基于一维卷积神经网络的粉煤灰混凝土氯离子质量分数预测  

Prediction of chloride concentration in fly ash concrete based on one-dimensional convolutional neural network

作　　者：章玉容 余威龙 王龙龙 唐科 ZHANG Yurong;YU Weilong;WANG Longlong;TANG Ke(College of Civil Engineering,Zhejiang University of Technology,Hangzhou 310023,China;Key Laboratory of Civil Engineering Structure&Disaster Prevention and Mitigation Technology of Zhejiang Province,Hangzhou 310023,China;Hangzhou Trade and Tourism Group Co.,Ltd.,Hangzhou 310003,China)

机构地区：[1]浙江工业大学土木工程学院,浙江杭州310023 [2]浙江省工程结构与防灾减灾技术研究重点实验室,浙江杭州310023 [3]杭州市商贸旅游集团有限公司,浙江杭州310003

出　　处：《浙江工业大学学报》2024年第2期156-163,共8页Journal of Zhejiang University of Technology

基　　金：浙江省水利防灾减灾重点实验室开放基金资助项目(FZJZSYS21004)。

摘　　要：为研究深度学习方法在氯离子质量分数预测中的应用,基于自然潮差环境下粉煤灰混凝土长期暴露试验获取了3150组自由氯离子质量分数数据,建立一维卷积神经网络(One-dimensional convolutional neural network,1D-CNN)模型用于预测粉煤灰混凝土氯离子质量分数。该模型分析了核函数和卷积层对1D-CNN预测精度的影响,研究了水灰比、暴露时间、粉煤灰掺量和渗透深度4个输入参数对粉煤灰混凝土自由氯离子质量分数预测结果的影响。实验结果表明:采用12个3×1卷积核及两层卷积层构建1D-CNN模型时,自由氯离子质量分数的预测结果最优;同时,应用最优的1D-CNN模型开展基于未测参数的自由氯离子质量分数预测,预测结果较为准确。因此,1D-CNN模型具有精度高和适用范围广泛的特点,能够为氯盐环境下混凝土中自由氯离子质量分数预测提供新的方法。To explore the application of deep learning theory to predicting chloride concentration,3150 groups of free chloride concentration data are obtained by a long-term exposure test of fly ash concrete under natural tidal environment.An one-dimensional convolutional neural network(1D-CNN)model is established to predict the free chloride concentration in fly ash concrete by considering the influence of four input parameters:water-cement ratio,exposure time,fly ash content and penetration depth.Besides,the effects of kernel function and number of convolutional layers on the prediction accuracy of the 1D-CNN are analyzed.Results show that the best prediction results can be obtained when the 1D-CNN model is constructed by using 123×1 convolution kernels and two convolutional layers.Meanwhile,the selected optimal 1D-CNN model isselected to predict the free chloride concentration based on the unmeasured parameters,and a relatively accurate precision is obtained.Therefore,the 1D-CNN model has the advantages of high precision and wide application scope,which can be used as a new method for predicting the free chloride concentration in concrete under chloride environment.

关 键 词：自由氯离子质量分数 一维卷积神经网络 粉煤灰混凝土 

分 类 号：TU528.1[建筑科学—建筑技术科学]