基于应变模态识别FRP层合梁板刚度的方法

发布时间：2018-03-10 20:03

本文选题：改进遗传算法　切入点：FRP梁板结构　出处：《武汉理工大学》2015年硕士论文　论文类型：学位论文

【摘要】：本课题来源于中央高校基本科研业务费专项资金资助“基于自适应混合遗传算法的复合材料及结构损伤反演研究”(项目号2014-Ia-036)。本学位论文基于应变模态对复合材料层合梁板进行刚度识别:正问题计算借助于ANSYS有限元计算软件、反问题计算利用MATLAB软件编写的改进遗传算法,对FRP梁板结构进行单元刚度识别。利用单纯型搜索结合自编自适应遗传算法的方式来加快收敛的速度和改善易陷入局部最优解的缺点。具体的研究内容如下:1、对有损悬臂梁结构进行应变模态理论公式推导,并且对悬臂树脂基玻璃纤维增强复合材料(FRP)层合梁结构进行应变模态实验,悬臂梁某处粘贴质量块以模拟刚度变化(损伤程度用刚度变化来表征);对有损和无损两种工况进行单元刚度识别研究,并检验识别结果的准确性。实验结果表明应变模态对单元损伤灵敏度高,能准确反映损伤位置。利用改进遗传算法对单元刚度的识别结果与有限元模拟实验吻合度高。2、对悬臂树脂基玻璃纤维增强复合材料层合板结构进行应变模态试验,获取其纵向方向的应变模态,并以实验数据为基础进行纵向刚度识别。对比研究改进自适应遗传算法和普通自适应遗传算法的识别性能,并研究单元尺寸对识别结果的影响。研究表明以实验模态为基础进行的识别,改进遗传算法识别的准确度更高,需要迭代的次数也较普通遗传算法少。减小识别单元的尺寸,能有效的提高识别的准确度,但是在工程中需要花费更多的工程量,才能获取全局的刚度信息。3、对悬臂树脂基玻璃纤维增强复合材料层合板结构进行有限元数值模拟实验,计算应变模态;在此基础上对板的纵横刚度进行识别。选取其中某些单元进行横向和纵向的刚度折减,以纵向和45°方向的应变模态为基础进行刚度识别研究。对识别结果进行对比研究,并讨论灵敏度对识别效果的影响。并通过对不同位置,不同程度的损伤识别,检验识别方法的实用性和可行性。研究发现以45°一阶应变模态为基础,识别的效果最佳,而且采用的应变模态对单元刚度的改变越敏感,识别效果越好。以45°方向一阶应变模态为基础对不同位置不同程度的损伤识别可以得到理想的识别结果。且损伤位置的多少对识别迭代速度和结果的精度无绝对关系。
[Abstract]:This topic comes from the central university basic scientific research business expense special fund to fund "the composite material and the structure damage inversion research based on the adaptive hybrid genetic algorithm" (project number 2014-Ia-036A). This dissertation is based on the strain mode to the composite material. Stiffness identification of laminated beams and plates: the forward problem is calculated with the help of ANSYS finite element software. Inverse problem calculation is based on the improved genetic algorithm written by MATLAB software. The element stiffness of FRP beam-plate structure is identified. The simplex search method combined with self-made adaptive genetic algorithm is used to accelerate the convergence speed and improve the disadvantage of falling into the local optimal solution easily. The specific research contents are as follows: 1. The theoretical formula of strain mode is derived for the structure of lossy cantilever beam. And the strain mode experiment of the cantilever resin matrix glass fiber reinforced composite (FRP) laminated beam structure was carried out. A mass block attached to a cantilever beam is used to simulate the change of stiffness (the degree of damage is characterized by the change of stiffness). The experimental results show that the strain mode is sensitive to the damage of the element. The result of identification of element stiffness by improved genetic algorithm is in good agreement with finite element simulation experiment. Strain mode test of cantilever resin matrix glass fiber reinforced composite laminated plate structure is carried out. The longitudinal strain modes are obtained, and the longitudinal stiffness is identified based on the experimental data. The performance of improved adaptive genetic algorithm (AGA) and conventional adaptive genetic algorithm (AGA) is compared. The effect of unit size on the recognition results is also studied. It is shown that the improved genetic algorithm has higher accuracy and less iterations than the common genetic algorithm, and reduces the size of the recognition unit. It can effectively improve the accuracy of identification, but it needs more engineering cost in engineering to obtain global stiffness information. The finite element numerical simulation experiment of cantilever resin matrix glass fiber reinforced composite laminated plate structure is carried out. The strain mode is calculated. On this basis, the longitudinal and horizontal stiffness of the plate is identified. Some of the elements are selected to reduce the transverse and longitudinal stiffness. Stiffness identification is carried out on the basis of longitudinal and 45 掳strain modes. The results of identification are compared, and the influence of sensitivity on the recognition effect is discussed. The practicability and feasibility of the identification method are tested. It is found that based on the 45 掳first-order strain mode, the best result is obtained, and the more sensitive the strain mode is to the change of element stiffness, The better the recognition effect is, the better the result is. Based on the first-order strain mode of 45 掳direction, we can obtain the ideal identification results for different positions and different degrees of damage, and there is no absolute relation between the number of the damage location and the accuracy of the identification iterative speed and the result.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TB33;TP18

【参考文献】