早期受冻混凝土耐久性预测模型研究  

Research on Prediction Model of Early Frost Durability of Concrete

作　　者：徐存东 陈家豪 李准 张鹏 连海东 XU Cun-dong;CHEN Jia-hao;LI Zhun;ZHANG Peng;LIAN Hai-dong(School of Water Conservancy,North China University of Water Resoures and Electrie Power,Zhengzhou 450046,China;Key Laboratory for Technology in Rural Water Management of Zhejiang Province,Hangzhou 310018,China;Henan Provincial Hydraulic Structure Safety Engineering Rescarch Center,Zhengzhou 4150046,China)

机构地区：[1]华北水利水电大学水利学院,河南郑州450046 [2]浙江省农村水利水电资源配置与调控关键技术重点实验室,浙江杭州310018 [3]河南省水工结构安全工程技术研究中心,河南郑州450046

出　　处：《水电能源科学》2022年第8期148-152,共5页Water Resources and Power

基　　金：国家自然科学基金项目(51579102);河南省高校科技创新团队支持计划(19IRTSTHN030);中原科技创新领军人才支持计划(204200510048);河南省科技攻关项目(212102310273);河南省高等学校重点科研项目计划(20A570006);浙江省重点研发计划(2021C03019)。

摘　　要：针对早期受冻混凝土从实验室拟定的配合比设计中获得具有针对性的成果非常耗时的问题,以甘肃省景电灌区水工建筑物群为试验原型,开展冻融循环、单轴压缩等室内材料试验,在混凝土受冻温度为-10℃、3.5 h后起冻的条件下,通过改变配合比研究其在盐冻环境中单轴抗压的强度变化,采用反向传播神经网络(BPNN)、广义回归神经网络(GRNN)两种方法预测其抗压强度,并建立了三种新的GRNN模型进行影响因素敏感度分析。结果表明,在四种回归评估指标中,GRNN全面优于BPNN,说明此类型混凝土抗压强度预测采用GRNN模型效果更好;水胶比为影响早期受冻混凝土最大的变量,而引气剂为影响最小的变量。Aiming at the problem that it is very time-consuming to obtain targeted results from the mix proportion design proposed by the laboratory for the early frozen concrete,taking the hydraulic building group in Jingdian Irrigation District in Gansu Province as the experimental prototype,this paper carried out indoor material tests such as freeze-thaw cycles and uniaxial compression.Under the condition that the freezing temperature of concrete is-10℃and the freezing time is 3.5 h and then freezing,the change of its uniaxial compressive strength in the salt-freezing environment was studied by changing the mix ratio.Back propagation neural network(BPNN)and generalized regression neural network(GRNN)were used to predict the compressive strength,and three new GRNN models were established to analyze the sensitivity of influencing factors.The results show that among the four regression evaluation indexes,GRNN is superior to BPNN in all aspects,indicating that GRNN model is better for the compressive strength model of this type of concrete.The water-binder ratio is the variable that affects the early frozen concrete the most,and the air-entraining agent is the least influential variable.

关 键 词：混凝土 早期受冻 盐冻环境 神经网络 预测模型 影响因子 

分 类 号：TV431[水利工程—水工结构工程]