有限观测绝对加速度响应下剪切框架在未知地震作用下损伤诊断

针对框架结构的绝对加速度响应部分观测、地震作用未观测的情况,提出一种多层剪切框架结构诊断的方法。该方法先依次采用扩展卡尔曼估计和递推最小二乘对一层以上结构的扩展状态向量和未知作用力进行递推;然后利用结构频率特征方程,对第一层的结构参数进行估计;最后基于数值求解一阶微分方程,识别未观测的地震荷载。算例表明, 该方法能够很好识别出结构参数和地震输入,通过跟踪结构刚度参数的退化,对未知地震作用下结构损伤进行诊断。     

基于子结构虚拟变形的损伤识别方法

针对大型土木结构损伤识别优化效率低的问题,提出了子结构虚拟变形方法。虚拟变形方法是一种结构快速重分析的方法,该方法利用单元的虚拟变形模拟结构的损伤,可以在不重新建立有限元模型的情况下,快速计算出结构参数改变后的结构响应。该文基于虚拟变形法的基本思想,对子结构的刚度矩阵进行分解和对损伤后结构运动方程进行整理,推导出利用子结构的虚拟变形刻画损伤的方法,扩展了虚拟变形方法的适用范围;并且给出了虚拟变形和结构响应的相关性计算公式,通过相关性分析提取主要的虚拟变形,减少参与计算的子结构虚拟变形的数目,提高计算效率;最后利用一个五十层框架的数值仿真验证方法的有效性。     

未知激励下的无迹卡尔曼滤波新方法

无迹卡尔曼滤波（UKF）是一种识别非线性系统的有效方法,然而传统的UKF方法需要观测外部激励,这限制了UKF的应用范围。迄今为止,国内外对未知激励情况下的UKF方法的研究还非常少。该文在传统UKF的基础上,推导出在未知激励情况下的无迹卡尔曼滤波（UKF-UI）方法的递推公式,通过对观测误差的最小化,利用非线性方程求解,识别未知外部激励,进而识别非线性结构系统状态与结构未知参数。进一步采用融合部分观测的加速度响应及位移响应,消除识别结果的漂移问题。分别通过白噪声和未知地震作用下识别非线性迟滞模型的两个数值算例,考虑观测噪声对非线性系统进行识别,从而验证提出新方法的有效性。结果表明,该文所提出的UKF-UI方法,能够在部分观测结构系统响应的情况下,有效地识别非线性结构参数和未知激励。     

有限观测下的结构损伤实时在线诊断

根据有限观测的结构动力响应,实时对结构发生损伤的时间、位置与损伤程度进行诊断,是目前土木工程健康监测的一个重要的任务。提出一种在对结构加速度部分观测的情况下,对结构刚度发生突变的结构损伤形式进行实时的追踪和诊断的算法。该算法的主要思想是首先基于扩展卡尔曼滤波识别结构刚度参数,然后通过观测加速度和计算预测的加速度误差平方和,对结构刚度发生突变时刻进行判断,再通过优化突变时的结构的刚度参数变化,对损伤位置与损伤程度进行识别。数值模拟和实验的结果都表明了,该方法能够有效地追踪时变参数,并实时诊断出损伤发生的时间、位置和程度。     

基于物理参数贝叶斯更新的桥梁剩余强度估计研究

考虑系统参数的随机性,将基于广义卡尔曼滤波的子结构法与贝叶斯更新方法相结合,提出了桥梁结构基于贝叶斯更新物理参数的剩余强度估计两步法:第一步,将子结构法与广义卡尔曼滤波算法相结合,成功识别出子结构及其相邻单元的物理参数;第二步,视识别出的结构物理参数为更新信息,对以蒙特卡罗仿真实验结果作为先验分布的参数进行贝叶斯更新并分别基于蒙特卡罗仿真参数和贝叶斯更新物理参数对结构进行了剩余强度估计。数值算例表明:基于贝叶斯更新物理参数估计得到的结构剩余强度明显低于基于蒙特卡罗仿真参数估计得到的结构剩余强度。该方法为测量响应信息不完备条件以及小样本抽样情况下桥梁结构剩余强度估计提供了一个较好的解决思路。     

