲面网壳结构形态创构方法研究
发布时间：2018-11-24 12:59
【摘要】：自由曲面结构作为一种新颖又美观的结构形式逐渐地成为了结构形态学研究的主要对象。自由曲面网壳结构作为当代空间结构发展的一种新趋势,对其进行更深入的创构研究是不可或缺的。本文基于遗传算法对自由曲面网壳结构进行了形态创构方法的研究。阐述了遗传算法的基本原理和一些遗传算法的高级实现方法,建立了基于遗传算法的自由曲面网壳结构的形态创构思路:首先,根据Loop网格细分法建立合理的自由曲面并通过完善生成自由曲面网壳结构的基结构,然后,采用遗传算法对基结构进行拓扑和尺寸优化,对得到的优化结果采用类模拟退火算法进行改进和完善。最后,采用敏感度解析法对最终优化结构进行节点移动,完成结构的形状优化。拓扑优化时,对稳定的三角单元进行编码,采用二进制0-1编码的方式表示存在或者不存在;尺寸优化时,建立包含不同截面尺寸的截面表,结构中的每根杆件的截面编码值对应于截面表中相应的截面编号,采用整数编码的方式;形状优化时,采用敏感度解析法对节点进行移动,可以只移动z坐标,也可以同时移动x、y和z坐标。本文基于遗传算法并结合多种优化方法对自由曲面网壳结构进行优化,既弥补了遗传算法局部搜索能力较差的弱点,又提高了求解效率。优化过程中提出了解码之后的结构中存在问题的处理方法和杆件自重荷载与均布荷载的等效转换措施。包括:必须存在的支座节点和加载点都存在的处理方法;结构中没有独立杆件团的处理方法,确保不会生成刚度矩阵奇异的机构。结构受均布荷载作用时,将结构的平面投影分成许多细小的正方形,均布荷载乘以小正方形的面积转换成集中荷载作用于小正方形的几何形心上,最后分配到距离其最近的节点上。当结构不满足材料强度、杆件长细比、构件稳定性和节点最大位移等约束条件时,采用罚函数法进行处理。在优化目标选取上,本文选取的两个优化目标分别是结构的质量最小和结构的质量与外力功乘积最小,通过算例分析和比较说明各自的优劣。并通过多个算例得到了受力性能良好而且新颖美观的结构,说明本文提出的形态创构方法的可行性和实用性。
[Abstract]:As a novel and beautiful structural form, free-form surface structure has gradually become the main object of structural morphology research. As a new trend of the development of the contemporary space structure, it is indispensable to study the free surface latticed shell structure more deeply. In this paper, the morphological method of free-form surface reticulated shell is studied based on genetic algorithm. The basic principles of genetic algorithm and some advanced implementation methods of genetic algorithm are described. The shape creation of free-form surface reticulated shell structure based on genetic algorithm is established. According to the Loop mesh subdivision method, a reasonable free form surface is established and the base structure of the free surface latticed shell structure is generated by perfecting it. Then, the topology and size of the base structure are optimized by genetic algorithm. The optimization results are improved and improved by using simulated annealing algorithm. Finally, the sensitivity analysis method is used to realize the shape optimization of the structure by moving the nodes to the final optimization structure. In topology optimization, the stable triangular units are encoded and binary 0-1 coding is used to indicate the existence or non-existence. When the size is optimized, a cross-section table containing different cross-section sizes is established. The cross-section coding value of each member in the structure corresponds to the corresponding section number in the cross-section table, and the method of integer coding is adopted. When the shape is optimized, the sensitivity analysis method is used to move the node, which can only move z coordinate, but also move xy and z coordinate simultaneously. In this paper, based on genetic algorithm and combined with various optimization methods, the free-form surface reticulated shell structure is optimized, which not only makes up for the weakness of the local search ability of genetic algorithm, but also improves the efficiency of solving the problem. In the process of optimization, the methods to deal with the problems existing in the decoded structure and the equivalent conversion measures between the deadweight load and the uniform load of the members are proposed. It includes: the processing method of supporting node and loading point that must exist, and the processing method of independent member cluster in the structure to ensure that the mechanism with singular stiffness matrix will not be generated. When the structure is subjected to uniform load, the plane projection of the structure is divided into many small squares. The uniform load multiplied by the area of the small square is converted into a concentrated load acting on the geometric center of the small square. Finally, it is assigned to the nearest node. The penalty function method is used when the structure does not satisfy the constraint conditions such as material strength, the aspect ratio of the members, the stability of the members and the maximum displacement of the joints. In the selection of optimization objectives, the two optimization objectives selected in this paper are the minimum quality of the structure and the minimum product of the mass of the structure and the product of the external force. The advantages and disadvantages of the two optimization objectives are illustrated through the analysis and comparison of the numerical examples. A novel and beautiful structure with good mechanical performance is obtained through several examples, which shows the feasibility and practicability of the proposed method.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU399
