基于单测点信号的桥梁损伤诊断与定位的混沌振子及重构相空间方法研究

The number of constructed bridges in China increases rapidly in the several decades. However, the overall situation of them is not reassuring since the number of dangerous bridges kept a high level in the past years. Current damage detection methods have many problems and challenges such as numerous measurement points, huge cost, uncertainty of ambient excitation, completeness and inaccuracy of measured data, et al. So, this project proposes a new method using chaotic attractor and reconstructed phase space with only one measurement point to try to solve part of these problems. Firstly, using the sensitivity of the critical state of chaotic oscillator, a choatic damage detection model will be constructed to diagnose the damage existence. Then, the method that using a single measutement point response to locate the damage of bridge under a moving traffic load will be studied. The more actual vehicle-bridge coupled system mechanical model will be constructed and the vibration mathematical model of the system will be derived, so that the vibration equation of the single point can be obtained. Using the response of the single point, the damage location detection model will be constructed based on the reconstructed phase space. Then the equation or empirical equation of bridge physical parameters, damage degree and location, moving load magnitude and velocity will also be obtained, realizing to diagnose and locate the bridge damage with single measuring point. The project plans to research combining theoretical analysis, numerical simulation and experimental study to develop a new, reliable, and applicable to the real engineering method in bridge health monitoring and damage detection.

我国桥梁飞速发展，但总体现状不容乐观，危桥数量呈上升趋势。现有损伤检测方法在实际应用中存在诸多问题：测点繁多、耗资巨大，环境激励的不确定性，实测数据不完备性和精确性等。本项目提出基于单测点信号的桥梁损伤诊断与定位的混沌振子及重构相空间方法有助于解决上述部分问题。首先利用混沌振子临界状态的灵敏性构建混沌检测模型来诊断损伤的存在，然后进行桥梁在移动交通荷载作用下用单测点信号进行损伤定位研究。构建更接近真实的车桥耦合力学模型，推导损伤桥梁在移动荷载作用下的振动数学模型，得到单点（测点）随位置变化的振动方程，并构建基于单测点响应的重构相空间损伤定位模型，以期得到与桥梁物理参数、损伤位置和程度、移动荷载大小和速度等的函数关系式或经验公式，实现单测点诊断和定位桥梁损伤。拟从理论推导、数值计算和仿真、实验研究三者相互补充验证的研究思路，发展一种新型、可靠、适用于实际工程的桥梁结构健康监测和损伤检测方法。

本项目自实施以来，发表高水平论文7篇，其中SCI论文3篇，1篇为土木与结构领域影响因子排名第一的国际期刊论文，影响因子5.79。另一篇发表在机械与结构领域顶级期刊（MSSP），影响因子4.03。EI论文2篇，国际会议论文2篇。申请国家发明专利2项，获批软件著作权1项。自实施以来，超额完成项目目标。. 当前桥梁健康监测系统受到工程界质疑，没有大面积推广和产业化的两个个重要原因是：（1）监测系统传感器过多、耗资巨大；（2）系统所测得海量数据难以处理。. 本项目针对这两个问题，展开了以下研究：.1、针对当前监测系统传感器过多、耗资巨大问题：. 1）研究了桥梁损伤检测的单传感器方法，本方法利用安装在桥梁上单测点信号的重构相空间拓扑结构的变化进行损伤定位，降低健康监测系统的成本；. 2）桥梁结构损伤定位的双传感器联合识别法，同样能够大大降低系统传感器的用量。. 3）研究发明一种基于移动窗函数的信号分离与去噪方法及装置，通过分离结构的振动响应信号，得到检测者感兴趣的分量信息。.2、针对当前系统所测数据海量难以处理问题：. 1）大数据的深度学习方法：提出利用大数据的深度学习自动特征提取的结构损伤监测方法，提供了一种新的解决该问题的框架——大数据思想，即直接分析数据得到结果的相关关系，而不去分析其因果关系。. 2）数据驱动方法：提出了移动主成分分析的数据驱动方法分析桥梁长期监测数据，评估桥梁状态。. 3）利用环境激励下的结构加速度响应二次协方差CoS（Covariance of Strain response）矩阵参数变化比法，对结构进行损伤定位。. 研究成果成功完成了成果转化，开发了桥梁健康检测系统，并成功应用到黄埔大桥，虎门大桥和九江大桥，并得到业界的高度认可。