ANALYSES ON STRUCTURAL DAMAGE IDENTIFICATION BASED ON CIMBINED PARAMETERS

The relative sensitivities of structural dynamical parameters were analyzed using a directive derivation method. The neural network is able to approximate arbitrary nonlinear mapping relationship, so it is a powerful damage identification tool for unknown systems. A neural network-based approach was presented for the structural damage detection. The combined parameters were presented as the input vector of the neural network, which computed with the change rates of the several former natural frequencies (C), the change ratios of the frequencies (R), and the assurance criterions of flexibilities ( A ). Some numerical simulation examples, such as, cantilever and truss with different damage extends and different damage locations were analyzed. The results indicate that the combined parameters are more suitable for the input patterns of neural networks than the other parameters alone.     

利用振动模态测量值和神经网络方法的结构损伤识别研究

提出了一种基于模态测量参数和神经网络的结构损伤检测方法,建造了两种输入方式的BP神经网络,即自振频率以及结合自振频率与振型,并讨论了不同数量的输入信息对结构损伤检测精度和计算效率的影响。证明了输入的参数越多,神经网络就越聪明,训练的收敛速度越快;以及在保证一定的测量精度的情况下,基于频率与振型的损伤识别结果要好于基于频率的检测结果。最后,通过对3层框架模型的4种损伤工况下的结构损伤检测结果的分析,认为利用模态测量参数和神经网络方法能够准确地识别结构损伤的位置,而且能较精确地识别结构损伤的大小。     

基于转角模态和小波神经网络的连续梁损伤识别研究

基于小波奇异性检测原理和神经网络非线性映射能力,结合结构基本模态参数,提出了一种结合小波神经网络与结构转角模态的损伤识别方法．首先,建立三跨连续梁的有限元模型获取结构模态参数,并对其进行Mexihat小波变换,通过系数图突变点判断结构损伤位置．然后,将小波系数模特征向量作为BP神经网络的输入,分别研究了该方法在单损伤和多损伤工况下的识别能力．最后将不同工况下神经网络预测值与结构实际损伤程度进行对比,得到单处损伤预测误差平均值为0.22%,多处损伤预测误差平均值分别为0.22%和0.18%,结果表明该方法在结构损伤识别方面的有较高有效性及精确度．     

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。     

Vibration response of rotating mechanical systems using experimental techniques and artificial neural networks

A neural network predictor investigation is presented for analyzing vibration parameters of a rotating system. The vibration parameters of the system, such as amplitude, velocity, and accelertion in the vertical direction, were measured at the bearing points. The system's vibration and noise were analyzed for different working conditions. The designed neural predictor has three layers, which are input, hidden, and output layers. In the hidden layer, 10 neurons were used for this approximation. The results show that the network can be used as an analyzer of such systems in experimental applications.     

时变环境与损伤耦合下桥梁结构频率及阻尼比的统计分析

对时变环境与损伤耦合下我国某斜拉桥的结构频率及阻尼比进行统计分析,以提高结构损伤识别的精度.首先,利用该桥的长期监测数据,采用环境激励技术结合特征系统实现算法识别该桥梁结构的频率及阻尼比;其次,利用人工神经网络算法建立该桥梁结构的环境温度与结构频率及阻尼比的关系模型;然后,通过统计分析,建立完好状态下该桥梁结构模态参数误差因子的概率分布模型;最后,通过分析不同时段与完好结构状态下该桥梁结构模态参数误差因子的相交概率比识别结构损伤,并利用该桥的实测结果验证所提算法的有效性.     

