基于MCMC算法的粗集料粒径分布推断研究

童欣; 陆艺; 李静伟; 范伟军

doi:10.3969/j.issn.1000-1158.2024.09.04

PDF(581 KB)

计量学报 ›› 2024, Vol. 45 ›› Issue (9) : 1273-1280. DOI: 10.3969/j.issn.1000-1158.2024.09.04

几何量计量

基于MCMC算法的粗集料粒径分布推断研究

童欣¹,陆艺¹,李静伟²,范伟军¹

作者信息 +

Coarse Aggregate Particle Size Distribution of Inference Based on MCMC Algorithm Research

TONG Xin¹，LU Yi¹，LI Jingwei²,FAN Weijun¹

Author information +

文章历史 +

摘要

运用机器视觉对振动盘掉落的粗集料进行数据采集，得到了动态集料的‘伪三维’图像。针对图像信息不能准确表达集料真实三维情况，引入贝叶斯统计推断的思想对集料进行粒径分布的推断，选取等效Feret短径作为图像特征，由于集料在大粒径情况下其Feret短径与实际粒径误差较大，故增加了等效椭圆短径作为第二特征。为获得准确的后验分布和高效的工程计算能力，采用了MCMC算法，解决了传统贝叶斯统计推断在高维计算不足的问题，从而得到了集料粒径分布的后验分布。实验结果表明：该方法的针对合格集料的粒径分布测量误差保持在±2.5%以内，不合格集料误差保持在±3.5%以内。

Abstract

Machine vision is used to collect data on the coarse aggregate dropped from the vibrating plate,and a ‘pseudothreedimensional’ image of the dynamic aggregate is obtained.Since the image information cannot accurately express the aggregate． Therefore,the idea of Bayesian statistical inference is introduced to infer the particle size distribution of aggregates.The equivalent Feret short diameter is selected as the image feature,but the error between the Feret short diameter and the actual particle size of the aggregate is large when the particle size is large,so the equivalent elliptical short diameter is added as the second feature.In order to obtain accurate posterior distribution and efficient engineering computing capabilities,the MarkovMonte Carlo (MCMC) algorithm is used,thus breaking through the problem of insufficient highdimensional calculations of traditional Bayesian statistical inference,and thus obtaining the aggregate Posterior distribution of particle size distribution.Experimental results show that the particle size distribution measurement error of this method for qualified aggregates is maintained within ±2.5%,and the error for unqualified aggregates is maintained within ±3.5%.

导出引用

童欣,陆艺,李静伟,范伟军. 基于MCMC算法的粗集料粒径分布推断研究[J]. 计量学报. 2024, 45(9): 1273-1280 https://doi.org/10.3969/j.issn.1000-1158.2024.09.04

TONG Xin，LU Yi，LI Jingwei,FAN Weijun. Coarse Aggregate Particle Size Distribution of Inference Based on MCMC Algorithm Research[J]. Acta Metrologica Sinica. 2024, 45(9): 1273-1280 https://doi.org/10.3969/j.issn.1000-1158.2024.09.04

