Non-linear filtering of ultrasonic signals using neural networks

Structure noise from inhomogeneous micro-structures makes the detection of flaws present in highly scattering materials difficult. Several techniques have been applied to improve the signal-to-noise ratio (SNR) in order to make flaw detection easier. Linear filtering does not provide good results because both structure noise and flaw signal concentrate energy in the same frequency band. Non-linear filtering can be used to reduce the structure noise of ultrasonic signals. Therefore, neural networks are applied in this work for this purpose. In order to use neural networks for non-linear filtering, dynamic structures must be applied. The easiest way to implement a neural network with the capability of processing temporal patterns is to consider them spatial ones, applying the signal into a tapped delay line of finite extension, that is the input of a static neural network (for example, a multi-layer perceptron). In this work, a dynamic neural network has been built to filter ultrasonic signals with structure noise, and has been trained with the real-time back-propagation algorithm, using as inputs 3000 synthetic ultrasonic signals of 896 samples each. Target signals for training are the same as the ones used as inputs but without noise. The neural network is trained in order to generate as output the target signal when the noisy input one is applied. For testing the performance of the non-linear filter, a new set of 500 noisy signals has been used. The SNR improvement is about 6 dB average. The results show that this non-linear filtering method is quite useful as pre-processing stage in flaw detection systems.     

An amplitude correlation and differencing method for the monitoring of flaws in repeat ultrasonic inspections

A new amplitude correlation and differencing (ACD) method for the detection of flaw changes using digital pulse-echo ultrasonic data is proposed. The ACD method is based on signal processing techniques applied to digitized pulse amplitudes from scans of the same transducer recorded during two separate ultrasonic inspections. Changes in flaws with a wide variety of sizes and other characteristics can be detected, provided the change is in a direction perpendicular to the centre line of the ultrasonic beam.Practical experience of this method was gained by analysis of ultrasonic data from two inspections of a weld, separated by a year. The effectiveness of the ACD method for flaw monitoring is compared with that of the 6 dB drop sizing method, before and after resolution improvement by the synthetic aperture focusing technique (SAFT).     

Signal-to-noise ratio enhancement based on wavelet filtering in ultrasonic testing

In ultrasonic non-destructive testing of materials with a coarse-grained structure the scattering from the grains causes backscattering noise, which masks flaw echoes in the measured signal. Several filtering methods have been proposed for improving the signal-to-noise ratio. In this paper we present a comparative study of methods based on the wavelet transform. Experiments with stationary, discrete and wavelet packet de-noising are evaluated by means of signal-to-noise ratio enhancement. Measured and simulated ultrasonic signals are used to verify the proposed de-noising methods. For comparison, we use signal-to-noise ratio enhancement related to fault echo amplitudes and filtering efficiency specific for ultrasonic signals. The best results in our setup were achieved with the wavelet packet de-noising method.     

Wavelet based noise suppression technique and its application to ultrasonic flaw detection

The noise suppression techniques with wavelet transform (WT) are widely used in non-destructive testing and evaluation (NDT&E), especially in ultrasonics. Complete reconstruction theory with hard or soft thresholds, reconstruction technique based on the singularities of noise and signal, matched filter with an impulse response, and optimal frequency-to-bandwidth ratio of wavelet technique have all been used to analyze ultrasonic signals for noise suppression. But a more simple and effective technique has been pursued for decades. This paper develops a new technique using WT for the right purpose. In this work, WT is treated as a band-pass filter whose central frequency and frequency bandwidth (CF&FB) are determined by the spectra distribution of an ultrasonic signal captured from real testing situation. For the purpose of matching their CF&FB well, a technique for evaluating the optimal scale of a daughter wavelet is carried out too. By acting this daughter wavelet as a band-pass filter, we can obtain excellent de-noising results, even when the signal to noise ratio (SNR) is below -18 dB. The performance of the technique has been done by ultrasonic signals with computer generated white noises. Finally, the experimental verification is performed on a pipeline specimen with man-made small flaws with good results obtained. The results show that the technique is more suitable for processing heavy noised ultrasonic signals, and it can also be used in automatic flaw detection.     