结构损伤的小波分形神经网络检测

用神经网络进行结构损伤检测、分析的有效性在很大程度上取决于训练样本的好坏。小波变换在时域和频域都具有表征信号局部特征的能力,小波包分析利用可以伸缩和平移的可变视窗能够聚焦到信号的任意细节,因此对有损伤的结构的非线性动力特性能进行有效的分析。利用分形几何方法不依赖于系统的数学模型的特点,将分形维数与小波分析相结合,建立了结构损伤的小波分形神经网络检测方法。研究结果表明,结构不同状态下的振动信号的各频段分形维数有明显的不同,可以将振动信号的各频段分形维数作为结构损伤检测的特征量,并用神经网络将结构的不同状态模式识别出来。     

Damage detection in structures using modified back-propagation neural networks

A nonparametric structural damage detection methodology based on neural networks method is presented for health monitoring of structure-unknown systems. In this approach appropriate neural networks are trained by use of the modal test data from a ‘healthy’ structure. The trained networks which are subsequently fed with vibration measurements from the same structure in different stages have the capability of recognizing the location and the content of structural damage and thereby can monitor the health of the structure. A modified back-propagation neural network is proposed to solve the two practical problems encountered by the traditional back-propagation method, i.e., slow learning progress and convergence to a false local minimum. Various training algorithms, types of the input layer and numbers of the nodes in the input layer are considered. Numerical example results from a 5-degree-of-freedom spring-mass structure and analyses on the experimental data of an actual 5-storey-steel-frame demonstrate that neural-networks-based method is a robust procedure and a practical tool for the detection of structural damage, and that the modified back-propagation algorithm could improve the computational efficiency as well as the accuracy of detection Project supported by the National Natural Science Foundation of China (No. 59908003) and the Natural Science Foundation of Hubei Province (No. 99J035).     

基于神经网络技术的管道机电阻抗健康状况定量评估研究

将机电阻抗法用于管道法兰和主干结构健康监测,并利用BP神经网络对结构损伤进行定量评估.首先实验研究了管道法兰与主干结构健康状况对阻抗谱的影响,不同的结构损伤可通过阻抗分析仪测量的阻抗实部谱变化反映出来;然后利用BP神经网络技术对管道不同工况下得到的阻抗实部谱进行定量分析.采用阻抗值实部作为输入样本对神经网络进行训练,并使用受训的神经网络实现了对管道中不同结构损伤状况的定量评估.研究结果表明,将机电阻抗法与神经网络数据处理技术结合起来用于复杂管道的结构健康监测,不仅可实现对不同类型损伤的定量评估,同时还具有较高的稳定性.     

组合神经网络在结构损伤检测中的应用

子结构的动态响应变化与整体结构相比,对结构内部损伤反应更为敏感。组合神经网络可以克服单个神经网络功能的单一局限性,实现更加全面综合的仿真识别功能。本文首先运用双协调自由界面模态综合法对结构进行模态分析,获取各子结构及整体结构的模态信息。然后,通过组合BP神经网络将损伤子结构与整体结构的模态频率变化率组合起来进行结构损伤检测。该方法在改善网络训练性能的同时,提高了检测结果的准确性和可靠性。文章最后通过数值算例验证了该方法的可行性和有效性。     

基于“AM-GoogLeNet+BP”联合数据驱动的混凝土细观模型压缩应力-应变曲线预测

本文结合GoogLeNet卷积神经网络和BP神经网络分别在图像数据挖掘和数据分析方面的良好性能,采用“AM-GoogLeNet+BP”联合数据驱动方法,对混凝土细观模型(含砂浆、骨料及孔隙)的单轴压缩应力-应变曲线进行了有效预测.通过引入力学参量对图像数据驱动的训练结果进行优化,从而提升了神经网络的物理可解释性.基于Python语言实现混凝土细观模型在Abaqus中的自动建模及细观图像生成过程,并将生成的细观图像数据库与相应的压缩应力-应变曲线作为训练数据集.在GoogLeNet中分别引入SENet, ECANet和CBAM三种代表性注意力机制并对三种注意力机制的性能进行对比和分析,以自适应方式提升神经网络对混凝土各相组分的分析能力,并以此得到混凝土细观模型的初步应力-应变预测曲线;将骨料体积分数、孔隙率及初步峰值应力等物理参量作为输入引入BP神经网络以改善峰值应力的预测精度,并与将物理参量直接引入卷积神经网络输入层的方法进行了对比,最后定量给出了骨料体积分数和孔隙率对峰值应力的影响权重.结果表明,对于不同骨料体积分数及孔隙率的混凝土细观模型,该方法均展现了较高的预测精度.本文采用的“...     

