运营荷载作用下大跨桥梁应变精细化建模与性能退化预警研究

In this project, the box-girder of long-span bridge is taken as the research object. Based on the long-term monitoring data and statistical analysis theory, the temperature affecting model of quasi-static strain response and the varying load affecting model of dynamic strain response are established for the main-girder, and its performance degradation alarming is realized. Under normal operations, the quasi-static strain is mainly affected by temperature (which is easy to measure). The overall correlation between multi-point temperature measurements and multi-point quasi-static strain measurements is analyzed comprehensively. The formula for calculating the representative temperature which describes the temperature field of the main-girder is deduced, and a precise temperature affecting model for the quasi-static strain response of the main-girder is established. Under normal operations, the dynamic strain is mainly affected by varying load (which is difficult to measure). The latent-variable analysis model is directly established for the dynamic strain measurements, and then the multiple coupling factors are separated. Based on the difference between statistical characteristics, the criteria for distinguishing varying load effects and random errors are proposed, and the principal latent-variables (representing the varying load) are selected to reconstruct varying load effects. The quasi-static and dynamic responses of strain measurement-points are taken as performance alarming indicators. The probability distribution functions of the errors of temperature affecting model and varying load affecting model are fitted to set the alarming threshold in normal service state. The extreme-value distribution functions of temperature and varying loads are fitted and their extreme-values are estimated during the design service period, and then the extreme-values of temperature and varying loads are predicted to set the alarming threshold under the limit service state.

以大跨桥梁箱型主梁为研究对象，利用长期监测数据和统计分析理论建立主梁准静态应变的温度作用模型和动态应变的变载作用模型，基于此实现主梁性能退化预警。正常运营状态下的准静态应变主要受温度作用（易测）影响，综合分析多测点温度和多测点准静态应变间的整体相关性，推导描述主梁温度场变化的表征温度计算公式，建立主梁准静态应变响应的精细化温度作用模型。正常运营状态下的动态应变主要受变载作用（难测）影响，直接对动态应变数据建立隐变量分析模型，分离其多重耦合影响因素，基于统计特征差异性提出区分变载效应和随机误差的准则，筛选主要隐变量（变载作用）以重构变载效应。以各应变测点的准静态响应和动态响应作为预警指标，拟合温度作用和变载作用模型误差的概率分布函数以设置正常使用状态下的预警阈值，拟合温度作用和变载作用的极值分布函数并估计它们在设计使用期内的极值，进而预测温度效应和变载效应的极值以设置极限状态下的预警阈值。

如何分析经年积累的监测数据建立精细化运营荷载作用及其效应模型以实现服役状态评估，是现阶段桥梁结构健康监测领域的核心问题之一。本项目提出了温度作用与准静态响应监测数据的多维相关分析方法，推导了桥梁表征温度计算公式，建立了考虑不确定性和可解释性的温度作用-准静态响应概率关系模型；改进了经典的主成分分析和独立分量分析方法，即子域加权主成分分析和动态独立分量分析，使之适应实际桥梁动态响应的非线性相关和非高斯分布特性，并基于此消除数据中的多重耦合影响因素；定义了时变运营荷载作用下桥梁服役性能的鲁棒预警指标，提出了相应的预警阈值设置和异常变量隔离方法。研究成果可为部署结构健康监测系统的大跨桥梁服役性能评估提供理论支撑。