基于感知信息的混凝土桥梁耐久性智能诊断方法  

A Method for Intelligent Diagnosing Durability of Concrete Bridge Based on Perceptual Information

作　　者：李殿斌 李伟 陈南[2] LI Dian-bin;LI Wei;CHEN Nan(China Highway Engineering Consulting Corporation,Beijing 100089,China;Research Institute of Highway,Ministry of Transport,Beijing 100088,China)

机构地区：[1]中国公路工程咨询集团有限公司,北京100089 [2]交通运输部公路科学研究院,北京100088

出　　处：《公路交通科技》2022年第8期70-75,共6页Journal of Highway and Transportation Research and Development

基　　金：交通运输部公路科学研究院交通强国试点项目(QG2021-1-3-2)。

摘　　要：混凝土桥梁结构于其生命周期中常因自然环境因素的影响而产生劣化与腐蚀,造成耐久性降低及使用年限减少,一般大气环境下桥梁结构混凝土碳化现象是造成混凝土桥梁耐久性下降的重要原因之一。结合桥梁结构混凝土材料的特点分析了碳化现象的原理及主要影响因素,提出了用于碳化深度预测的自回归时间序列模型(BP-AR),并利用依托工程中采集的感知数据对结构实际发生的碳化深度进行了预测。开展室内试验并将试验结果与模型测算结果进行比较,验证了BP-AR预测算法的预测精度和可靠性。结果表明:混凝土碳化深度的影响因素复杂多变且交叉耦合,在感知信息数据有限的条件下,利用BP-AR模型由于在回归过程中考虑了碳化现象的时效性,因而能够获得比一般神经网络模型(BP)更准确的预测结果,具有较高的预测精度;建立的模型充分考虑了时间依赖性来实现碳化深度的神经网络预测,即把碳化反应随时间的变化与其他影响因素统筹考虑,形成一种BP-AR融合算法,保证了对碳化深度的预测和修正,达到了提高模型预测精度的目的,揭示了结构混凝土碳化机制。Concrete bridge structure is often degraded and corroded in its life cycle due to the influence of natural environmental factors,resulting in the reduction of durability and service life.The carbonation phenomenon of bridge structure concrete in general atmospheric environment is one of the important reasons for the decline of durability of concrete bridges.According to the characteristics of concrete materials in bridge structure,the principle and main influencing factors of carbonation phenomenon are analyzed,an autoregressive time series model(BP-AR) for predicting carbonation depth is proposed,and the actual carbonation depth of the structure is predicted by using the sensing data in in the supporting project.The laboratory test is carried out,and the test result is compared with the model calculation result to verify the prediction accuracy and reliability of the BP-AR prediction algorithm.The result shows that(1) The influencing factors of concrete carbonation depth is complicated and variable and cross-coupling.Under the condition of limited perceptual information data,using BP-AR model can obtain more accurate prediction result than the general neural network model(BP) because it considers the timeliness of carbonization phenomenon in the regression process,it has higher prediction accuracy.(2) The established model fully considers the time dependence to realize the neural network prediction of carbonation depth,that is,the carbonation reaction changes over time and other influencing factors are considered as a whole,and a BP-AR fusion algorithm is formed,which ensured the prediction and correction of carbonation depth,achieved the purpose of improving the prediction accuracy of the model,and revealed the carbonation mechanism of structural concrete.

关 键 词：桥梁工程 碳化深度 神经网络 混凝土桥梁 人工智能 

分 类 号：U448.27[建筑科学—桥梁与隧道工程]