A methodology for structural health monitoring with diffuse ultrasonic waves in the presence of temperature variations

Diffuse ultrasonic waves for structural health monitoring offer the advantages of simplicity of signal generation and reception, sensitivity to damage, and large area coverage; however, there are the serious disadvantages of no accepted methodology for analyzing the complex recorded signals and sensitivity to environmental changes such as temperature and surface conditions. Presented here is a methodology for applying diffuse ultrasonic waves to the problem of detecting structural damage in the presence of unmeasured temperature changes. This methodology is based upon the prediction and observation that the first order effect of a temperature change on a diffuse ultrasonic wave is a time dilation or compression. A multi-step procedure is implemented to (1) record a set of baseline waveforms from the undamaged specimen at temperatures spanning the expected operating range, (2) select a waveform from the baseline set whose temperature is the closest to that of a subsequently measured signal, (3) adjust this baseline waveform to best match the signal, and (4) calculate an error parameter between the signal and the adjusted waveform and compare this parameter to a threshold to determine the structural status. This procedure is applied to experimental data from aluminum plate specimens with artificial flaws. Probability of detection and the minimum flaw size detected are presented as a function of the size of the baseline waveform set. It is shown that a probability of detection of over 95% can be achieved with a small number of baseline waveforms.     

A new similarity index for nonlinear signal analysis based on local extrema patterns

Common similarity measures of time domain signals such as cross-correlation and Symbolic Aggregate approximation (SAX) are not appropriate for nonlinear signal analysis. This is because of the high sensitivity of nonlinear systems to initial points. Therefore, a similarity measure for nonlinear signal analysis must be invariant to initial points and quantify the similarity by considering the main dynamics of signals. The statistical behavior of local extrema (SBLE) method was previously proposed to address this problem. The SBLE similarity index uses quantized amplitudes of local extrema to quantify the dynamical similarity of signals by considering patterns of sequential local extrema. By adding time information of local extrema as well as fuzzifying quantized values, this work proposes a new similarity index for nonlinear and long-term signal analysis, which extends the SBLE method. These new features provide more information about signals and reduce noise sensitivity by fuzzifying them. A number of practical tests were performed to demonstrate the ability of the method in nonlinear signal clustering and classification on synthetic data. In addition, epileptic seizure detection based on electroencephalography (EEG) signal processing was done by the proposed similarity to feature the potentials of the method as a real-world application tool.     

Quasi pulse-echo ultrasonic technique for flaw classification

One of the important flaw characterization tasks in the field of ultrasonic non-destructive evaluation (NDE) is to provide flaw type information by analysing the flaw responses acquired during an inspection. Here a new quasi pulse-echo ultrasonic classification technique is presented which utilizes the time separation and amplitude difference of mode-converted diffracted signals to distinguish between smooth and sharp-edged flaw geometries. Experiments with cylindrical cavities, surface-breaking fatigue cracks and slag inclusions have been used to test the practicality of this approach. All results of these tests show good consistency in the separation of smooth and sharp-edged flaws, provided that the signal-to-noise ratio is sufficient. Furthermore, the scattering feature used for classification in this method is also verified by detailed elastodynamic scattering calculations.     

Sparse signal representation and its applications in ultrasonic NDE

Many sparse signal representation (SSR) algorithms have been developed in the past decade. The advantages of SSR such as compact representations and super resolution lead to the state of the art performance of SSR for processing ultrasonic non-destructive evaluation (NDE) signals. Choosing a suitable SSR algorithm and designing an appropriate overcomplete dictionary is a key for success. After a brief review of sparse signal representation methods and the design of overcomplete dictionaries, this paper addresses the recent accomplishments of SSR for processing ultrasonic NDE signals. The advantages and limitations of SSR algorithms and various overcomplete dictionaries widely-used in ultrasonic NDE applications are explored in depth. Their performance improvement compared to conventional signal processing methods in many applications such as ultrasonic flaw detection and noise suppression, echo separation and echo estimation, and ultrasonic imaging is investigated. The challenging issues met in practical ultrasonic NDE applications for example the design of a good dictionary are discussed. Representative experimental results are presented for demonstration.     

Application of adaptive time-frequency decomposition in ultrasonic NDE of highly-scattering materials

In the paper, adaptive time-frequency decomposition by basis pursuit (BP) is utilized to improve ultrasonic flaw detection in highly-scattering materials as an alternative to the Wavelet Transform technique. The detection of ultrasonic pulses using the BP is described. Computer simulation was performed to verify the signal detection improvements for an ultrasonic wave embodied in white noise, and numerical results show good detection even for signal-noise ratio (SNR) of -18 dB. The improvement in detection is experimentally verified using cast steel samples with artificial flaws.     

