影像轮廓线叠差分析获取桥梁全息变形及结构状态演绎方法研究

Two common methods of bridge safety monitoring at home and abroad are regular manual safety inspection and on-line monitoring of contact sensor. The regular manual safety inspection lacks quantitative scientific basis. The on-line monitoring of contact sensor has many problems, such as high cost, difficult maintenance, short life and large noise. The loading test of large scale steel truss concrete composite beam under different damage conditions is carried out to research high-precision image stitching algorithm of bridge under uncertain photography. The mapping relation between the gray level space of image and the potential space of data field of structural image contour is constructed; The distribution, noise and vague edges characteristics of coupled data fields in the feature space of structural image contour are analyzed to obtain the probabilistic edge from uncertain edge by morphological processing; Based on the differential analysis of the bridge image contour which is successively collected, the method of obtaining the corresponding holographic deformation is proposed; Based on the known bridge holographic deformation, the analytical theory about structural state, damage location and degree is established; According to the uncertain input variables of actual bridge, a multidimensional database of holographic deformation which contains environmental, live load, material properties and other uncertain factors was preliminarily constructed; The structural state deduction method which is based on grey input and the structural identification method of abnormal deformation position and degree which is based on large data are explored. Finally, the purpose of this project is to break through the bridge safety monitoring technology and theory which is economic and effective.

采用经常性人工安全巡检和接触式传感器在线监测是目前国内外桥梁安全监测的两种常用手段，前者缺乏量化的科学依据，后者存在费用高、维护难、噪声大、信息不全等问题。本项目拟通过多种损伤状况下大尺度钢桁-砼组合梁的加载试验，研究不定性摄影下桥梁影像的高精度拼接算法，构建图像灰度值空间至结构影像轮廓线数据场势值空间的映射关系；分析结构影像轮廓线特征空间耦合数据场的分布、噪声及模糊边缘的特性，对不确定性边缘进行形态学处理获得概率边缘，提出依据先后采集桥梁影像轮廓线的叠差分析获得全息变形的方法;建立基于已知桥梁全息变形反演结构状态，识别损伤部位及程度的分析理论；针对实际桥梁作用输入的不定性，初步构建包含环境、活载、材料性能等不定性因素的全息变形多维数据库框架；探索基于灰色作用输入下的结构状态演绎方法和基于大数据视角的结构异常变形部位及程度识别方法。实现较高频次经济有效的桥梁日常安全监测技术和理论的新突破。

基于机器视觉对桥梁结构服役状态进行安全监测，较传统人工安全巡检方法和接触式传感器监测系统具有全息、方便、经济的突出优势，并高度契合未来数字信息的发展趋势。本项目针对高噪声及复杂光照条件下实际桥梁图像的高精度融合算法开展理论研究，建立了与桥梁图像尺度不变特征变换相适应的噪声信号分布模型，根据噪声信号分布模型和尺度空间内极值点特征，提出了桥梁图像的配准和融合算法；研发了针对桥梁结构监/检测的定轴旋转采集图像硬件系统，推导了投影中心偏移引起桥梁图像像点坐标偏差方程，揭示了投影中心偏移对桥梁图像像点坐标和图像拼接精度的作用机理，提出依据先后采集桥梁影像轮廓线的叠差分析获得全息变形的方法；试验论证了开裂损伤会导致结构相应部位位移矢量转角发生突变，提出利用位移矢量转角变化率识别结构的损伤部位，建立了结构边缘裂缝两侧位移转角差与裂缝高度的理论联系；推导了基于曲率模态描述结构损伤部位和程度的数学表达式，建立了基于已知桥梁全息变形反演结构状态，识别损伤部位及程度的分析理论；基于物理、力学及结构的先验知识构建了融合物理模型与数据驱动的桥梁数字孪生模型，概率化地解释了驱动过程，初步实现了数字空间的结构孪生模型的动态更新。针对实际桥梁作用输入的不定性，初步构建了包含环境、活载、材料性能等不定性因素的全息变形多维数据库框架；用计算机视觉获得的结构全场密集点测值，丰富和拓展了常规SHM监测中结构形态测试状态向量，据此推导了计算机视觉架构下在物理空间与模态空间的结构动力学测试方程；提出概率论与信息论框架下的结构信息全息度配置方法。探索了基于灰色作用输入下的结构状态演绎方法和基于大数据视角的结构异常变形部位及程度识别方法。实现了较高频次经济有效的桥梁日常安全监测技术和理论的新突破。共发表论文33篇，其中SCI收录6篇；申报发明专利13项；获得重庆市科技进步一等奖和全国博士后创新创业大赛金奖各一项。

内容获取失败，请点击重试