A wavelet-based technique for damage quantification via mode shape decomposition

In this study, a neuro-wavelet technique was proposed for damage identification of cantilever structure. At first, damage localisation was accomplished through mode shape decomposition using discrete wavelet transforms. Subsequently, a damage indicator was defined based on the detail coefficients of the decomposed signals. It was found that distinct patterns relate the damage indicators to damage locations. Considering this property, a neural network ensemble was developed for damage quantification. Damage indicators and damage locations were selected as input parameters for the neural networks. Three individual neural networks were trained by input samples obtained from different combinations of decomposed mode shapes. Then, the outcomes of the individual neural networks were fed to the ensemble neural network for damage quantification. The proposed method was tested on a cantilever structure both experimentally and numerically. Six different damage scenarios including three different damage locations and three different damage severities were introduced to the structure. The results revealed that the proposed method was able to quantify different damage levels with a good precision.     

Seismic damage detection of tall airport traffic control towers using wavelet analysis

In this study, the applicability of continuous wavelet transform (CWT) and discrete wavelet transform (DWT) for seismic damage detection of tall airport traffic control (ATC) towers was investigated. Nonlinear finite element (NFE) model of Kuala Lumpur International Airport (KLIA) ATC tower with the height of 120 m was created using discrete moment-curvature hinges. Three different strong ground motions excited the tower and three different damage scenarios were then obtained. Response accelerations at four strategically selected locations were analysed by CWT and DWT to detect the damage scenarios. It was found that CWT successfully detects seismic-induced damage even when the signals are polluted by noises. On the other hand, DWT is quite sensitive to noisy signals and successful damage detection by DWT depends on noise level and sampling interval. Moreover, it was observed that DWT is more sensitive to the change in the stiffness of the tower structural elements than CWT.     

用空间分布信号的小波变换识别岩石材料的损伤

以空间分布信号的小波变换技术为基础，提出了一个识别岩石材料损伤的新方法。这个方法的理论基础是：岩石材料的损伤将使得结构的测量信号在损伤部位产生扰动。尽管测量信号中包含着材料损伤的信息，但很难直接从测量信号中发现损伤的确切位置，不过可以从测量信号的小波变换中发现损伤的位置，并用实验验证了这种方法的可靠性，即用声发射实验对含有预制裂纹的圆柱形水泥砂浆试样进行检测，可以得到沿试样轴向不同位置处的初至波幅度，对初至波幅度信号进行小波变换，结果表明，损伤位置与小波系数出现较大扰动的位置相对应，这说明本方法是可行的。     

Wavelet packet based damage detection in beam-like structures without baseline modal parameters

In this paper, a new approach for damage detection in beam-like structures is presented. The method can be used without the need for baseline modal parameters of the undamaged structure. Another advantage of the proposed method is that it can be implemented using a small number of sensors. In the proposed technique, the measured dynamic signals are decomposed into the wavelet packet decomposition (WPD) components, the power spectrum density (PSD) of each component is estimated and then a damage localisation indicator is computed to indicate the structural damage. The proposed method is firstly illustrated with a simulated beam and the identified damage is satisfactory with assumed damage. Then, the method is applied to a steel beam. The effect of damage location and the effects of wavelet type and the decomposition level are examined. The results show that the proposed method has great potential in crack detection of beam-like structures.     

A novel machine learning‐based algorithm to detect damage in high‐rise building structures

A novel model is presented for global health monitoring of large structures such as high‐rise building structures through adroit integration of 2 signal processing techniques, synchrosqueezed wavelet transform and fast Fourier transform, an unsupervised machine learning technique, the restricted Boltzmann machine, and a recently developed supervised classification algorithm called neural dynamics classification (NDC) algorithm. The model extracts hidden features in the frequency domain of the denoised measured response signals recorded by sensors on different elevations or floors of a structure. The extracted features are used as an input of the NDC to detect and classify the global health of the structure into categories such as healthy, light damage, moderate damage, severe damage, and near collapse. The proposed model is validated using the data obtained from a 3D 1:20 scaled 38‐story reinforced concrete building structure. The results are compared with 3 other supervised classification algorithms: k‐nearest neighbor (KNN), probabilistic neural networks (PNN), and enhanced PNN (EPNN). NDC, EPNN, PNN, and KNN yield maximum average accuracies of 96%, 94%, 92%, and 82%, respectively.     

