基于随机森林的混凝土早期碳化预测  

Prediction of Early Carbonization of Concrete Based on Random Forest

作　　者：吴贤国[1] 王雷 邓婷婷 胡毅 袁福银 李铁军[4] WU Xian-guo;WANG Lei;DENG Ting-ting;HU Yi;YUAN Fu-yin;LI Tie-jun(School of Civil Engineering&Mechanics,Huazhong University of Science and Technology,Wuhan 430074,China;CCCC Second Harbour Engineering Co Ltd,Wuhan 430040,China;Road&Bridge International Co Ltd,Beijing 100027,China;China Communications Construction Co Ltd,Beijing 100088,China)

机构地区：[1]华中科技大学土木工程与力学学院,湖北武汉430074 [2]中交第二航务工程局有限公司,湖北武汉430040 [3]中交路桥建设有限公司,北京100027 [4]中国交通建设股份有限公司,北京100088

出　　处：《土木工程与管理学报》2020年第2期14-19,26,共7页Journal of Civil Engineering and Management

基　　金：国家重点研发计划(2016YFC0800208);国家自然科学基金(51378235,71571078,51308240)。

摘　　要：碳化会降低混凝土结构的耐久性,减少混凝土结构的使用寿命,因此快速准确预测混凝土碳化对于混凝土的质量评估具有重要意义。本文基于随机森林算法,以松原至通榆段高速公路项目为研究背景,建立随机森林(RF)算法预测模型。构建混凝土早期碳化影响因素指标体系,选择碳化深度作为碳化性能评价指标,根据原始数据建立训练集和测试集,利用Pearson相关性矩阵图分析影响因素相关度,并建立随机森林训练模型,结果表明,利用随机森林预测模型对混凝土碳化进行预测是一种新的有效方法。Carbonation will reduce the durability of concrete structure and the service life of concrete structure.Therefore,it is of great significance to quickly and accurately predict concrete carbonation for the quality assessment of concrete.Based on the random forest algorithm,taking the highway project of Songyuan-Tongyu section as the research background,the random forest(RF)algorithm prediction model is established.The index system of influencing factors of early carbonation of concrete is constructed,the carbonation depth is selected as the evaluation index of carbonation performance,the training set and test set are established according to the original data,the Pearson correlation matrix is used to analyze the correlation degree of influencing factors,and the random forest training model is established to predict the test set.Compared with the prediction errors of BP artificial neural network and wavelet neural network model,the random forest algorithm has better prediction effect.The results show that using the random forest prediction model to predict concrete carbonation is a new and effective method.

关 键 词：混凝土 早期碳化 碳化深度预测 随机森林 

分 类 号：TU528[建筑科学—建筑技术科学]