THEORETICAL MODEL AND NUMERICAL METHOD ON ONLINE IDENTIFICATION OF DYNAMICAL CHARACTERISTICS OF STRUCTURAL SYSTEM （Department of Engineering Mechanics, Southeast University, Nanjing 210096, P.R. China）

An online method of identification of dynamic characteristics only using measured ambient response of structural dynamic system is widely focused on. The Ibrahim and ARMA ( AutoRegressive Moving Average ) methods are basic identification methods. A model on dynamic system suffered by random ambient excitation was researched into, and a subspace decomposition methodwas introduced. Robustness and effectiveness of this approach on identification of vibration characteristics are demonstrated on numerical experiment.     

Modal parameter identification of flexible spacecraft using the covariance-driven stochastic subspace identification (SSI-COV) method

In this paper, the on-orbit identification of modal parameters for a spacecraft is investigated. Firstly, the cou-pled dynamic equation of the system is established with the Lagrange method and the stochastic state-space model of the system is obtained. Then, the covariance-driven stochas-tic subspace identification (SSI-COV) algorithm is adopted to identify the modal parameters of the system. In this algo-rithm, it just needs the covariance of output data of the system under ambient excitation to construct a Toeplitz matrix, thus the system matrices are obtained by the singular value decom-position on the Toeplitz matrix and the modal parameters of the system can be found from the system matrices. Finally, numerical simulations are carried out to demonstrate the validity of the SSI-COV algorithm. Simulation results indi-cate that the SSI-COV algorithm is effective in identifying the modal parameters of the spacecraft only using the output data of the system under ambient excitation.     

Structural damping identification method using normal FRFs

All structures exhibit some form of damping, but despite a large literature on the damping, it still remains one of the least well-understood aspects of general vibration analysis. The synthesis of damping in structural systems and machines is extremely important if a model is to be used in predicting vibration levels, transient responses, transmissibility, decay times or other characteristics in design and analysis that are dominated by energy dissipation. In this paper, new structural damping identification method using normal frequency response functions (NFRFs) which are obtained experimentally is proposed and tested with the objective that the damped finite element model is able to predict the measured FRFs accurately. The proposed structural damping identification is a direct method. In the proposed method, normal FRFs are estimated from the complex FRFs, which are obtained experimentally of the structure. The estimated normal FRFs are subsequently used for identification of general structural damping. The effectiveness of the proposed structural damping identification method is demonstrated by two numerical simulated examples and one real experimental data. Firstly, a study is performed using a lumped mass system. The lumped mass system study is followed by case involving numerical simulation of fixed–fixed beam. The effect of coordinate incompleteness and robustness of method under presence of noise is investigated. The performance of the proposed structural damping identification method is investigated for cases of light, medium, heavily and non-proportional damped structures. The numerical studies are followed by a case involving actual measured data for the case of a cantilever beam structure. The results have shown that the proposed damping identification method can be used to derive an accurate general structural damping model of the system. This is illustrated by matching the damped identified FRFs with the experimentally obtained FRFs.     

反求压电薄板智能结构荷载的有限元逆逼近方法

基于压电薄板智能结构提出一种有限元逆逼近反求结构荷载参量的方法，逼近目标函数以压电电荷响应为参量，迭代初值采用单位荷载预逼近方法；数值算例表明了该方法的有效性，可用于解决压电智能结构的荷载识别问题。     

密集模态挠性结构模型多变量频域辨识和控制

密集模态挠性结构的模态不仅阻尼小而且耦合程度高,给模型辨识带来了很大困难.频域辨识是获取空间挠性结构模型的一个有效方法,但是目前频域辨识要么模型结构定义有缺陷,要么是集中在单变量情形,不适合作大型空间挠性结构多变量模型的辨识.本文提出了多变量极大似然频域辨识,给出了其模型结构定义、算法推导及实现,并将其应用于H型密集模态挠性板的模型辨识,根据辨识结果,设计主动控制律,实现了对密集模态挠性板振动的有效抑制,表明了辨识算法及主动控制的可行性.     

基于模态数据的长细结构小损伤精确识别研究

本文针对现有的损伤识别方法不能满足部分结构损伤识别精度要求的现状，对结构的小损伤精确识别方法开展研究．以长细结构为研究对象，对具有不同损伤位置和损伤程度的圆柱形的轻阻尼梁结构进行了数值分析和实验研究，应用数值计算方法和实验确定的特征向量和特征频率对长细结构裂缝参数进行识别计算．本文在研究过程中编制了一个创新性的预测程序，通过其一次性生成目标函数图来选择合适的初始参数，从而对识别结果进行分析．研究结果表明，应用本文提出的识别方法，裂缝位置的识别误差可以控制在0.05 %~0.28 %范围内，裂缝深度识别误差低于7 %．     