Application of stress wave propagation technique for condition assessment of timber poles

Abstract

The stress wave propagation technique can be effectively used to assess the condition of timber utility poles. However, reliable detection of damage based on the reflected wave within the time domain is not always possible. Therefore, various signal processing methods such as frequency-domain analysis and time-frequency analysis can be adopted to overcome this problem depending on the application. In this paper, Hilbert–Huang and continuous wavelet transforms are selected as signal processing methods to analyse the reflected wave. The signal is initially subjected to an empirical mode decomposition process prior to the computation of instantaneous frequencies of the decomposed signals using the Hilbert–Huang transformation. The anomalies in the instantaneous frequency plots can be used to identify any damage and its location along the pole. Additionally, the decomposed signals are subjected to a wavelet transformation to further confirm the existence of damage. The combined Hilbert–Huang and continuous wavelet transform technique is applied to the stress wave signal recorded from the in-service poles to assess the accuracy of the proposed method. This method increases the confidence level of defect identification of timber utility poles.     

Decentralized damage identification using wavelet signal analysis embedded on wireless smart sensors

In this paper, a decentralized damage identification method using wavelet signal analysis tools embedded on wireless smart sensors (Imote2) has been proposed and experimentally validated. The damage identification analysis is decentralized by calculating discrete wavelet coefficients for acceleration in Imote2 sensors and transmitting the wavelet coefficients to a base station for damage identification through wavelet entropy indices. The wavelet entropy is modified to serve as a damage-sensitive signature that can be obtained both at different spatial locations and time stations to indicate existence of damage. It is known that wavelet-based approaches have clear advantages over Fourier transform-based ones for damage identification, since the wavelet transform allows for a wider choice of basis functions. This flexibility allows the wavelet transform to isolate changes in a signal that may be difficult to detect using other transform methods. To assess the reliability of the measurement signals, the wireless sensors have been compared with reference wired sensors. The proposed decentralized method for damage identification is verified via experimental tests using two laboratory structures: a three-story shear building structure and a three-dimensional truss bridge structure.     

多段微差爆破振动信号频带能量分布特征的小波包分析

爆破振动分析是研究爆破振动危害控制的基础,也是控制爆破振动危害的前提。根据爆破振动信号具有短时非平稳的特点,利用小波包分析技术对满足分析要求的多段微差爆破振动信号的能量分布特征进行研究。首先,简略地介绍了小波变换与小波包分析的特点;其次,对6条多段微差爆破振动信号进行小波包分析,得到了爆破振动信号在不同频带上的能量分布图;最后,总结了多段微差爆破振动信号频带能量的分布特征。该分析手段为综合研究爆破地震效应特别是为将来构建振动速度–频率相关安全准则提供了一种有效的分析技术。     

Combination of neural network and wavelet transform for islanding detection of distributed generation in a small-scale network

In this paper, an optimum and intelligent method is proposed for islanding detection using wavelet transform. The suggested relay is based on neural network (NN) in which different heuristic algorithms are used for training the NN. In the proposed method, the appropriate signals for detection procedure as well as mother wavelet are selected optimally, based on the mean square error (MSE) concept. Lately, the desired relay is trained by the optimally selected signals using four different algorithms and the optimum condition of the fault detector is identified. Simulation results approved that non detection zone (NDZ) has a significant reduction utilising the proposed intelligent technique. The contributions of the proposed method include presenting an appropriate signal selection method based on MSE, selecting optimum number of relay input signals using the proposed technique, fast training of intelligent relay by using least information, solving threshold selection problem and reduction of NDZ approximately to zero.     

