基于视觉与加速度测量的结构动态位移融合估计  

Fusion Estimation of Structural Dynamic Displacement Based on Visionand Acceleration-Based Measurements

作　　者：熊春宝[1] 孙长保 牛彦波[1,3] Xiong Chunbao;Sun Changbao;Niu Yanbo(School of Civil Engineering,Tianjin University,Tianjin 300072,China;China Oilfield Services Limited,Tianjin 300450,China;College of Civil Engineering and Architecture,Zhejiang University,Hangzhou 310058,China)

机构地区：[1]天津大学建筑工程学院,天津300072 [2]中海油田服务股份有限公司,天津300450 [3]浙江大学建筑工程学院,杭州310058

出　　处：《天津大学学报（自然科学与工程技术版）》2024年第9期891-901,共11页Journal of Tianjin University：Science and Technology

基　　金：国家重点研发计划资助项目(2023YFC3009301);中国博士后科学基金资助项目(2021M692835);中国-巴基斯坦重大基础设施智慧防灾一带一路联合实验室开放基金资助项目(2023CPBRJL-05).

摘　　要：结构动态位移测量与精准估计对于结构安全运营和性态评估具有重要意义.基于计算机视觉的位移监测方法具有精度高、非接触式、成本低、设备安装简便等优点,在实际复杂工程环境中,设备难以架设,视觉测量方法较传统接触式位移监测方法具有明显的优势.图像分辨率和拍摄帧率等因素在一定程度上限制了视觉方法的使用.针对视觉位移测量技术高频振动识别精度低的问题,提出了一种基于视觉与加速度测量的结构动态位移重构方法,通过融合视觉低频与加速度高频振动响应信号,实现结构动态位移精准识别.首先,利用光流法从结构振动视频数据中提取结构位移响应,引入前后向误差与离群值过滤机制,提升特征点追踪精度,避免漂移问题.然后,利用逐次变分模态分解方法分别从视觉位移与加速度二次积分得到的位移信号中提取相应的本征模态函数(intrinsic mode function,IMF)分量.最后,基于互相关函数筛选机制,确定融合模态分量,融合基于视觉测量的低阶IMF与基于加速度测量的高阶IMF,重构结构位移响应.通过一个钢筋混凝土框架结构振动台试验,对提出的位移融合估计方法进行了试验验证.结果表明:与单一视觉测量方法相比,所提出的方法能够更为准确地估计结构动态位移,并且通过引入加速度测量中的动态位移分量,融合后的位移比基于视觉测量的结果具有更宽的频率范围.The measurement and accurate estimation of structural dynamic displacements are of significance for the safe operation and performance evaluation of structures.The computer vision-based displacement monitoring approaches have advantages such as a high accuracy,non-contact,a low cost and easy installation,and they are superior to the traditional contact-type displacement monitoring methods in a complicated engineering environment in practice,where the equipment is difficult to install.However,the use of vision-based methods is limited by factors including image resolution and shooting frame rate to some extent.In this paper,to address the issue of low accuracy in high-frequency vibration recognition using the vision-based displacement measurement technique,a structural dynamic displacement reconstruction approach based on vision-and acceleration-based measurements is put forward by fusing the vision-based low-frequency and acceleration-based high-frequency vibration response signals,thus realizing an accurate recognition of structural dynamic displacements.First,an optical flow method is adopted to extract the structural displacement responses from the video data of structural vibration,and a forward-and-backward error and an outlier filtering mechanism are introduced to improve the feature point tracking accuracy and avoid the problem of drift.Then,a successive variational mode decomposition approach is used to extract the corresponding intrinsic mode functions(IMFs)from displacement signals obtained by quadratic integration of the vision-and acceleration-based displacements,respectively.Finally,the fused modal components are determined according to a crosscorrelation function filtering mechanism,and the structural displacement responses are reconstructed by fusing the vision-based low-order and acceleration-based high-order IMFs.A shaking table test was conducted on a reinforced concrete frame structure to experimentally verify the proposed fusion estimation approach for displacements,and results indicate that t

关 键 词：数据融合 计算机视觉 光流法 逐次变分模态分解 互相关函数 

分 类 号：TU196.4[建筑科学—建筑理论]