Exploring human rhythmic gait movement in the role of cerebral cortex signal

The rhythmic movement is a spontaneous behavior due to the central pattern generator(CPG).At present,the CPG model only shows the spontaneous behavior,butdoes not refer to the instruction regulation role of the cerebral cortex.In this paper,a modified model based on the Matsuoka neural oscillator theory is presented to better show the regulation role of the cerebral cortex signal to the CPG neuronal network.The complex interaction between the in put signal and other parameters in the CPG networkis establish...     

结构试验机的神经网络内模自适应跟踪控制系统

针对结构试验系统的非线性和不确定性特性，研究了一种基于神经网络的非线性内模自适应加载控制方法。引入的神经网络内模可跟踪学习对象的时变动力学，控制器的设计较少依赖于对象的先验知识，控制参数的调整是基于被控过程的测量信息，利用导出的神经网络算法来实现的。实验结果证明该系统具有良好的控制效果。     

Exploring human rhythmic gait movement in the role of cerebral cortex signal

The rhythmic movement is a spontaneous behavior due to the central pattern generator (CPG). At present, the CPG model only shows the spontaneous behavior, but does not refer to the instruction regulation role of the cerebral cortex. In this paper, a modified model based on the Matsuoka neural oscillator theory is presented to better show the regulation role of the cerebral cortex signal to the CPG neuronal network. The complex interaction between the input signal and other parameters in the CPG network is established, making all parameters of the CPG vary with the input signal. In this way, the effect of the input signal to the CPG network is enhanced so that the CPG network can express the self-regulation movement state instead of being limited to the spontaneous behavior, and thus the regulation role of the cerebral cortex signal can be reflected. Numerical simulation shows that the modified model can generate various movement forms with different modes, frequencies, and interchanges between them. It is revealed in theories that the cerebral cortex signal can regulate the mode and frequency of the gait in the course of the gait movement.     

模糊—神经网络控制算法及其在离心力—振动复合环境试验系统中的应用

针对离心－振动复合环境试验系统所存在的耦合性、非线性和不确定性提出了一种模糊－神经网络控制算法,利用被控对象输入输出信息离线、在线相结合学习系统的动态特性,对时变、非线性系统进行跟踪控制,并研究了该算法在系统中的实现方法。实现表明了控制系统具有良好的跟踪能力。该算法也适用于快速变化这类系统的实时控制。     

Combined neural network and reduced FRF techniques for slight damage detection using measured response data

Summary  This paper deals with structural damage detection using measured frequency response functions (FRF) as input data to artificial neural networks (ANN). A major obstacle, the impracticality of using full-size FRF data with ANNs, was circumvented by applying a data-reduction technique based on principal component analysis (PCA). The compressed FRFs, represented by their projection onto the most significant principal components, were used as the ANN input variables instead of the raw FRF data. The output is a prediction of the actual state of the specimen, i.e. healthy or damaged. A further advantage of this particular approach is its ability to deal with relatively high measurement noise, which is a common occurrence when dealing with industrial structures. The methodology was applied to detect three different states of a space antenna: reference, slight mass damage and slight stiffness damage. About 600 FRF measurements, each with 1024 spectral points, were included in the analysis. Six 2-hidden layer networks, each with an individually-optimised architecture for a specific FRF reduction level, were used for damage detection. The results showed that it was possible to distinguish between the three states of the antenna with good accuracy, subject to using an adequate number of principal components together with a suitable neural network configuration. It was also found that the quality of the raw FRF data remained a major consideration, though the method was able to filter out some of the measurement noise. The convergence and detection properties of the networks were improved significantly by removing those FRFs associated with measurement errors. Received 9 March 2000; accepted for publication 12 December 2000     

