Modeling method of concrete material at mesoscale with refined aggregate shapes based on image recognition
Modeling method of concrete material at mesoscale with refined aggregate shapes based on image recognition
复制标题
基于图像识别的细化骨料形状混凝土材料介观建模方法
DOI：
10.1016/j.conbuildmat.2019.01.157
复制
发表时间：
2019-04-20
影响因子：
7.4
通讯作者：
Yin, Zhiyi
中科院分区：
文献类型：
--
作者：
Ruan, Xin;Li, Yue;Yin, Zhiyi
混凝土材料的数值模拟已经被广泛接受，而基于实际材料的骨料建模是关键。然而，颗粒形状控制仍需改进。提出了一种基于图像识别的基于统计规则的精化集合体建模方法。从样品的横截面上提取了4407个真实粒子的几何数据，研究了粒子尺寸和长径比的联合概率分布。统计规则是显著和稳定的。给出了几何参数的推荐分布函数，以指导模拟。将骨料模型细化为基于颗粒大小、长宽比和表面纹理的三层框架，并提出了相应的具体建模方法。为了考察该方法的效果，对800个模型进行了数值试验。结果表明，精确控制粒子的纵横比是进行精确模拟的必要条件。(C)2019爱思唯尔有限公司。保留所有权利。
Numerical simulation of concrete materials has become widely accepted, and modeling the aggregate based on the actual material is critical. However, the particle shape control still needs to be improved. In this paper, a refined aggregate modeling method is proposed using statistical rules based on image recognition. The geometrical data of 4407 real particles are extracted from the specimen cross-sections, and the joint probability distributions of particle size and aspect ratio are investigated. The statistical rules are remarkable and stable. The recommended distribution functions for geometrical parameters are provided to guide the simulation. The aggregate model is refined to a three-level framework based on the particle size, aspect ratio, and surface texture, and a corresponding concrete modeling method is also proposed. To investigate the method effect, numerical experiments on 800 models are conducted. The results show that it is necessary to precisely control the particle aspect ratio for accurate simulation. (C) 2019 Elsevier Ltd. All rights reserved.