基于数字图像技术的混凝土细观层次力学建模

Meso-level analysis model for concrete based on digital image processing

应用数字图像技术和MATLAB图像处理工具,针对二维混凝土骨料的边界提取进行研究,并进行细观层次的有限元分析。依据灰度直方图选取[120,210]为阈值对数字照片样本进行分段变换,然后依据骨料的尺寸和骨料内部杂质大小等因素选取16×16像素的方形模块对图片进行中值滤波。应用形态学理论选取半径为6像素的"disk"构造结构元素对图片作开运算和闭运算,最终应用Canny边缘检测方法得到了与原始数字图片相近的骨料边界。基于商业软件ABAQUS对提取的骨料建立混凝土细观有限元分析模型,提出面积级配曲线并按此曲线分别生成与真实骨料模型相似的圆形骨料模型和任意凸多边形模型,对三种骨料模型的四点弯拉和轴拉断裂过程进行非线性有限元分析。计算结果表明:面积级配能够较好的衡量骨料模型之间的相似性;骨料分布越均匀,承载能力越高;多边形骨料模型加载的软化段离散性较大;对于形状为非圆滑类骨料,多边形骨料模型的软化段与真实骨料模型具有更好的相似性。

The edge detection method for aggregates in concrete section digital photo was developed based on digital image technology and the image processing tools of MATLAB for carrying out the finite element analysis of the boundary identification at meso-level. Firstly, the digital image samples were converted sec- tionally according to the gray-scale colmnn diagram with the threshold value [120, 210], then a square model 16×16 pixels was selected according to the dimension and internal impurity＇s size of the aggregates, and the noise was eliminated by median filter. Based on the morphology theory, a disk tectonic structure with radius of 6 pixels was selected for the open operation and close operation of the photos. Finally, the aggregates edges approximate to the original photo were obtained by Canny Edge Detection Method. The me- so-level FEM model for aggregates is created based on the commercial nonlinear finite element code ABAQUS. The area gradation curve was proposed, and the round shape aggregate models and arbitrary convex polygon aggregate models which have similar area gradation curves to real aggregate model are present- ed in this paper. The four-point flexural-tension and axial-tension fracture processes of models for three kinds of aggregate were analyzed by nonlinear FEM and arc-length method. Numerical results show that the area gradation curves is useful to characterize the similarity of different aggregate models. When the aggre- gates are distributed more homogeneous, the load-bearing capacity of the models are higher. The softening stage of arbitrary convex polygon aggregate models is more discrete. For the non-circular aggregates, the arbitrary convex polygon aggreate model have more similar softening stages to the real aggregate model.