基于非接触远程智能感知的桥梁形态监测试验

为进一步提高利用非接触式影像监测大型桥梁安全状态的精度与效率，实现结构健康监测系统兼有经济、可信且全息的技术理论优势，提出依据结构全息影像序列数据的非接触式机器视觉远程智能感知进行桥梁结构全息几何形态监测的方法。通过单机自动巡转桥梁立面动静影像全息监测系统试验装置获取试验桥在各损伤/作用工况下的原始动静影像数据，根据序列数据在时间、空间上具有强关联信息的特性，分别构建降噪及抗扰动单元、欧拉运动放大单元及运动信息提取单元进行全息几何形态测量，以历次试验监测数据为样本数据集，利用分层思想依次建立数据抽样和结构几何信息间映射的数学网络模型，经结构设计不断训练、调试与优化桥梁力学行为智能感知网络，最后获取试验桥在试验过程中的全息变形、全息变形包络谱、全息位移时程曲线。研究结果表明：该方法获得的数据与传统的常规接触式传感器实测值基本吻合，试验桥在各工况下的曲线变化趋势基本一致，全息变形测量值平均误差12.21%，全息变形包络谱测量值平均误差9.06%，全息位移时程曲线测量值平均误差8.55%，对环境规律噪声信号筛除效率为81.9%，基于非接触远程智能感知的桥梁形态监测真实、连续、敏感、较为准确地反映了结构在各损伤/作用工况下的真实形态变化，可为后续进一步研究结构状态演绎以及损伤智能化识别方法奠定基础。

Experiment of Structural Morphology Monitoring for Bridges Based on Non-contact Remote Intelligent Perception Method

In order to further improve the precision and efficiency of the safety state technology and theory of large bridges, non-contact video monitoring is expected to realize a new breakthrough in structural safety state monitoring with economic, credible, high frequency, and holographic advantages. Based on the holographic series data, a non-contact remote intelligent perception method was proposed to monitor the bridge holographic geometric shape. Using a single automatic camera patrol experimental device, original segmental dynamic and static video monitoring data of a model bridge under various damage/activities were collected. According to the temporal and spatial characteristics of the series data, de-noising, and disturbance-rejection units, the amplifying unit of Eulerian motion and the extracting unit of motion information were respectively constructed for the holographic measurement of geometric shape. A mathematical network model of data sampling and mapping between the structural geometric information was established using layering thought. The intelligent perception network of bridge mechanical behavior was trained, debugged, and optimized by structural design. Finally, the holographic deformation, envelope spectrum of holographic deformation, and time history curve of holographic displacement of the test bridge were obtained. The results show that the measured values of the test bridge are consistent with those of the conventional contact sensors, and the curve variation trend of the test bridge under various working conditions is essentially the same. The average error of the measured values of holographic deformation, holographic deformation envelope spectrum, holographic displacement time history curve, and screening efficiency of the environmental regularity noise signal are 12.21%, 9.06%, 8.55%, and 81.9%, respectively. Bridge shape monitoring based on non-contact remote intelligent perception is real, continuous, and sensitive, and accurately reflects the real shape changes of the structure under various damage/action conditions; this can lay a foundation for further research on structural state deductive models and intelligent damage identification methods.