【作者】 曹建新；

【导师】 刘洋；

【作者基本信息】 哈尔滨工业大学 ， 土木工程， 2022， 博士

【摘要】 梁桥集群是指由城市高架或公路交通线路内多座中小跨径梁桥所构成的结构群。运营环境下桥梁结构响应监测数据是车辆荷载、温度荷载以及收缩徐变等各种耦合作用的综合反应,仅利用单座桥梁结构响应监测数据通常难以有效剔除复杂耦合因素对桥梁结构损伤定位的影响,因此,如何实现运营环境下集群内全部桥梁结构的损伤定位是一个具有挑战性的难题。针对上述难题,分别以结构相同的梁桥集群、相似的梁桥集群及缺少健康监测数据的梁桥集群为研究对象,利用梁桥集群结构所受环境及车辆荷载相似的特点,融合集群内多座桥梁结构响应的监测数据,开展了运营环境下中小跨径梁桥集群结构损伤定位方法的系统研究,主要研究内容如下:针对时变环境下结构相同梁桥集群的损伤定位问题,提出集群内相同梁桥结构损伤定位的概率SDDLV(Stochastic dynamic damage locating vector)方法。讨论传统SDDLV方法在单桥损伤定位中的局限性,探索基于梁桥集群结构特性的损伤定位向量构造算法;考虑环境温度时变性的影响,研究基于高斯混合聚类和概率有限元模型的结构应力场计算方法;构建基于损伤定位指标概率特征的损伤定位阈值计算方法,结合假设检验和交叉验证策略,实现集群内结构相同梁桥的损伤交叉定位。相对于传统单桥SDDLV方法,解决了时变环境下仅利用稀疏测点即可完成梁桥上部结构全部区域损伤定位的问题。在前述研究的基础上,为了避免模态参数识别误差,进一步提出集群内相同梁桥结构损伤定位的应变投影差值比方法。利用集群内相同梁桥结构关键受力断面的应变监测数据,研究基于应变投影差值比的桥梁结构损伤定位特征构造算法;讨论所提损伤定位特征的统计特性与桥梁结构相似性程度之间的相互关系,探索算法的适用条件;构建基于损伤特征向量空间夹角的损伤定位指标,进而结合核密度估计与交叉验证策略,实现对集群内全部相同梁桥关键受力断面处微小损伤的交叉定位。相对于传统单桥结构损伤定位方法,该方法可更加有效剔除运营环境对损伤定位精度的影响。在解决集群内相同梁桥结构损伤定位的基础上,针对时变环境下集群内相似梁桥结构的损伤定位问题,提出集群内相似梁桥结构损伤定位的应变时空关联模型方法。理论分析集群内相似梁桥结构之间应变监测数据的时空关联性,揭示应变监测数据间复杂的时变映射关系;探索联合卷积神经网络和长短时记忆网络的深度学习架构,研究梁桥集群结构应变监测数据时空关联模型的构建算法;将该模型作为应变变化趋势预测框架,可有效剔除复杂耦合因素的影响,准确定位集群内全部相似梁桥结构的损伤位置。针对既有相似梁桥集群结构损伤的快速定位问题,提出无参考模型的集群内相似梁桥结构损伤快速定位方法。理论推导移动荷载作用下简支梁桥和连续梁桥的应变时程曲线面积方程;基于集群内不同桥梁各监测分位点处应变时程曲线面积的构成特征,构建基于应变时程曲线面积比的梁桥结构损伤定位指标;探讨归一化损伤定位指标与主梁结构损伤的关系,实现短暂封闭交通下梁桥集群结构损伤的快速定位。该方法不需桥梁历史健康数据和有限元模型作为参考,可将集群内不同位置不同变化幅值的应变时程曲线转化为仅反应结构刚度变化的统一标准化指标,特别适用于长距离梁桥集群工程的结构损伤快速定位。上述研究探索了融合多桥监测数据进行结构损伤定位的思路,对于提高运营环境下梁桥集群结构损伤定位的准确性及精度,实现中小跨径梁桥集群结构损伤定位的实际工程应用具有一定的理论意义与实用价值。更多还原