一种适用于大型工程结构的分散振动控制方法

基于瞬时最优控制算法,提出一种适用于大型结构振动控制的分散控制方法。根据大型结构的有限元模型,将其划分为若干子结构,相邻子结构间的作用力视为对子结构的"附加未知扰动"。在子结构加速度响应部分观测情况下,依次运用卡尔曼预测估计和最小二乘估计法,分别对子结构的状态和子结构间的作用力进行估计,并在此基础上基于瞬时最优控制算法确定各子结构的最优控制力,继而施加于结构对其进行振动控制。将提出的分散振动控制方法应用于美国土木工程师协会(ASCE)提出的20层控制基准问题(benchmark problem),并与传统的集中振动控制方法进行相关振动控制指标的比较。结果表明提出的分散振动控制方法能取得满意的振动控制效果。因此,提出的分散振动控制方法适合应用于对大型结构的振动控制。     

互谱能量法在激励力未知条件下对结构的诊伤研究

众多结构损伤诊断的方法都是基于结构有限元模型，或者是基于结构激励力已知条件下进行的系统参数识别技术。然而，由于大型结构物激励力不易测试，这些方法应用于土木工程结构物存在着一定的局限性。因此。发展基于环境激励下抗噪声性能较好的结构损伤诊断方法对于土木工程结构物安全监测显得尤为重要。提出一个新的互谱能量法，实现了在环境激励力未知条件下对结构进行损伤诊断。互谱能量法诊断结构损伤无需已知结构本身的材料特性，无需系统参数识别，不是基于结构的有限元模型，仅基于结构动力可测试部分输出响应信号以及结构单元拓扑分布即可实现结构的损伤诊断。为了验证所提互谱能量法的有效性，数值模拟两跨连续梁结构，仅仅利用有限的动力测试信号，并且结构测试信号中加入了较大的噪声模拟实际的测试环境，研究结果表明互谱能量法能够有效地诊断出结构单一，多损伤位置的损伤，抗噪声能力较强。     

桥梁结构物理参数识别的双单元子结构法

为了解决大型桥梁结构局部区域的健康监测问题,提出了一种用于桥梁结构物理参数识别的双单元子结构法。该方法是一种基于有限元的时域系统识别方法,其一次仅能识别出两个相邻单元的物理参数。将桥梁结构划分为若干个单元,选取任意两相邻单元为研究对象,建立其状态方程和观测方程。应用广义卡尔曼滤波,可得到该两相邻单元的质量、刚度和阻尼等物理参数。以一座三跨连续梁桥为例进行仿真分析,识别出不同信噪比情况下其子结构的物理参数。结果表明参数识别精度高,收敛速度快,从而验证了方法的有效性。     

一种未知激励下土木工程结构模态参数识别新方法

未知激励下的土木工程结构响应信号通常是随机的且噪声水平较高,因此对其进行参数识别具有挑战性。从未知激励下的振动响应信号出发,结合随机减量技术、解析模态分解、希尔伯特变换和卡尔曼滤波理论提出一种新的未知激励下土木工程结构模态参数识别新方法。该方法首先采用随机减量技术将实测的振动响应信号转换成自由振动响应信号;其次,运用解析模态分解理论将转换后的自由振动响应信号分解成各阶独立的模态分量信号;最后,采用希尔伯特变换估计出各阶分量信号的固有频率和模态阻尼比。然后运用卡尔曼滤波算法对估算出的频率和阻尼比进行滤波和平滑以得到更精确的识别值。通过一个含有密集模态分量的合成信号和一个未知激励作用下4层钢框架结构试验验证了该方法的有效性,研究结果表明:该方法在未知激励情况下仍然能够准确有效识别结构固有频率和阻尼比。     

约束子结构损伤识别的时序方法研究与试验

针对大型复杂结构的整体监测常常面临测量信息不足等困难,提出只利用局部动态响应进行子结构损伤识别的约束子结构方法。约束子结构方法是通过子结构响应的卷积组合限制子结构边界的响应为零,来实现施加虚拟支座,从而将子结构分离出整体,然后利用构造的相应子结构内部响应,进行子结构损伤识别。该文利用先分段提取结构响应的子时间序列,再延时排列Toeplitz矩阵的方式,使基于不同响应的构造约束子结构的方程具有相同表达式,统一了约束子结构方法的基本思想。通过测量悬臂梁的局部动力响应,利用局部响应的时间序列实现了子结构的快速准确地分离和识别,验证了方法的实用性和有效性。     

