基于数字图像技术振动台试验结构位移信息捕捉技术研究  

Capturing technology for displacement information of shaking table test structure based on digital image technique

作　　者：常铭宇 申玉生[1,2] 张熙[1,2] 高登 骆阳 王浩鱇 CHANG Mingyu;SHEN Yusheng;ZHANG Xi;GAO Deng;LUO Yang;WANG Haokang(MOE Key Lab of Transportation Tunnel Engineering,Southwest Jiaotong University,Chengdu 610031,China;National Engineering Research Center of Geological Disaster Prevention Technology in Land Transportation,Southwest Jiaotong University,Chengdu 610031,China)

机构地区：[1]西南交通大学交通隧道工程教育部重点实验室,成都610031 [2]西南交通大学陆地交通地质灾害防治技术国家工程研究中心,成都610031

出　　处：《振动与冲击》2024年第5期139-148,158,共11页Journal of Vibration and Shock

基　　金：国家自然科学基金项目(52278414);四川省自然科学基金项目(2022NSFSC0424)。

摘　　要：针对振动台试验模型箱位移、围岩错动以及隧道衬砌变形监测的需求,将YOLO-V2、改进Canny边缘检测、Hough变换及倾斜矫正技术相结合,提出了一种倾斜影像的自动矫正技术,实现了将倾斜影像修正为相机光轴垂直于测点运动平面的图像分析;提高了试验数据结果的处理精度,扩大了目标追踪算法在结构位移和变形监测的使用范围。同时通过核相关滤波技术完成对修正后的影像进行追踪,形成一套完整的地震动信号捕捉与监测系统。采用该系统进行了强震作用下穿越断层隧道韧性结构减震抗错性能试验研究,分别对模型箱位移、围岩错动和隧道衬砌变形进行信号识别与捕捉,同时对比分析了高精度拉线式位移计和位移传感器采集的数据,验证了系统的可靠性。结果表明:在监测点位设有标记物时,首先对采集结果进行倾斜矫正后追踪,通过监测结果与位移传感器进行对比,两者的2-范数小于0.005;而在无标记物时对结构表面特征物进行追踪,精度达到了97.5%;试验数据证明通过倾斜影像自动矫正技术来完成非对正对标记物时的影像进行追踪具有较高可靠性。Here,aiming at needs of monitoring displacement of shaking table test model box,surrounding rock dislocation and tunnel lining deformation monitoring,a tilt image automatic correction technology was proposed by combining YOLO-V2,improved Canny edge detection,Hough transform and tilt correction technology to realize image analysis by correcting tilt image to make camera’s optical axis perpendicular to motion plane of measurement point.Processing accuracy of test data results was improved,and application range of target tracking algorithm in structural displacement and deformation monitoring was expanded.At the same time,the corrected image was tracked using nuclear correlation filtering technology to form a complete system for capturing and monitoring seismic signals.This system was used to conduct test study on seismic reduction and anti-dislocation performance of tunnel ductile structures crossing faults under strong earthquakes.Signal recognition and capturing were performed for model box displacement,surrounding rock dislocation and tunnel lining deformation,respectively.Data collected with high-precision guyed displacement meter and displacement sensor were analyzed contrastively to verify the reliability of the system.The results showed that when markers are located at monitoring points,the collected results are tracked after tilt correction,by comparing monitoring results with those using displacement sensor,the 2-norm of the two is less than 0.005;when there are no markers,the tracking accuracy of structural surface features reaches 97.5%;using the automatic correction technology to track tilted images not aligning markers has higher reliability.

关 键 词：振动台试验 变形监测 CANNY边缘检测 倾斜影像自动修正 目标追踪 

分 类 号：TH212[机械工程—机械制造及自动化] TH213.3