摘要
进行了18组不同配合比的再生混凝土立方体试块的抗压强度试验,分析了水胶比、再生骨料取代率和粉煤灰掺量对再生混凝土抗压强度的影响规律。建立了基于深度学习神经网络的抗压强度预测模型,以水胶比、再生粗骨料取代率、再生细骨料取代率和粉煤灰掺量为输入变量预测再生混凝土试块的抗压强度。结果表明:与传统神经网络模型相比,基于深度学习的预测模型具有高精度、高效率和高泛化能力的优点,可以作为再生混凝土强度计算经验式的一种新方法。
The compressive strength tests of recycled concrete speciments with18different sets of mix ratios are carried out.The influence of water gel ratio,recycled aggregate replacement ratio and fly ash content on the compressive strength of recycled concrete speciments is analyzed.A prediction model of compressive strength based on depth learning neural network is established,in which model the compressive strength of recycled concrete speciments was predicted with water gel ratio,recycled coarse aggregate replacement rate,recycled fine aggregate replacement rate and fly ash content as input variables.Compared with the traditional neural network model,the results showed that prediction of deep learning model has advantages of high precision and high efficiency and high generalization ability,whicn can be based on a new method for calculating the strength of recycled concrete as.
作者
高蔚
AO Wei(Jiangxi College of Foreign Studies,Nanchang 330099,China;Department of Civil and Architecture,Nanchang University College of Science and Technology,Nanchang 330031,China)
出处
《混凝土》
CAS
北大核心
2018年第11期58-61,70,共5页
Concrete
基金
国家自然科学基金(61663013)
江西省重点研发计划(20161BBE50076)
江西省自然科学基金(20171BAB206045)
江西省教育厅科学技术项目(GJJ160491)
关键词
再生混凝土
抗压强度
预测模型
深度学习
卷积神经网络
recycled concrete
compressive strength
prediction model
deep learning
convolutional neural network