输入未知条件下基于自适应广义卡尔曼滤波的结构损伤识别

发展一种输入未知条件下的自适应广义卡尔曼滤波(Adaptive Extended Kalman Filter with Unknown Inputs,AEKF-UI)方法,在线复合反演系统参数与未知输入,结合基于改进粒子群优化算法的自适应技术实现系统时变参数追踪,进而识别结构损伤,包括损伤发生的时间、位置和程度。建立基础隔震结构实验模型及理论模型,其中隔震层的非线性动力学特性通过Bouc-Wen模型描述。对基础隔震结构进行振动实验研究,采用刚度元件装置模拟时间、位置和程度不同的结构损伤,基于测得的加速度响应和AEKF-UI方法进行实时系统参数与未知输入的同步反演。研究结果表明:在两种典型地震波激励下,AEKF-UI方法得到的识别值与参考值相一致,验证了该方法在系统辨识中的有效性和准确性。     

结构物理参数时域识别的子结构方法研究

研究了输入、输出信息皆不完备情况下的结构参数识别以及荷载反演问题。阐述了一种通用的子结构动力方程及其参数识别方程建立的基本原理和方法,并针对实际工程检测中子结构参数识别方程的输入特性,分别采用一种与之相适应的分解反演算法或统计平均算法。子结构技术与分解算法或统计平均算法的有效结合,为有限测点条件下的结构参数识别及荷载反演问题提供了一个较好的解决方案。大量的数值计算结果表明,本文提出的方法具有很好的参数识别精度及荷载反演效果。     

Health Monitoring of Aerospace Structures Using Fibre Bragg Gratings Combined with Advanced Signal Processing and Pattern Recognition Techniques

Abstract: The main purpose of this work is to develop an innovative system for structural health monitoring of aerospace composite structures based on real‐time dynamic strain measurements. The dynamic response of a composite panel, which represents a section of a typical aeronautical structure, is measured using fibre Bragg grating (FBG) dynamic sensors. Damage is simulated by slightly varying locally the mass of the panel at different zones of the structure. A finite element model of the structure has been developed to simulate the dynamic behaviour based on the modal superposition principle. The numerical model was calibrated against experimental results, and it was used for the placement of the FBG sensors. The proposed damage detection algorithm utilises the collected dynamic response data, and through various levels of data processing, an artificial neural network identifies the damage size and location. Feature extraction is the first step of the algorithm. Novel digital signal processing techniques, such as the wavelet transform, are used for feature extraction. The extracted features are effective indices of damage location and its extension. The classification step comprises a feed‐forward back propagation network, whose output determines the simulated damage location. Finally, dedicated training and validation activities are carried out by means of numerical simulations and experimental procedures.     

基于有限元模型修正的土木结构损伤识别方法

基于有限元模型修正的结构损伤识别方法计算过程直观、物理意义明确,能同步识别结构损伤位置和损伤程度,是结构损伤评估最直接的依据。该文介绍了基于有限元模型修正的损伤识别方法基本原理与过程,总结近三十年来有限元模型修正技术的发展。土木工程结构模型复杂性(包含大量单元、节点、待修正参数等)导致有限元模型修正过程效率低,详细介绍了基于子结构有限元模型修正的土木工程损伤识别方法在提高计算效率方面的优势。将两种有限元模型修正方法应用于一栋超高层建筑数值模型的损伤识别,说明两类方法在土木结构损伤识别中的特点。土木工程尺度大而结构损伤通常发生在局部区域,子结构方法将整体结构的分析转换为对若干独立子结构的分析,通过少数局部子结构的模型修正实现损伤识别,避免重复分析大尺寸整体结构,为大型结构基于有限元模型修正的损伤识别开辟高精度和高效率的途径。     