本文编号：2353805
[Abstract]:As a novel and beautiful structural form, free-form surface structure has gradually become the main object of structural morphology research. As a new trend of the development of the contemporary space structure, it is indispensable to study the free surface latticed shell structure more deeply. In this paper, the morphological method of free-form surface reticulated shell is studied based on genetic algorithm. The basic principles of genetic algorithm and some advanced implementation methods of genetic algorithm are described. The shape creation of free-form surface reticulated shell structure based on genetic algorithm is established. According to the Loop mesh subdivision method, a reasonable free form surface is established and the base structure of the free surface latticed shell structure is generated by perfecting it. Then, the topology and size of the base structure are optimized by genetic algorithm. The optimization results are improved and improved by using simulated annealing algorithm. Finally, the sensitivity analysis method is used to realize the shape optimization of the structure by moving the nodes to the final optimization structure. In topology optimization, the stable triangular units are encoded and binary 0-1 coding is used to indicate the existence or non-existence. When the size is optimized, a cross-section table containing different cross-section sizes is established. The cross-section coding value of each member in the structure corresponds to the corresponding section number in the cross-section table, and the method of integer coding is adopted. When the shape is optimized, the sensitivity analysis method is used to move the node, which can only move z coordinate, but also move xy and z coordinate simultaneously. In this paper, based on genetic algorithm and combined with various optimization methods, the free-form surface reticulated shell structure is optimized, which not only makes up for the weakness of the local search ability of genetic algorithm, but also improves the efficiency of solving the problem. In the process of optimization, the methods to deal with the problems existing in the decoded structure and the equivalent conversion measures between the deadweight load and the uniform load of the members are proposed. It includes: the processing method of supporting node and loading point that must exist, and the processing method of independent member cluster in the structure to ensure that the mechanism with singular stiffness matrix will not be generated. When the structure is subjected to uniform load, the plane projection of the structure is divided into many small squares. The uniform load multiplied by the area of the small square is converted into a concentrated load acting on the geometric center of the small square. Finally, it is assigned to the nearest node. The penalty function method is used when the structure does not satisfy the constraint conditions such as material strength, the aspect ratio of the members, the stability of the members and the maximum displacement of the joints. In the selection of optimization objectives, the two optimization objectives selected in this paper are the minimum quality of the structure and the minimum product of the mass of the structure and the product of the external force. The advantages and disadvantages of the two optimization objectives are illustrated through the analysis and comparison of the numerical examples. A novel and beautiful structure with good mechanical performance is obtained through several examples, which shows the feasibility and practicability of the proposed method.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU399
【参考文献】
相关期刊论文 前5条
1 沈世钊;武岳;;结构形态学与现代空间结构[J];建筑结构学报;2014年04期
2 倪丽君;吕爱民;;Dymaxion——富勒生态设计思想的启示[J];华中建筑;2009年01期
3 宋晓霞;;遗传算法中初始群体技术的改进与实现[J];计算机工程与设计;2007年22期
4 崔昌禹;严慧;;自由曲面结构形态创构方法——高度调整法的建立与其在工程设计中的应用[J];土木工程学报;2006年12期
5 崔昌禹;严慧;;结构形态创构方法——改进进化论方法及其工程应用[J];土木工程学报;2006年10期
，本文编号：2353805
本文链接：https://www.wllwen.com/jingjilunwen/jianzhujingjilunwen/2353805.html
