【作者】 赵秀阳；

【导师】 尹衍升；

【作者基本信息】 山东大学 ， 材料学， 2006， 博士

【摘要】 复合材料细观力学的核心任务是建立复合材料宏观性能及其组分和细观结构之间的定量关系。进行复合材料细观力学研究首先需要引入复合材料代表单元的概念(RVE)。目前的复合材料有限元细观力学研究绝大部分是建立在将增强相简化为球形，椭球或六面体等规则的几何形状的基础上，这种简化为复合材料的有限元细观力学理论体系的建立发挥了巨大的作用。但是对于微观结构非常复杂的颗粒增强，特别是增强相和基体发生了反应的复合材料，采用传统的有限元细观力学进行宏细观力学性能的分析，其结果就不一定十分正确。随着计算机科学技术的发展，特别是科学计算可视化技术的进步，复合材料的微观结构就可以运用科学计算可视化技术呈现在人们面前。三维重构技术很早以来就已经应用在医学领域，而且取得了很大的成功。但是在材料科学领域，起步较晚，尤其是国内更晚。即使在国外，采用三维重构技术获得材料的微观形貌，采用的也多是医学或者通用的软件，而针对材料科学领域开发的三维重构软件，还少之又少。 本文基于这种现状，系统研究了复合材料科学的三维重构技术，并将其与复合材料有限元细观力学相结合，建立了基于真实微观结构的复合材料有限元细观力学模型，并将该模型预测的复合材料有效性能和基于规则几何体的复合材料有限元细观力学模型预测的有效性能进行了比较，结果表明本文建立的模型的预测结果更接近实验结果，且误差很小。 本文具体做了以下工作： 采用序列切片方法，获得了颗粒增强复合材料的序列显微图像，通过单轴拉伸实验获取了SiC／Al复合材料的应力-应变曲线。为了解决在高倍显微镜下获取的显微图像视野较小的问题，对复合材料显微图像进行了基于特征点提取的自动拼接。采用Harris算子和Canny算子提取两幅待配准显微图像的特征点，采用互相关方法获取得到相对应的特征点序列，然后进行基于特征点序列的SVD-ICP配准，确定精确的对应关系，拼接这两幅显微图像得到过渡光滑自然，达到无缝拼接要求的足够大的复合材料显微图像。 研究了获取复合材料第二相颗粒轮廓的方法。利用小波多分辨率原理对复合材料第二相颗粒的边缘进行了检测，从高分辨率到低分辨率逐步搜索边缘，最后得出边缘位置准确的轮廓图。针对小波边缘检测方法获取的颗粒边缘不连续的缺点，在灰度复杂的金相显微照片中即使GVF Snake算法仍要求初始轮廓必须靠近颗粒边缘的情况下，提出了一种新的Snake初始轮廓的定义方法。该方法利用B样条插值和重采样技术，在利用小更多还原

【Abstract】 The key assignment of composite micro-mechanics is to build the quantitative relation of various macro-performance and the microstructure. Traditionally simplified unit cell models based on the finite element analysis have been used to resolve micro-mechanics problems such as the prediction of the onset and growth of evolving damage in composite materials. While these models provide valuable insights into the microstructural processes, simple morphologies idealize actual microstructures for many engineering materials that bear little relationship to the actual stereographic features. However, for complicated composite, many work has shown that traditionally unit cell models tend to give unsatisfactory predictions for the mechanical response of particle reinforced materials. It is the central precept of materials science that microstructure influences the properties and performance of materials. For composite materials, the degree of property enhancement depends on morphological factors such as volume fraction, size, shape and spatial distribution of the reinforcements, in addition to the constituent material and interface properties. With the development of computer science, direct observations of three-dimensional microstructure are of interest. For opaque materials, 3D reconstruction from serial sectioning has been used in the past to obtain information concerning three-dimensional microstructural geometry. Currently There are few special software aim to the 3D reconstruction of material science.The aim of this paper is to study the key technologies in the 3D reconstruction for composite are studied systemically, and predict the effective properties based on the true microstructure.The calculating result shows that the data calculated by the 3D microstructure model are more accurate than the datas acquired from 3D spherical particles model and analytical model.With the serial sectioning method, the serial images of particle reinforced metal matrix composites are acquired and the strain-stress curve of SiC/Al composite is obtained by trial.To solve the problem that only very limited field of view is available got when highly magnifying lens are used in microscopes, Image mosaic is processed to align the serial images. A novel method is adopted to mosaic two SiC/Al microscopic images to avoid the drawbacks of the classical methods. Feature points are extracted by using Harris corner detector and filtered through Canny edge detector. A 40 × 40 feature template is chosen by sowing a seed in the更多还原