基于小波变换的时-能密度法优选微差延期时间

在微差爆破工程中,能否顺利实施微差爆破的关键在于确定合理的微差延期时间。首先,利用基于小波分析的时-能密度法,通过从实测爆破震动信号中识别出各段雷管实际的起爆时刻点,得到了爆破中所用雷管的实际延期时间。其次,利用信号时-频转换技术,从实测微差爆破震动信号中分离出各分段震波。最后,通过比较各分段震波在不同延期时间下的叠加效果,可以得到微差爆破的较优延期时间。以某地下工程爆破震动信号进行分析为例,对本方法的有效性作了检验。该方法具有较高的理论和应用价值,为系统开展爆破震动危害控制和预测研究奠定了理论和技术基础。     

基于小波变换的微差爆破震动信号分离法

采用震波叠加模拟法确定合理的微差延期时间,关键在于能否获取组成实测微差爆破震波的分段震波。基于小波变换的微差爆破震动信号分离法能较好地解决上述问题。首先,利用基于小波分析的时-能密度法,通过从实测爆破震动信号中识别出各段雷管实际的起爆时刻点,得到了爆破中所用雷管的实际延期时间。其次,利用信号时一频转换技术。从实测微差爆破震动信号中分离出各分段震波。最后,通过比较各分段震波在不同延期时间下的叠加效果,可以得到微差爆破的较优微差延期时间。以某地下工程爆破震动信号进行分析为例,对本方法的有效性作了检验。     

基于小波变换时频能量分析技术的岩石声发射信号时延估计

声发射源定位技术是根据各传感器接收声发射信号的时差来实现损伤定位的,时延估计精确程度直接影响声发射源定位的精度。首先,在对不同损伤的声发射波形模式和频率识别的基础上,利用小波变换提取相应的单一频率或某一很窄频率段内的波形,并据此实现不同传感器在该频段的时延估计,为声发射源定位提供一个更为科学的方法;然后,使用Hyperion超声波系统,对单轴加载条件下岩石破裂过程中的声发射信号进行监测,并使用仪器自带的定位算法实现声发射源定位;最后,基于小波变换的方法对岩石试样声发射信号的时频能量分布特征进行分析,实现声发射源定位,并将定位结果与试样的真实破裂模式进行比较。试验结果表明,基于小波变换时频能量分析技术有利于减小声发射源定位的误差,从而提高材料损伤定位的精度。     

岩体爆破损伤声波测试信号频谱特征的小波(包)分析

 岩体爆破损伤特性除了影响声波速度外,同时造成声波能量衰减和频谱特征的变化。为弥补单纯声波速度分析的不足,更好地利用岩体声波信号携带的丰富信息,在某地下工程围岩中开展10次小药量模拟爆破岩体损伤声波测试研究。针对傅立叶分析的缺陷,运用小波(包)变换方法,对声波测试信号的频谱特征进行分解分析。研究结果表明：(1) 经小波(包)变换得到的爆破前后岩体声波频谱特征变化规律非常明显;(2) 2–4,2–5和2–6尺度下小波分量的幅值和功率谱密度远大于其他尺度下的小波分量的幅值和功率谱密度,对岩体爆破损伤的敏感性较好;(3) 岩体爆破损伤作用导致声波测试信号的能量集中区和最大能量分布百分比对应的频段(频率)向低频方向偏移。研究成果对于揭示岩体爆破损伤与声波测试信号频谱特征之间的内在联系具有一定的指导意义。     

耦合台风天气预报模式和实测数据的神经网络风速预测

基于准定常假定,风荷载与风速平方成正比.为了实现对结构的台风动力效应进行分析预测,建立了耦合数值天气预报(weather research and forecast,WRF)模式和现场实测数据的风速预测神经网络模型,开展台风短期风速的高精度预测.利用该模型对2017年"泰利"和2018年"康妮"的台风风场进行模拟和预测...     

