单向纤维增强复合板弹性常数的反演方法研究

利用小波变换和人工神经网络技术,对激光激发单向纤维增强复合板产生的兰姆波信号进行分析,进而反演出单向纤维增强复合板的弹性常数。对小波变换和神经网络算法加以优化以提高反演结果的准确性,并分别对纤维方向和垂直于纤维方向的实验波形进行弹性常数的反演,两个方向的反演结果十分吻合。研究表明,利用该方法可直接从激光激发兰姆波的瞬态实验波形中反演出复合板的弹性常数,既不用事先知道兰姆波各模式传播的群速度,也不必采用传统的最小二乘法拟合色散曲线来反演弹性常数。     

An Inverse method of elastic constants for unidirectional fiber-reinforced composite plate

An inverse method is presented to determine the elastic constants of an experimental sample, a titanium graphite unidirectional fiber-reinforced composite plate, using wavelet transform and neural networks. Optimal algorithms of wavelet transform and neural networks are given here in order to improve the accuracy of inversion results. Coherent results were shown in both fiber direction and cross fiber direction, proving the feasibility of this method. Neither the group velocity of the Lamb wave modes are needed, as in the conventional method, and no direct least-square fitting of the experimental waveforms is necessary. __________ Translated from Acta Acustica, 2005, 30(4) (in Chinese)     

Elastic constants measurement of anisotropic Olivier wood plates using air-coupled transducers generated Lamb wave and ultrasonic bulk wave

A hybrid elastic wave method is applied to determine the anisotropic constants of Olive wood specimen considered as an orthotropic solid. The method is based on the measurements of the Lamb wave velocities as well as the bulk ultrasonic wave velocities. Electrostatic, air-coupled, ultrasonic transducers are used to generate and receive Lamb waves which are sensitive to material properties. The variation of phase velocity with frequency is measured for several modes propagating parallel and normal to the fiber direction along a thin Olivier wood plates. A numerical model based mainly on an optimization method is developed; it permits to recover seven out of nine elastic constants with an uncertainty of about 15%. The remaining two elastic constants are then obtained from bulk wave measurements. The experimental Lamb phase velocities are in good agreement with the calculated dispersion curves. The evaluation of Olive wood elastic properties has been performed in the low frequency range where the Lamb length wave is large in comparison with the heterogeneity extent. Within the interval errors, the obtained elastic tensor doesn’t reveal a large deviation from a uniaxial symmetry.     

Characterization of mechanical properties of a hollow cylinder with zero group velocity Lamb modes

Hollow cylinders used in the industry must be regularly inspected. Elastic guided waves, similar to Lamb modes in a plate, can propagate in the axial direction or around the circumference. They are sensitive to geometrical and mechanical parameters of the cylindrical shell. The objective of this paper is to show that zero group velocity (ZGV) Lamb modes can be used to bring out anisotropy and to measure elastic constants of the material. This study provides experimental and numerical investigations on a Zirconium alloy tube extensively used by the nuclear industry in reactor core components. A non-contact method, based on laser ultrasound techniques and ZGV Lamb modes, demonstrates that the difference observed between axial and circumferential guided waves cannot be explained by an isotropic model. Then, a transverse isotropic model is used for the Zircaloy tube. Four of the five elastic constants are directly extracted from ZGV resonance frequencies. The last one is deduced from the measured dispersion spectra. With this complete set of constants, a good agreement is obtained between theoretical and experimental dispersion curves for both axially and circumferentially propagating guided waves.     

不锈钢折弯板裂纹Lamb波检测技术

金属弯板广泛应用于车辆、船舶、飞机等大型装备结构件中,折弯处易形成应力集中产生裂纹,直接影响构件的使用安全和寿命.该文开展了不锈钢弯板裂纹缺陷Lamb波检测技术研究,计算了钢板的频散曲线,优选3种不同频率的S0模态:0.25 MHz、0.5 MHz、1 MHz.利用COMSOL仿真软件建立频域仿真模型,模拟了Lamb波...     

Time-frequency analysis of the dispersion of Lamb modes.