Influence of structural interfaces on the statistics of corrosion microcracks

The amplitude distributions of acoustic emission signals generated during the formation of corrosion microcracks in loaded welded joints connecting two corrosion-resistant steel tubes are studied. For acoustic emission signal amplitudes of less than 0.6 mV and a microcrack concentration of ～10 mm^?3, the distribution density of the signal amplitudes is described by a gamma function. For acoustic emission signal amplitudes exceeding 1.0 mV and a microcrack concentration of greater than or equal to 10² mm^?3, the distribution density of the amplitudes exhibits two maxima, whose shape is described by a Gaussian function. The mean amplitudes of acoustic emission signals differ by a factor of 3. The change in the amplitude distribution of acoustic emission signals is explained by the effect exerted by the weld-metal interface on the microcrack formation.     

一种用于界面接触状态评价的超声信号相似度分析方法*

针对常规超声反射系数对界面接触状态灵敏度受限问题,论文发展一种用于界面接触状态评价的超声信号相似度分析方法。针对不同压力下承压界面超声检测信号,研究了激励信号带宽、相似度函数类型及分析域对相似度分析方法对界面接触状态评价的影响。结果表明,检测信号的相似度指标随压力增加呈现明显规律性。激励信号带宽、相似度函数类型和分析域对界面接触状态评价效果有很大的影响。宽带激励下的时域欧式距离指标对界面压力微变化最为敏感,可用于界面接触状态的定量评价。     

虚拟式声显微镜及应用技术研究

本文描述一台虚拟式扫描声显微镜及应用技术研究，它的硬件特别简单，仅由一台带有超高速A/D卡和快脉冲收发卡的PC机和机械扫描系统组成，主要的功能均由软件完成，脉冲收发卡仅包括脉冲产生和高频放大器，A/D卡采用PCI总线，采样率1GSPS，数据传输率100MB/s，系统能采集并实时显示频率达百兆的未检波回波信号，采修订本以的信号经软件处理，可在屏幕上实时显示样品的A、B、C型幅度像或相位像，仪器利用信号处理技术提高信号的信噪比；进行特征提取，识别，该仪器在材料识别，电子、光电子器件和焊接质量检测等方面功能较强，具有广阔的开发应用前景。     

Dynamic evaluation of flaw size in ultrasonic testing of steel plates

The paper deals with the problems of ultrasonic non-destructive testing of steel plates, such as distance sensitivity, acoustic coupling, flaw position dependence and their influence upon the flaw size evaluation, and describes equipment for the automated testing of circular stampings and their classification according to the size and number of flaws detected.     

超声探头的响应特性及暂态回波模型

建立检测系统的数学模型,可以更好地理解超声检测的物理本质。分析了超声波从产生、介质中传播、缺陷耦合以及接收的全过程,将缺陷回波表示为探头响应函数与缺陷响应的时域卷积。利用空间脉冲响应和基尔霍夫近似建立了超声波与平面型缺陷的耦合模型,用大平面试块底面回波和大平面响应进行反卷积求得了探头的响应函数,并详细分析了探头在不同偏置位置时不同大小缺陷响应的特点,发现缺陷回波由直达回波和边缘回波组成,直达回波和边缘回波极性相反,且直达回波的幅值远远大于边缘回波。      

An improved automated ultrasonic NDE system by wavelet and neuron networks

Despite of the widespread and increasing use of digitized signals, the ultrasonic testing community has not realized yet the full potential of the electronic processing. The performance of an ultrasonic flaw detection method is evaluated by the success of distinguishing the flaw echoes from those scattered by microstructures. So, de-noising of ultrasonic signals is extremely important as to correctly identify smaller defects, because the probability of detection usually decreases as the defect size decreases, while the probability of false call does increase. In this paper, the wavelet transform has been successfully experimented to suppress noise and to enhance flaw location from ultrasonic signal, with a good defect localization. The obtained result is then directed to an automatic Artificial Neuronal Networks classification and learning algorithm of defects from A-scan data. Since there is some uncertainty connected with the testing technique, the system needs a numerical modelling. So, knowing the technical characteristics of the transducer, we can preview which are the defects that experimental inspection should find. Indeed, the system performs simulation of the ultrasonic wave propagation in the material, and gives a very helpful tool to get information and physical phenomena understanding, which can help to a suitable prediction of the service life of the component.     