基于Sym小波和BP神经网络的基桩缺陷智能化识别

为提高基桩低应变动测信号的分析水平,采用一种新的时频域分析方法——小波分析。利用Sym小波对基桩速度响应时程曲线进行小波分解,对指定频带上的信号分量进行特征值提取,提取的特征值为反映各频带范围内体现能量分布的功率谱均值,提取的特征值可构成反映信号特征的特征向量,同时利用BP人工神经网络的非线性映射特性建立特征向量和基桩缺陷类别之间的一种对应关系。通过数值模拟的方法可以得到大量不同缺陷类型的基桩的桩顶速度响应时程曲线,对这些数值模拟信号进行小波分解得到的特征向量为神经网络的学习提供大量训练样本。最后,利用实测信号小波分解后得到的特征向量对训练过的神经网络进行检验,其识别结果表明,训练后的神经网络能根据实测信号的特征向量对基桩缺陷进行智能化的识别。     

交通荷载下路面振动响应信号的时频特征分析

本文在简单分析了Fourier变换、短时Fourier变换和小波变换的基础上,以实测的交通荷载路面振动信号为模拟信号,采用小波包对模拟信号进行时频特征分析,并结合Fourier变换,给出了小波包各分解重构信号随时间的衰减规律及其在各子频带的频谱图,因此,结合Fourier变换对信号进行小波包分析是一种有效的信号处理途径.研究结果对今后交通荷载作用下路面路基动力响应计算模型的初始边界条件的确定提供依据,为路面路基动力响应的计算奠定基础.     

基于小波分析与支持向量机的地铁隧道沉降预测

根据地铁隧道监测点沉降变化中非线性、不确定、时变性的特点,建立了基于小波分析的支持向量机预测模型。首先运用小波分析将监测点沉降序列分解为低频近似分量和高频细节分量,然后对各分量分别进行支持向量机预测,最后将各分量预测结果进行小波重构得到监测点的沉降预测曲线。预测结果表明,在相同样本数和短周期预测条件下,Wavelet—SVM模型的预测精度优于BP神经网络方法。对地铁沉降监测提前进行预警预报有一定的参考价值。     

小波网络应用于空调负荷预测

准确预测空调负荷不仅对蓄能空调高效运行意义重大,而且也是冷热电三联产技术发挥优势的关键所在。本文提出一种小波网络应用于空调负荷的预测模型,通过小波分解,把空调负荷序列分解为不同频段的小波系数序列,再将各层的小波系数子序列重构到原尺度上,然后对小波系数序列采用相匹配的BP神经网络模型进行预测,最后合成空调负荷序列的最终预测结果。该预测模型中的低频小波系数a3和中频小波系数d3的神经网络输入变量为前1天小波系数值和对应时刻的温度、相对湿度、风速、总辐射量、天气状况和星期几编码共7个因子,并采用主成分分析法进行输入变量的降维;高频小波系数d2和d1以前几日的小波系数为输入因子。经过对西安市某综合楼的空调负荷进行预测,证明了预测值和实际运行值拟和很好,相对误差为-10%~8%。该预测模型具有预测精度较高、推广能力较强及计算速度较快的优点。     

基于小波分解的城市供水管网短期水量负荷预测

利用小波分解和人工神经网络相结合的方法建立了城市供水管网短期水量负荷的组合预测模型。该方法首先利用小波分解技术将时负荷水量分解为相对简单的带通分量信号，然后根据各分量信号的特点分别建立独立的神经网络预测模型，最后将预报结果集成。由于小波分解后各分量的频率相对单一，因而可有效缩短网络训练时间，提高计算速度。仿真计算结果表明，该方法建模合理、计算量适中，可准确预测管网水量。     

卷积神经网络在结构损伤诊断中的应用

对卷积神经网络(CNN)在工程结构损伤诊断中的应用进行了深入探讨; 以多层框架结构节点损伤位置的识别问题为研究对象,构建了可以直接从结构动力反应信号中进行学习并完成分类诊断的基于原始信号和傅里叶频域信息的一维卷积神经网络模型和基于小波变换数据的二维卷积神经网络模型; 从输入数据样本类别、训练时间、预测准确率、浅层与深层卷积神经网络以及不同损伤程度的影响等多方面进行了研究。结果表明:卷积神经网络能从结构动力反应信息中有效提取结构的损伤特征,且具有很高的识别精度; 相比直接用加速度反应样本,使用傅里叶变换后的频域数据作为训练样本能使CNN的收敛速度更快、更稳定,并且深层CNN的性能要好于浅层CNN; 将卷积神经网络用于工程结构损伤诊断具有可行性,特别是在大数据处理和解决复杂问题能力方面与其他传统诊断方法相比有很大优势,应用前景广阔。