Accurate knowledge of the velocity dispersion of Lamb modes is important for ultrasonic nondestructive evaluation methods used in detecting and locating flaws in thin plates and in determining their elastic stiffness coefficients. Lamb mode dispersion is also important in the acoustic emission technique for accurately triangulating the location of emissions in thin plates. In this research, the ability to characterize Lamb mode dispersion through a time-frequency analysis (the pseudo-Wigner-Ville distribution) was demonstrated. A major advantage of time-frequency methods is the ability to analyze acoustic signals containing multiple propagation modes, which overlap and superimpose in the time domain signal. By combining time-frequency analysis with a broadband acoustic excitation source, the dispersion of multiple Lamb modes over a wide frequency range can be determined from as little as a single measurement. In addition, the technique provides a direct measurement of the group velocity dispersion. The technique was first demonstrated in the analysis of a simulated waveform in an aluminum plate in which the Lamb mode dispersion was well known. Portions of the dispersion curves of the A0, A1, S0, and S2 Lamb modes were obtained from this one waveform. The technique was also applied for the analysis of experimental waveforms from a unidirectional graphite/epoxy composite plate. Measurements were made both along and perpendicular to the fiber direction. In this case, the signals contained only the lowest order symmetric and antisymmetric modes. A least squares fit of the results from several source to detector distances was used. Theoretical dispersion curves were calculated and are shown to be in good agreement with experimental results.     

基于BP神经网络的复杂彩色图像文本定位

针对复杂彩色图像中文本的特征,提出了基于小波变换和BP神经网络甄别文本区域的算法.该算法首先利用文本块的边缘特征遴选出备选图像块,而后采用小波变换提取备选图像块的纹理特征,把这些纹理特征参量连同图像块的颜色特征和笔画特征参量输入训练好的BP神经网络,判断备选图像块是否包含文本.该方法运算简单,定位时间短.采用专用的文本定位比赛用图进行实验的结果表明,定位准确率可达到92%,召回率为87.4%.     

神经网络反演算法在超声测量单向复合纤维材料特性中的应用

提出了一种神经网络反演算法,用于超声测量单向复合纤维材料的参数。算法基于Chistoffel方程,将单向复合纤维材料视作六角对称晶体,通过测量纤维样品中的声速,可以反演出五个独立的弹性系数C₁₁,C₁₂,C₁₃,C₃₃和C₄₄。反演方法则利用了神经网络反复学习不断用较好样例更新的训练样本集,克服了由于学习样例数目有限而使神经网络在预测过程中所产生的误差,提高了计算精度。并将该方法用于单向复合材料S2/DET85的弹性系数的实验测量和反演,所得声速与实验测量值吻合较好。该方法实现了超声测量单向复合纤维材料特性的自动递进反演过程。     

基于BP神经网络的复杂彩色图像文本定位*

针对复杂彩色图像中文本的特征,提出了基于小波变换和BP神经网络甄别文本区域的算法.该算法首先利用文本块的边缘特征遴选出备选图像块,而后采用小波变换提取备选图像块的纹理特征,把这些纹理特征参量连同图像块的颜色特征和笔画特征参量输入训练好的BP神经网络,判断备选图像块是否包含文本.该方法运算简单,定位时间短.采用专用的文本定位比赛用图进行实验的结果表明,定位准确率可达到92％,召回率为87.4%.     

Laser-ultrasonic generation of Lamb waves in the reaction force range

The Laser-ultrasonic generation of Lamb waves in an elastic plate is investigated theoretically and experimentally for a laser source whose intensity is high enough to create reaction forces (normal tractions) on the illuminated surface of the specimen. The analytical solutions for transient waves are derived using the integral transform method first by considering an arbitrary source shape and time excitation function, and then specifically for circular and line source shapes. The simulation study allows us not only to predict the behavior of individual wave modes but also to construct the overall responses; thus it helps us better understand the wave excitation mechanisms. The dispersive and multi-modal nature of laser-generated Lamb waves is presented by showing the spatiotemporal Fourier transform of displacements obtained by the simulation study. The transform, displayed in the frequency-wave number domain, enunciates the characteristics of the propagating individual Lamb wave modes. The simulation results are then compared with the 2-D Fourier transform of a set of experimental data obtained by scanning an aluminum plate specimen.     