基于爆炸切割试验的有机玻璃本构模型参数反演

为了获取爆炸切割数值模拟中有机玻璃（PMMA）的材料本构模型参数，建立了一种基于神经网络的有机玻璃Johnson Holmquist ceramics (JH-2)本构模型参数反演方法：基于从爆炸切割试验和现有研究得到的JH-2本构模型经验参数，确定本构模型参数的调整区间；使用LS-DYNA数值模拟软件对2.5 mm宽爆炸切割索切割14 mm PMMA平板过程进行数值模拟并收集平板损伤数据集；建立PMMA平板本构模型参数与损伤数据之间的神经网络模型；通过训练完成的神经网络模型对PMMA平板的JH-2本构模型参数进行反演。为验证通过反演参数的可靠性，进行了4.2 mm宽爆炸切割索切割19 mm PMMA平板试验和有限元数值模拟，计算结果中的平板损伤情况与实验结果相差较小，表明通过反演获得的JH-2本构模型参数能较好地应用于PMMA平板爆炸切割数值模拟。传统材料参数获取方法，该参数反演方法相较于可以通过较少的试验及测试，获得比较准确的材料本构模型参数。     

基于代理模型的箭体连接结构极值响应分析与优化

螺栓法兰连接结构在航空航天等工程领域中广泛应用，其力学性能在不同工况和装配情况下十分复杂。由于拉压刚度差异，含连接结构的箭体动力学响应呈现明显的非线性特征。因此，考虑不同连接参数及工况下的连接非线性动力学响应，对结构优化设计有着重要意义。本文针对以双线性弹簧表征螺栓法兰连接非线性的箭体等效动力学模型，基于径向基函数(RBF)神经网络和响应面法分别建立其连接面处的极值响应代理模型，对比发现RBF神经网络模型在较高精度上可以实现对动响应极值的预测及分析；同时分析了不同载荷参数及刚度变化对连接结构动响应极值的影响；最后，利用RBF神经网络代理模型，开展了连接面加速度极值响应与螺栓弹簧力最小化为目标的连接结构参数优化。     

水下发射水动力的多尺度预测网络研究

空泡的演化和水动力特征的预测在航行体发射的设计中有非常重要的意义. 人工智能技术已经成为了参数预测的重要手段.为了能够快速预测航行体水下发射过程的尾部压力的复杂变化, 提出了一种多尺度深度学习网络.该网络模型以一维卷积网络(1DCNN)为基础,构建了一种编码-解码型网络结构,通过不同的采样频率将原始数据分解为光滑部分和脉动部分,进而训练低保真度的大尺度网络和高保真度的小尺度网络.从而实现对不同物理过程的响应和捕捉.首先,通过数值模拟获得了不同发射条件下的尾部压力曲线,并结合空泡的理论机理构建了具有物理性的输入数据集.其次,将数据集进行分解处理,分别训练了两个尺度的深度学习网络. 最终将两组输出数据整合在一起,建立了底部压力预测模型.并通过测试和验证说明本文提出的多尺度网络对于多种常见的发射条件,能够实现航行体受力特征的快速准确的预测,光滑曲线、压力突变、震荡的频率和幅值都和数值模拟的结果吻合.证明本文的方法能够为运动和弹道的预测提供依据.      

爆破地震波三要素对多层砌体结构弹塑性地震响应的影响

利用Matlab编写了一套Newmark时程分析法与刚度退化二线型恢复力模型相结合求解多层砌体结构体系弹塑性地震响应的程序.以一个具体的4层砌体结构为分析对象,输入荷载用实测及人工模拟爆破地震波作用的形式,并按某原则变换输入爆破地震波的特性参数,根据所求出的结构各弹塑性地震响应幅值结果,分别讨论了爆破地震波幅值、主频和...