多自由度系统自然激励技术的统一模型

结构运行状态下的模态参数提取是结构健康监测系统需要解决的关键问题之一.自然激励技术的提出为大型复杂工程结构运行状态下的模态参数提取提供了一条新的途径.原始自然激励技术给出了单输入白噪声激励下利用结构位移响应的互相关函数进行模态识别的理论模型,对于多输入情况则缺乏相应的理论模型.本文在单输入理论模型的基础上进一步发展了自然激励技术:推导了多输入独立白噪声激励下多自由度系统结构位移响应的互相关函数的解析公式,并分析了它与单输入情况下互相关函数之间的关系;基于此互相关函数定义了一个新函数,证明它含有结构各阶模态信息,可以表达为一系列衰减正弦函数之和,并且各个组分正弦函数的频率等于各阶模态的有阻尼固有圆频率;提出了以新函数为核心的同时适用于单输入和多输入情况的模态识别算法,建立了自然激励技术的统一模型.     

Bending tests to estimate the axial force in slender beams with unknown boundary conditions

This paper presents a static method for the axial load identification of slender prismatic beams with uncertain length and unknown boundary conditions as is typical of struts and ties of truss structures or tie-rods of arches and vaults. The proposed method requires the knowledge of the beam flexural rigidity only. Flexural displacements or curvatures are measured at five cross sections of the beam subjected to an additional concentred lateral load. Unlike analogous dynamic methods, any set of experimental data may be used in the identification algorithm. The proposed algorithm is verified by means of many numerical and experimental tests on beams having different boundary conditions. Excellent estimates of the axial forces are obtained when the greatest possible distance between sensors is adopted, even if in the presence of high values of the axial forces very accurate displacement measurements are required.     

环境温度影响下基于支持向量机与强化飞蛾扑火优化算法的结构稀疏损伤识别

结构处于自然环境中常会受到环境温度变化的影响,引起实测动力响应出现较大误差,进一步影响对结构健康状况的判定.另外,基于优化算法的损伤识别在反演损伤位置及量化损伤程度时,易出现局部最优解,且计算效率低下.针对以上难题,本文提出一种结合支持向量机与强化飞蛾扑火优化算法的损伤识别方法,用于对环境温度影响下的结构稀疏损伤进行识...     

Global analysis and model validation in nonlinear system identification

The analysis of parameterised nonlinear models is considered. In particular the emphasis is placed on models produced as a result of applying nonlinear system identification techniques. A qualitative approach is taken, calling upon ideas from bifurcation theory. When combined with an existing numerical method this approach enables the analysis of a variety of nonlinear model forms to be carried out. The combination of qualitative and numerical aspects provides a flexible framework for analysis and in addition adds a global perspective that is difficult to achieve using analytical methods alone. When combined with the well tried methods of system identification, the approach allows the validation of nonlinear models from a qualitative viewpoint to be carried out. The method contrasts well with the statistical model validation techniques traditionally used. The technique is applied to a number of examples.     

部分输入未知时求解动力复合反演问题的补偿算法

将作者提出的全量补偿算法[8] 推广为一类适用于一般多自由度系统的时域补偿识别算法。该方法是基于最小二乘原则的一类迭代计算方法 ,对于结构上的部分作用力已知的情况 ,可用来同时识别结构的参数并反演输入。文中应用不同类型的结构分析实例说明了此方法适用于实际工程动力检测的可能性。     

基于L1正则化的随机梁式结构静力损伤识别方法

提出了一种结合摄动法和L1正则化方法的随机梁式结构静力损伤识别方法。考虑初始模型误差和测量误差的影响,建立了关于随机损伤指数的控制方程,并将摄动法和L1正则化方法相结合,对随机损伤指数的控制方程进行求解,进而从概率的角度对结构的损伤进行识别。损伤试验结果表明,和传统的最小二乘求解法相比,本文方法能够更为准确地识别多处局部损伤的位置及大小,对实际结构损伤检测具有较好的参考价值。     

随机梁式结构静力损伤识别的一种改进方法

针对已有的损伤识别方法会出现损伤误识别的问题,本文在已有方法的基础上发展了一种随机梁式结构静力损伤识别的改进方法。假定静力荷载下梁式结构初始模型参数（如弹性模量和几何尺寸等）及测量误差为随机量,给出已有的基于随机有限元模型的梁式结构静力损伤识别方法,并进一步提出了一种改进方法。该方法通过设定损伤概率指标的阈值和反复迭代对结构损伤识别进行改进。数值算例和简支梁静力试验表明,考虑初始模型的不确定性以及静力响应测量误差,本文方法相较已有方法可以更有效地识别梁式结构的损伤。     