Lamb波多特征参数的复合材料损伤程度 评估方法*

航空航天飞行器、风电等工程机械结构的安全性保障至关重要,复合材料等新材料的大量应用,进一步提升了对结构健康监测的难度和要求。裂纹、脱层等复合材料结构典型损伤形成机理复杂、易扩展,损伤程度累积严重时会造成结构的突然失效。面向复合材料结构的损伤监测与诊断,现有的方法大多很难实现对结构损伤程度的跟踪评估,进而无法对结构的安全性评价提供依据。采用Lamb波监测理论,基于其传播距离远、对小损伤敏感等特点,通过研究分析典型损伤对Lamb波结构响应信号传播的多个特征作用过程和机理,采用多尺度分析,提取结构损伤时频域多特征参数;借助于神经网络的非线性映射能力,构建多特征复合材料结构损伤程度评估模型,实现对不同程度下典型结构损伤的评估。在环氧玻璃纤维复合材料板结构上的实验验证结果表明：提取的Lamb波时频域多特征参数,对不同程度典型损伤有一定的敏感性,所构建的损伤程度评估系统可以较为有效的实现损伤的程度估计。     

Normal mode expansion method for laser-generated ultrasonic Lamb waves in orthotropic thin plates

A quantitative theory for modeling the laser-generated transient ultrasonic Lamb waves, which propagates along arbitrary directions in orthotropic plates, is presented by employing an expansion method of generalized Lamb wave modes. The displacement field is expressed by a summation of the symmetric and antisymmetric modes in the surface stress-free orthotropic plate, and therefore the theory is particularly appropriate for waveform analyses of Lamb waves in thin plates because one needs only to evaluate several lower modes. The transient waveforms excited by the thermoelastic expansion and the oil-coating evaporation are analyzed for a transversely isotropic thin plate. The results show that the theory provides a quantitative analysis to characterize anisotropic elastic stiffness properties of orthotropic plates by laser-generated Lamb wave detection.     

小波变换与神经网络融合法在油页岩近红外光谱分析中的应用

便携式近红外光谱分析技术可实现油页岩含油率的原位检测,在油页岩资源现场勘查中发挥着重要作用。但是,由于其测得的原始光谱数据量大、冗余信息多,直接建模会影响速度与精度。因此提出一种小波变换与神经网络融合法,先将油页岩全谱数据进行db8小波3级分解,提取其近似系数形成输入矩阵,然后再进行神经网络建模。为了验证有效性,利用30个油页岩合成样品,从中随机选择20个用于训练,另外10个用于预测,并分别使用全谱数据与小波特征数据进行了10次神经网络建模。结果表明,全谱数据建模速度均值为570.33 s,预测残差平方和及相关系数均值分别为0.006 012及0.843 75;而小波神经网络法对应的以上均值为3.15 s, 0.002 048及0.953 19。由此说明小波神经网络法优于全谱数据建模法,为油页岩含油率的快速、高精度检测提供了一种新方法。     

混杂复合材料等效热传导性能预测的小波-机器学习混合方法

混杂复合材料是一种新型复合材料,其复杂的细观结构导致预测其等效热传导性能极富挑战性.本文结合渐近均匀化方法、小波变换方法和机器学习方法发展了一种新的可以有效预测混杂复合材料等效热传导性能的小波-机器学习混合方法.该方法主要包括离线多尺度建模和在线机器学习两部分.首先借助渐近均匀化方法通过离线多尺度建模建立了混杂复合材料...     

稀疏降噪自编码结合高斯过程的近红外光谱药品鉴别方法

提出一种稀疏降噪自编码结合高斯过程的近红外光谱药品鉴别方法。首先对近红外光谱数据进行小波变换以消除基线漂移,然后用稀疏降噪自编码(SDAE)网络提取光谱特征并降维表示,最后采用高斯过程(GP)进行二分类,其中GP选用光谱混合(SM)核函数作为协方差函数,记此分类网络为wSDAG_SM。自编码网络具有很强的模型表示能力,高斯过程分类器在处理小样本数据时具有优势。wSDAG_SM网络通过稀疏降噪自编码学习得到维数更低但更有价值的特征来表示输入数据,同时将具有很好表达力的光谱混合核作为高斯过程的协方差函数,有利于更准确的光谱数据分类。以琥乙红霉素及其他药品的近红外光谱为实验数据,将该方法与经过墨西哥帽小波变换的BP神经网络(wBP)、支持向量机(wSVM), SDAE结合Logistic二分类(wSDAL)、SDAE结合采用平方指数(SE)协方差核的GP二分类(wSDAG_SE),以及未采用小波变换的SDAG_SM网络等方法进行对比。实验结果表明,对光谱数据进行墨西哥帽小波变换预处理能有效提升SDAG_SM网络的分类准确率和稳定性。wSDAG_SM方法无论从分类准确率还是分类结果稳定性方面,都优于其他分类器。     

