基于BP神经网络的新老混凝土界面黏结强度预测  

Prediction of bond strength between new and old concrete based on BP neural network

作　　者：陈昊 范颖芳 CHEN Hao;FAN Yingfang(Department of Civil Engineering,College of Transportation Engineering,Dalian Maritime University,Dalian 116026,China)

机构地区：[1]大连海事大学交通运输工程学院土木工程系,辽宁大连116026

出　　处：《混凝土》2024年第7期43-49,共7页Concrete

基　　金：国家自然科学基金(51578099)。

摘　　要：为探寻能够考虑多因素综合影响的新老混凝土界面黏结强度预测方法,在综合考察国内外新老混凝土界面黏结强度试验资料基础上,搭建试验资料数据库;对比分析了国内外规范和几种常用简化计算方法对新老混凝土界面黏结强度计算的优缺点;基于神经网络技术,建立8因素(老混凝土龄期、抗压强度、新混凝土抗压强度、黏结界面面积、界面饱水程度、界面粗糙度、界面剂、试验方法)影响的新老混凝土界面黏结强度BP神经网络模型,模型预测值与试验值能够很好地吻合。The prediction method of the interface bond strength between old and new concrete is examined.Based on hundreds of experimental data reported in the literature worldwide,a database is then built up.The bond strength between the old and new concrete calculated by the standard methods and the simplified methods are compared with the experimental results.Based on neural network technology,a BP neural network model is established for the prediction of the bond strength between new and old concrete.Eight factors(old concrete age,compressive strength ofold concrete,compressive strength of new concrete,interfacial area,interface saturation degree,interface roughness,interface agent,test method)are taken into consideration.The predicted results obtained by the BP neural network achieve good agreement with the test results.

关 键 词：BP神经网络 新老混凝土界面 黏结强度 预测 

分 类 号：TU528.01[建筑科学—建筑技术科学]