弹性——粘弹性复合结构的动态载荷识别

提出了一种识别弹性－－粘弹性复合结构动态载荷频域方法。首先将复合结构利用约对值模态应变能方法（ＡＶＭＳＥ）进行模态分析而获得模态参数，其次在频域中通过模态变换法求得动态载荷谱。通过两个算例说明方法的应用，其结果表明该方法是有铲的，识别精度是好的。并对该方法应用中的一些问题进行了讨论。     

多种群遗传算法在PBX本构模型参数识别中的应用

利用多种群并行结构对标准遗传算法SGA进行并行化处理,引入移民算子和精华种群形成多种群遗传算法MPGA,并设计了自适应交叉和变异概率对算法的收敛速度进行改进。结合ABAQUS软件和改进的多种群遗传算法,建立了材料本构模型参数识别方法。采用该方法对PBX炸药黏弹性损伤本构模型参数进行了模拟识别,并同基于标准遗传算法的参数识别方法进行了比较。结果证明,基于改进多种群遗传算法IMPGA的方法对克服算法未成熟收敛有显著的效果,识别结果更稳定。同时该方法的收敛速度更快,寻优能力更强,适合复杂非线性问题的优化,此方法可以被应用到其他材料本构模型的参数识别中。     

基于吸收边界条件的有限元方法和离散切比雪夫展开的二维缺陷识别

结合考虑了吸收边界条件的有限元正演方法,通过离散切比雪夫多项式在特定离散点的展开逼近待识别函数,实现了无限区域内二维缺陷的识别。正问题计算中引入的吸收边界条件虽然不能严格地模拟无限域对近场波动的影响,但能够以满足实际需要的精度模拟这一影响,且具有解耦特性,极大地简化了数值计算,在反演计算中,离散点取在切比雪夫多项式的零点处,使逼近时最大偏差最小,数值算例表明此方法的有效性。     

环境温度影响下基于LSTM神经网络识别结构损伤

环境温度的改变会引起模态参数的变化,其变化程度会掩盖或部分掩盖损伤引起的变化量,导致结构健康监测系统发出假阳性或假阴性的误判,因此,消除温度效应是提高损伤识别精度的关键。本文基于LSTM神经网络提出了一种环境温度影响下识别结构损伤的方法。充分利用LSTM神经网络的非线性映射优势,建立多元温度-模态频率的相关模型,在此基础上采用数据标准化方法消除温度效应,并结合控制图判断模态频率异常变化以确定损伤状况。最后将所提方法在数值模型和实际桥梁中加以应用,结果表明,方法能够有效消除温度效应;结合控制图能识别损伤时刻,并具有一定的抗噪性;在实桥数据分析中仍能表现出较好的损伤敏感性。     

优化方法用于模型校核

在工程实际中，常常会遇到需要选择本构模型和校准参数的模型校核问题。文中采用优化方法实现了复杂模型校核的自动化。文中，首先将各种本构模型映射为整数编号，并根据这些编号来激活相应的本构模型和本构模型参数，解决了本构模型的参数化问题；另外，根据模型校核问题的特点选择了试验结果和计算结果相对误差的平方和作为目标函数。通过这些步骤，用实现了优化方法模型校核自动化。     

腐蚀环境下结构系统疲劳主要失效模式识别

疲劳主要失效模式识别是结构系统疲劳可靠性理论研究的主要内容之一。本文研究了腐蚀环境下结构系统疲劳主要失效模式识别方法:(1)建立了腐蚀环境下结构元部件损伤随机模型,对该模型的适用性作了讨论;(2)研究了识别结构系统疲劳主要失效模式的全局分枝约界法,与结构元部件损伤随机模型结合,实现腐蚀环境下结构系统疲劳主要失效模式识别,并对算法执行过程作了调整;(3)引入了一种安全裕量方程确定方法。理论分析和数值算例表明,本文所提算法可实现腐蚀环境下结构系统疲劳主要失效模式识别。     

基于改进云推理算法的塔架结构损伤识别

为了解决塔架结构的损伤识别问题,提出了基于应变能和改进云推理算法的损伤识别方法。首先描述了云模型的基本理论和数字特征,并给出了模态应变能的基本公式;然后分析了X条件云发生器和Y条件云发生器的基本算法和运行步骤,借助灰云模型建立相应的前件云和后件云规则,考虑了测量噪声的影响,利用云发生器生成多组云滴,并利用多模式下云滴的确定度和生成值构建了基本云推理算法及其损伤识别指标。基本云推理算法中常会产生不均匀发散的云滴,从而使计算结果产生一定的偏差,为了降低云滴发散产生的偏差影响,提出了基于损伤模式数量加权的云推理改进策略。计算结果表明:云推理算法可以较好地应用于塔架结构的损伤识别,其识别结果明显优于传统的应变能耗散率指标方法;而改进云推理算法进一步提高了识别的精度,优于基本云推理算法。