DETERMINATION OF ELASTIC CONSTANTS OF ANISOTROPIC LAMINATED PLATES USING ELASTIC WAVES AND A PROGRESSIVE NEURAL NETWORK

In this paper, a procedure is suggested to inversely determine the elastic constants of anisotropic laminated plates using a progressive neural network (NN). The surface displacement responses are used as the inputs for the NN model. The outputs of the NN are the elastic constants of anisotropic laminated plates. The hybrid numerical method (HNM) is used to calculate the displacement responses of laminated plates to an incident wave for given elastic constants. The NN model is trained using the results from the HNM. A modified back-propagation learning algorithm with a dynamically adjusted learning rate and an additional jump factor is developed to tackle the possible saturation of the sigmoid function and to speed up the training process for the NN model. The concept of orthogonal array was adopted to generate the representative combinations of elastic constants, which reduces significantly the number of training data while maintaining its data completeness. Once trained, the NN model can be used for on-line determination of the elastic constants if the dynamic displacement responses on the surface of the laminated plate can be obtained. The determined elastic constants are then used in the HNM to calculate the displacement responses. The NN model would go through a progressive retraining process until the calculated displacement responses using the determined results are sufficiently close to the actual responses. This procedure is examined for an actual glass/epoxy laminated plate. It is found that the present procedure is very robust and efficient for determining the elastic constants of anisotropic laminated plates.     

Influence of thresholding procedures in ultrasonic grain noise reduction using wavelets

Wavelet transform based techniques are used for signal-to-noise ratio (SNR) enhancement in ultrasonic non-destructive testing and evaluation of strong sound scattering materials. The overall denoising performance of a wavelet signal processor is conditioned by several processing parameters, including the type of wavelet, thresholding method, and threshold selection rules. Different thresholding procedures and threshold selection rules are analysed in this paper using the discrete wavelet transform and decomposition level dependent thresholds. Global performance is evaluated by means of the SNR enhancement using synthetic grain noise registers with an incrusted flaw signal, with different values of the input SNR, and experimental ultrasonic traces acquired from a carbon fibre reinforced plastic composite block.     

Measurements of dispersion curves of leaky Lamb waves using a lensless line-focus transducer.

This paper presents a new method of measuring dispersion curves of leaky Lamb waves propagating in a thin plate. The measurement system is based on a lens-less line-focus transducer and its defocusing measurement technique. A new waveform processing method known as V(f, z) method is introduced for analyzing the measured waveforms and for extracting the dispersion relation. For two thin-plate samples, namely a 150 microm thick alumina plate and a 320 microm thick stainless steel plate, the dispersion curves of several lower order modes of leaky Lamb waves have been experimentally determined in the frequency range of few to 25 MHz. The experimental dispersion curves have then been compared with the theoretical ones. Excellent agreement is observed. It is then verify that this method indeed provides a convenient and accurate way for experimentally measuring dispersion curves of leaky Lamb waves of thin-plate samples.     

单向复合纤维材料弹性系数的超声测量

本文将单向复合纤维材料视作六角对称晶体，认为其弹性特征由五个独立的弹性系数来描述。利用Christoffel方程，我们得到这一各向异性介质中不同方向的声波速度与弹性系数的关系．我们还建立了一套计算机控制的实验测量系统，通过它对纤维板样品中的声速进行测量，反演得到了五个独立的弹性系数C11、C33、C13、C14和C12。理论计算得到的声速和实验测量值吻合较好，表明我们的方法是有实用意义的。     

Elastic constants determination by direct measurement of the beat wavelength between A0 and S0 Lamb modes with pulsed TV holography

A whole-field optical technique of high sensitivity, namely pulsed TV holography with phase evaluation by the Spatial Fourier Transform Method, is applied to the acquisition of instantaneous displacement fields of ultrasonic Lamb waves in aluminium plates a few millimetres thick. Two Lamb modes, the A0 and S0 ones, are simultaneously generated in the plate, producing a clear beating. Several values of the beat wavelength, corresponding to different excitation frequencies, are obtained by direct measurement of the distance between nodes in the wave field. The obtained values are fitted to the theoretical Rayleigh–Lamb frequency spectrum in order to determine the elastic constants of the plate material. We conclude that it is necessary to know the value of another parameter to univocally solve the problem, and so the bulk longitudinal wave velocity is measured by the pulse-echo method. Then the Poisson's ratio is obtained and, from these two parameters, the Young's modulus can also be determined.