利用人工神经网络实现缺陷类型识别

本文在对各向同性均匀固体中横穿孔，平底孔和裂和裂缝缺陷超声散射特性分析的基础上，分别用回波幅度谱和去郑积幅度谱作为特征量，利用人工神经网络对缺陷类型进行识别。     

Wigner distribution of a transducer beam pattern within a multiple scattering formalism for heterogeneous solids

Diffuse ultrasonic backscatter measurements have been especially useful for extracting microstructural information and for detecting flaws in materials. Accurate interpretation of experimental data requires robust scattering models. Quantitative ultrasonic scattering models include components of transducer beam patterns as well as microstructural scattering information. Here, the Wigner distribution is used in conjunction with the stochastic wave equation to model this scattering problem. The Wigner distribution represents a distribution in space and time of spectral energy density as a function of wave vector and frequency. The scattered response is derived within the context of the Wigner distribution of the beam pattern of a Gaussian transducer. The source and receiver distributions are included in the analysis in a rigorous fashion. The resulting scattered response is then simplified in the single-scattering limit typical of many diffuse backscatter experiments. Such experiments, usually done using a modified pulse-echo technique, utilize the variance of the signals in space as the primary measure of microstructure. The derivation presented forms a rigorous foundation for the multiple scattering process associated with ultrasonic experiments in heterogeneous media. These results are anticipated to be relevant to ultrasonic nondestructive evaluation of polycrystalline and other heterogeneous solids.     

基于双高斯衰减模型的超声回波处理方法

信号降噪与特征提取是超声检测数据处理的关键技术.基于超声信号有特定结构而噪声和超声信号的结构无关,本文提出一种旨在解决强噪声背景下超声回波的参数估计和降噪问题的方法.该方法将超声回波的参数估计和降噪问题转换为函数优化问题,首先根据工程经验建立超声信号的双高斯衰减数学模型,然后根据观测回波和建立的超声信号模型确定目标函数,接着选择人工蜂群算法对目标函数进行优化从而得到参数的最优估计值,最后由估计出的参数根据建立的超声信号数学模型重构出无噪的超声估计信号.通过仿真和实验表明本文方法可以准确估计出信噪比大于-10 dB的含噪超声回波中的无噪信号,且效果优于基于自适应阈值的小波降噪方法和经验模态分解方法;此外相比常用的指数模型和高斯模型,本文提出的双高斯衰减超声信号模型与实测超声信号更接近,其均方误差为9.4×10~(-5),波形相似系数为0.98.     

面向电容式微机械超声换能器器件的32通道收发 电路设计与测试*

针对以电容式微机械超声换能器(CMUT)阵列为探头的超声成像系统,设计了一种兼具现场可编程门阵列(FPGA)控制、脉冲驱动以及微弱电流信号检测功能的电路。利用FPGA产生64路控制信号,控制脉冲信号的频率、脉冲个数、占空比等参数;脉冲电路在FPGA以及直流电压的控制下,产生32路脉冲信号;接收电路通过跨阻放大结构实现32路电流信号检测;32通道收发电路利用脉冲产生芯片内自带的T/R开关进行高压隔离。通过搭建测试平台,对收发电路功能及一致性进行测试,并连接CMUT进行自发自收测试。测试结果表明,32通道收发电路具有良好的一致性,电路可以实现基于CMUT阵列的32通道超声信号的发射,检测回波信号,并对CMUT器件的带宽进行测试。电路具有功能完善,结构稳定的优点,为基于CMUT阵列的超声成像系统的应用提供了硬件支持。     

Detecting small flaws near the interface in pulse-echo

Flaw detection near the interface surface is a common problem in many pulse-echo NDT applications due to interference with the interface echo, orders of magnitude above the flaw echoes. Several digital signal processing techniques like deconvolution, Hilbert transform and cepstrum analysis have been proposed to improve axial resolution. However, they require strict linearity, which takes a large portion of the system dynamic range just to fit the interface echo, thus reducing the dynamic range available for flaw detection. This work presents a new alternative based on the time-domain phase analysis of the received signals. Differently from conventional approaches, it works quite well with saturated signals resulting when a high gain is applied to detect small flaws. These can be detected in a range of a fraction of one wavelength from the interface surface, even using narrow-band transducers, as it has been experimentally verified. The method can be easily hardware implemented for real-time processing.