基于数据增强和机器学习的铁尾矿混凝土抗压强度预测  

Predicting compressive strength of iron ore tailings concrete using data augmentation and machine learning

作　　者：荣兆川 胥孝川[1,2] 李志军 戴佳豪 RONG Zhaochuan;XU Xiaochuan;LI Zhijun;DAI Jiahao(School of Resources and Civil Engineering,Northeastern University,Shenyang 110819,China;Technology Innovation Center for Intelligent Water Resources and Environment,Northeastern University,Shenyang 110819,China)

机构地区：[1]东北大学资源与土木工程学院,辽宁沈阳110819 [2]东北大学智慧水利与资源环境科技创新中心,辽宁沈阳110819

出　　处：《新型建筑材料》2025年第3期95-101,141,共8页New Building Materials

基　　金：国家自然科学基金重点项目(52234004);2022年度沈阳市科学技术计划社会治理科技专项项目(22-322-3-02);中国建筑股份有限公司2022年度科技研发课题重点课题(CSCEC-2022-Z-21)。

摘　　要：研究了铁尾矿(IOTs)对混凝土28 d抗压强度的影响,并采用4种机器学习模型进行分析。通过选择6个关键变量,对145组数据进行描述性统计,采用变分自编码器(VAE)对数据进行增强,并比较了梯度提升、自适应增强、随机森林和高斯过程4种模型的性能。结果表明,梯度提升模型在各项评估指标上均表现最佳,特别是在决定系数、均方误差等关键指标上。基于梯度提升模型预测混凝土的28 d抗压强度,与实际数据吻合度高,泛化能力强,该模型为学者们提供了一种高效、经济、可靠的预测方法。This research examines the influence of Iron Ore Tailings(IOTs)on the compressive strength of concrete at 28 days.Employing a selection of four machine learning algorithms,a comprehensive analysis was undertaken.An initial selection of six pivotal variables facilitated the descriptive of 145 data sets,succeeded by data enhancement through the Variational Autoencoder(VAE).During the model development stage,a comparative assessment was made between Gradient Boosting,Adaptive Boosting,Random Forest,and Gaussian Process methodologies.Empirical findings underscored the Gradient Boosting algorithm's preeminence in performance metrics,notably in coefficient of determination and mean squared error criteria.The gradient boosting model-based prediction of the 28-day compressive strength of concrete has a high degree of consistency with actual data and strong generalization ability,provides scholars with an efficient,economical,and reliable prediction method.

关 键 词：铁尾矿 抗压强度 机器学习 数据增强 梯度提升 

分 类 号：TU528[建筑科学—建筑技术科学]