A statistical average algorithm for the dynamic compound inverse problem

 A new method named as Statistical Average Algorithm is developed for solving the structural dynamic compound inverse problem, which means to identify structural parameters with unknown input or to inverse input time history with unknown structural parameters. By taking the mechanical characteristic of the ground motion as an additional condition, an iterative algorithm based on the least-squares technique is developed so that the input process and structural parameters can be correctly determined using only output measurements. The procedure of the proposed method is discussed in detail, and a mathematical proof is presented as well in the appendix. A practical input situation such as earthquake, ambient vibration is considered in the numerical examples to verify the accuracy, reliability and robustness of the proposed algorithm. Considering the background of the practical application, both of the noise-free and noise-included output responses are considered in the numerical examples. In all cases, the proposed method identifies the structural parameters and reconstructs the input process rationally. Received: 12 March 2001 / Accepted: 16 September 2002     

用BP神经网络诊断结构破损

提出了一种基于BP神经网络的结构破损诊断方法,该方法以结构残余力向量作为破损诊断的网络输入。对网络训练样本采用广义空间格点法进行了变换,从而较好地解决了由于系统响应样本在数据空间分布不均对网络收敛速度及网络诊断精度的影响问题。应用实例表明,本文方法能准确诊断结构破损位置与严重程度,是一种有效的结构破损诊断方法。     

高层建筑损伤识别中的补偿算法

本文在已有研究工作的基础上，提出了处理激励测量信息不完备情况下的补偿算法。此方法主要针对高层建筑动力检测中强迫激励的情况，是一种基于时域的识别方法，该方法实际上是利用客观工程实际中结构系统激励所具有的物理持征作为识别计算的辅助条件，来解决测量信息的不完备性问题。算例分析表明了该方法的可行性。     

An experimental study of the feasibility of phase-based video magnification for damage detection and localisation in operational deflection shapes

Optical measurements from high-speed, high-definition video recordings can be used to define the full-field dynamics of a structure. By comparing the dynamic responses resulting from both damaged and undamaged elements, structural health monitoring can be carried out, similarly as with mounted transducers. Unlike the physical sensors, which provide point-wise measurements and a limited number of output channels, high-quality video recording allows very spatially dense information. Moreover, video acquisition is a noncontact technique. This guarantees that any anomaly in the dynamic behaviour can be more easily correlated to damage and not to added mass or stiffness due to the installed sensors. However, in real-life scenarios, the vibrations due to environmental input are often so small that they are indistinguishable from measurement noise if conventional image-based techniques are applied. In order to improve the signal-to-noise ratio in low-amplitude measurements, phase-based motion magnification has been recently proposed. This study intends to show that model-based structural health monitoring can be performed on modal data and time histories processed with phase-based motion magnification, whereas unamplified vibrations would be too small for being successfully exploited. All the experiments were performed on a multidamaged box beam with different damage sizes and angles.     

Neural networks to model dynamic systems with time delays

An algorithm is proposed to identify a neural network model that represents a nonlinear dynamic system with a multivariate time delay response. The algorithm consists of two major parts. The first one identifies the time delay vector for a given neural network structure. This task is accomplished by using an exhaustive integer enumeration algorithm that minimizes a statistical parameter to assess the performance of the neural network model. The second part uses a cross-validation strategy to identify the best neural network model. Since the structure that models a nonlinear system is usually unknown, the identification strategy consists of selecting several neural network structures and identifying the best time delay vector for each network. The modeling process starts with the simplest structure and progressively the complexity of the network is increased to end up with a complex structure. Finally, the network that offers the simplest structure with the best network performance is the one that exhibits the appropriate neural network structure with the corresponding optimal time delay vector. The Monte Carlo simulation technique was used to test the performance of the algorithm under the presence of linear and nonlinear relationships among several variables of dynamic systems and with a different time delay applied to each input variable. The introduced algorithm is used to detect a chemical reaction delay among enriched amyl acetate, acetic acid, water, and the pH of erythromycin sail. An appropriate neural network model was designed to model the pH of the erythromycin during a continuous extraction process. To the best of the authors knowledge the proposed algorithm is the only one currently available to identify time delay interactions in the multivariate input output variables of a system. The major drawback of the introduced algorithm is that it becomes very slow as the number of system inputs increases. This algorithm works efficiently in a system that involves five inputs or less.