中图分类号： TB92

参考文献

［1］唐健钦. 施工中粗集料级配对道路混凝土的影响分析［J］. 华东科技(学术版), 2013(12): 95.
TANG J Q. During the construction of the coarse aggregate level matching analysis of the influence of road concrete ［J］. Journal of east China science and technology (academic), 2013(12): 95.
［2］袁强. 骨料形状对沥青混合料性能影响的离散元研究［D］. 大连: 大连理工大学, 2015.
［3］粱汝恒, 刘星, 林礼跃. 粗骨料粒形对混凝土性能的影响［J］. 广东建材, 2003(8): 11-13.
LIANG R H, LIU X, LIN L Y. Coarse aggregate grain shape effect on the properties of concrete ［J］. Guangdong Building Materials, 2003(8): 11-13.
［4］史源. 基于LabVIEW的集料颗粒检测与级配分析［D］. 西安: 长安大学, 2014.
［5］程永春, 马健生, 颜廷野, 等. 基于数字图像处理技术的沥青混合料级配检测方法［J］. 科学技术与工程, 2017, 17(32): 332-338.
CHENG Y C, MA J S, YAN T Y, et al. Based on digital image processing technology of asphalt mixture gradation test method ［J］. Science Technology and Engineering, 2017, 17(32): 332-338.
［6］陈泽琦. 基于多视角粒径信息融合的集料级配检测研究［D］. 杭州: 中国计量大学, 2021.
［7］陈泽琦, 范伟军, 郭斌, 等. 基于形态重建和反向跟踪的粗集料级配视觉检测［J］. 计量学报, 2021, 42(6): 710-717.
CHEN Z Q, FAN W J, GUO B, et al. A Visual Detection Method for Particle Size of Moving Coarse Aggregate Based on Morphological Reconstruction and Reverse Tracking ［J］. Acta Metrologica Sinica, 2021, 42(6): 710-717.
［8］董成烨, 李东方, 冯槐区, 等. 基于机器视觉和机器学习技术的浙贝母外观品质等级区分［J］. 浙江大学学报(农业与生命科学版), 2023, 49(6): 881-892.
DONG C Y, LI D F, FENG H Q, et al. Classification of Fritillaria thunbergii appearance quality based on machine vision and machine learning technology ［J］. Journal of Zhejiang University (Agriculture and Life Sciences),2023, 49(6): 881-892.
［9］周建华, 房怀英, 杨建红, 等. 图像法集料粒径检测表征参数的选择及实验研究［J］. 计量学报, 2018, 39(6): 783-790.
ZHOU J H, FANG H Y, YANG J H, et al. Study on Characterization Parameters of Aggregate Particle Size Using Image Analysis ［J］. Acta Metrologica Sinica, 2018, 33 (6): 783-790.
［10］IGATHINATHANE C, MELIN S, SOKHANSAMJ S, et al. Machine vision based particle size and size distribution determination of airborne dust particles of wood and bark pellets ［J］. Powder Technology, 2009, 196(2): 202-212.
［11］邵建林, 徐东, 王兰州, 等. 一种新的预测蛋白质二级结构的模型—贝叶斯神经网络［J］. 计量学报, 2006, 27(3): 281-285.
SHAO J L, XU D, WANG L Z, et al. BNN: A Novel Model for Prediction of Protein Secondary Structure ［J］. Acta Metrologica Sinica, 2006, 27(3): 281-285.
［12］张玉敏, 韩学山, 杨明, 等. 基于狄利克雷模型的分布鲁棒机组组合［J］. 中国电机工程学报, 2019, 39(17): 5074-5084,5288.
ZHANG Y M, HAN X S, YANG M, et al. Distributionally Robust Unit Commitment Based on Imprecise Dirichlet Model ［J］. Proceedings of the CSEE, 2019, 39(17): 5074-5084,5288.
［13］许士芳, 谢立, 刘济林. 基于MCMC粒子滤波的机器人定位［J］. 浙江大学学报(工学版), 2007, 41(7): 1083-1087.
XU S F, XIE L, LIU J L. Robot localization based on MCMC particle filter ［J］. Journal of Zhejiang University(Engineering Science), 2007, 41(7): 1083-1087.
［14］胡雨. 基于MCMC算法的贝叶斯网络结构学习［D］. 金华: 浙江师范大学, 2018.
［15］那威, 邴佳乐, 刘品妍. 基于贝叶斯MCMC方法的城市建筑供热能耗基准［J］. 浙江大学学报(工学版), 2023, 57(10): 2106-2115.
NA W, BING J L, LIU P Y. Urban building heating energy consumption benchmark based on Bayesian MCMC method ［J］. Journal of Zhejiang University(Engineering Science), 2023, 57(10): 2106-2115.
［16］胡红波, 孙桥, 杜磊. GUM S1与基于贝叶斯方法的不确定度评估比较［J］. 计量学报, 2017,38(4): 517-520.
HU H B, SUN Q, DU L. Comparison of Uncertainty Evaluation between GUM S1 and Bayesian Analysis ［J］. Acta Metrologica Sinica, 2017, 38(4): 517-520.