【Abstract】 The beam bridges monitored within one cluster refer to several medium-and small-span beam bridges located in highway or urban elevated corridor.The structural health monitoring data of bridge under operation environment are the comprehensive response of various coupling effects such as vehicle load,temperature load,shrinkage and creep.It is usually difficult to eliminate the influence of complex coupling factors on damage localization of bridge by using only structural monitoring data from a single bridge.Consequently,it is challenged to localize the structural damage in all bridges within one cluster under operating environment.To address this issue,the following three types of bridges within one cluster are taken as research objects in this study,i.e.,the beam bridges with same structure,the beam bridges with similar structure and the beam bridges without health monitoring data.Based on the characteristics that the beam bridges within one cluster bear similar environment and vehicle load,the damage localization methods of medium-and small-span beam bridges within one cluster under the operation environment was systematically studied by fusing the monitoring data from multiple bridges.The main contents of this study are described as following.Aiming at the issue of damage localization of beam bridges with same structure under time-varying environment,a probabilistic SDDLV method for localizing damage in same beam bridges monitored within one cluster is proposed.Fitst,the limitation of traditional SDDLV method for damage localization of a single bridge is discussed,and the damage localization vector based on the structural characteristics of beam bridges within one cluster is established.Then,considering the influence of time-varying temperature,a method for calculating structural stress field based on Gaussian mixture clustering and probabilistic finite element model is studied.Finally,the damage localization threshold based on the probability characteristics of damage localization index is calculated,and then by incorporating hypothesis testing and a cross-validation strategy,the structural damage of all monitored same bridges within one cluster is localized.Compared with the empirical damage localization threshold of the traditional SDDLV method,the proposed method completes the damage localization in the whole area of the superstructure of beam bridges by using only sparse measuring points under the time-varying environment.Based on the above research,to avoid modal parameter identification error,a damage localization method for same bridges within one cluster using the difference ratio of projected strain monitoring data is proposed from the perspective of strain time domain analysis.First,a damage diagnosis feature is established by using the difference ratio of projected strain monitoring data obtained from the key stress sections of same beam bridges monitored within one cluster.On this basis,the relationship between the statistical characteristics of the proposed damage diagnosis feature and the degree of structural similarity between two bridges are discussed in detail,and the applicable conditions of the algorithm is explored.Then,a damage localization index is presented by calculating the subspace angle between two damage features.Finally,combined with kernel density estimation and a cross-validation strategy,the proposed index is implemented to localize the damage in the key stress sections of all same bridges monitored within one cluster.Compared with the traditional damage localization methods for a single bridge,this method can eliminate the influence of operation environment on damage localization accuracy more effectively.Aiming at the issue of damage localization of beam bridges with similar structure under time-varying environment,a damage localization method for similar beam bridges monitored within one cluster is proposed based on a spatiotemporal correlation model of strain monitoring data.First,a deep learning architecture combining a convolutional neural network with a long short-term memory network is established,which can reveal the complex time-varying mapping relationship between the strain monitoring data for similar bridges within one cluster to obtain an accurate spatiotemporal correlation model.Then,a strain prediction framework is presented that uses the proposed spatiotemporal correlation model after training.On this basis,the predicted and measured strains can be utilized to calculate a damage localization index that is not affected by complex coupling factors.Finally,the proposed index is implemented to accurately localize damage in all similar bridges within one cluster.Aiming at the issue of damage fast localization of existing beam bridges with similar structure,a damage fast localization method is proposed for similar beam bridges monitored within one cluster without reference model.First,an area equation of the strain time-history curve is derived theoretically under a moving load for a simple beam bridge and a continuous beam bridge.Then,a damage localization index is established based on the area-ratio of the strain time-history curve by analyzing the constitutive characteristics of the area equation of the strain time-history curve at each measurement point.On this basis,the relationship between the normalized damage localization index and the damage of girder is discussed.Finally,damage fast localization for beam bridges within one cluster is realized under temporary closed traffic.The proposed method does not need to reference historical health data or a finite element model and can transform strain time-history curves with different amplitudes from different positions along bridges into a unified normalized index reflecting only structural stiffness changes,which is especially suitable for structural damage localization of long-distance bridges within one cluster.This study explored the idea of structural damage localization by integrating monitoring data from multiple bridges,which has good theoretical significance and practical value for improving the accuracy of damage localization of beam bridges under the operation environment and realizing the practical engineering application for damage localization of medium-and small-span beam bridges within one cluster.更多还原