Ndt & E International最新文献

{"title":"Physics-informed interpretation and classification of acoustic emission signals using explainable deep convolutional neural networks","authors":"Xuhui Huang,&nbsp;Obaid Elshafiey,&nbsp;Ming Han,&nbsp;Yiming Deng","doi":"10.1016/j.ndteint.2025.103487","DOIUrl":"10.1016/j.ndteint.2025.103487","url":null,"abstract":"<div><div>Model interpretability remains a critical challenge for deep learning applications in Acoustic Emission (AE) signal characterization, limiting their trustworthiness in structural health monitoring. The proposed approach integrates explainable Convolutional Neural Network with physics-informed segmentation to enhance both classification accuracy and interpretability. By segmenting signals based on the theoretical arrival times of fundamental Lamb wave modes (<span><math><mrow><msub><mi>S</mi><mn>0</mn></msub></mrow></math></span> and <span><math><mrow><msub><mi>A</mi><mn>0</mn></msub></mrow></math></span>), and employing Class Activation Mapping (CAM), Gradient-weighted CAM (Grad-CAM), and Dimension-wise CAM (DCAM), we provide quantitative insights into the model's decision-making process. Using 200 AE signals from pencil break tests, our model identifies distinct features for different events. Visualizations show the model focuses on <span><math><mrow><msub><mi>S</mi><mn>0</mn></msub></mrow></math></span>-<span><math><mrow><msub><mi>A</mi><mn>0</mn></msub></mrow></math></span> transition and post-A₀ regions, with DCAM highlighting significant importance in the <span><math><mrow><msub><mi>S</mi><mn>0</mn></msub></mrow></math></span>-<span><math><mrow><msub><mi>A</mi><mn>0</mn></msub></mrow></math></span> transition region for the closest test point. In this paper, we address the 'black box' nature of deep learning by offering quantitative, physics-based intuition, correlating model outputs to specific AE signal segments and underlying physical processes such as Lamb wave mode interactions and dispersion. By bridging the gap between deep learning performance and human-interpretable insights, our method enhances the reliability of AE-based structural health monitoring.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103487"},"PeriodicalIF":4.1,"publicationDate":"2025-07-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144670539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Data-driven deep representation of acoustic signals amplifies the accuracy of coin-tap test for non-destructive detection of defects","authors":"Yonglin Wu ,&nbsp;Hongyu Li ,&nbsp;Peng Jiang,&nbsp;Tiejun Wang","doi":"10.1016/j.ndteint.2025.103488","DOIUrl":"10.1016/j.ndteint.2025.103488","url":null,"abstract":"<div><div>Coin-tap test is a simple way for non-destructive detection of defects, and has long been used in engineering structures. However, improving the accuracy of coin-tap test is challenging. In this work, we propose a data-driven deep representation method for acoustic signals to amplify the accuracy of coin-tap test. We design an incremental dense one-dimensional convolutional neural network (IDCNN) with two feature aggregation blocks to organize deep representations. We introduce six types of defects to three types of bi-layered structures, use coin-tap tests to obtain acoustic signals, and train the IDCNN. The results show that the IDCNN performs well for deep representation of acoustic signals and significantly amplifies the accuracy of coin-tap test. The accuracy rate for defect detection ranges from 98.42 % to 99.06 %. The rates of missing and false alarms for defects are extremely low, ranging from 0.94 % to 1.58 % and from 0.72 % to 1.32 %, respectively. The results show that the data-driven deep representation of acoustic signals results in an effective coin-tap test for non-destructive detection of defects. The proposed method has potential for broad applications in acoustic-based non-destructive tests.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103488"},"PeriodicalIF":4.1,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144696871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Three-dimensional spatial coherent imaging based on two-dimensional laser ultrasonic row-column addressing array","authors":"Long Chen ,&nbsp;Zenghua Liu ,&nbsp;Xiaoyu Liu ,&nbsp;Yanping Zhu ,&nbsp;Cunfu He","doi":"10.1016/j.ndteint.2025.103486","DOIUrl":"10.1016/j.ndteint.2025.103486","url":null,"abstract":"<div><div>Three-dimensional (3D) imaging represents a crucial advancement in ultrasonic non-destructive testing (NDT). Compared with one-dimensional (1D) arrays, full matrix capture (FMC) based on two-dimensional (2D) arrays experiences exponential growth in required elements and sampling channels, fundamentally constraining array aperture and imaging resolution. The row-column addressing (RCA) array innovatively reduces the element count from <em>N</em> × <em>N</em> to <em>N</em> + <em>N</em> through alternating row-column addressing, significantly decreasing channel requirements and sampling data volume. This study pioneers a fully optical approach for 2D RCA array, aiming to achieve 3D non-contact imaging of body defects. Methodologically, the research progresses through three phases: First, to generate uniform energy cylindrical waves, an optical modulation technique was developed to create elongated homogenized laser spots, enabling laser ultrasonic orthogonal compound scanning for array data acquisition. Second, a 3D spatial coherence imaging algorithm was established for multimode characterization of artificial blind holes and simulated corrosion defects. Finally, systematic comparisons of several array configurations and imaging algorithms were conducted, with optimization strategies analyzed. Experimental results demonstrate that laser ultrasonic RCA array imaging effectively reconstructs 3D defect geometries and dimensional features. Multimode visualization successfully resolves geometrical variations across defect regions. This work introduces a novel non-contact array inspection methodology, facilitating the transition from 2D to 3D in ultrasonic NDT. The findings provide critical insights into adaptive algorithm selection and system optimization for high-resolution volumetric inspection.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103486"},"PeriodicalIF":4.1,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144656365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-linear feature based damage localization in riveted plates using FBG sensors for guided wave sensing utilizing instantaneous baseline","authors":"Rohan Soman ,&nbsp;Farzam Omidi Moaf ,&nbsp;Maciej Radzienski ,&nbsp;Marzena Kurpińska ,&nbsp;Tomasz Wandowski ,&nbsp;Pawel Kudela","doi":"10.1016/j.ndteint.2025.103483","DOIUrl":"10.1016/j.ndteint.2025.103483","url":null,"abstract":"<div><div>Structural health monitoring (SHM) systems ensure safety and reliability of structures. They may be used for maintenance planning which reduces the maintenance cost and potentially can extend lifetime. As a result, there is a lot of active research in the area for SHM of structures. The SHM system should be low cost, suitable for continuous monitoring, able to detect small levels of damage. Guided waves (GW) based SHM techniques allow monitoring of large plate-like structures with a few sensors and have been identified as the most promising of techniques for SHM.</div><div>One of the challenges of GW based techniques is that most of the developed techniques depend on comparison of the current measurements with a known baseline. Often the baseline measurement is not available or is acquired under different operating conditions. As a result, the applicability of these techniques although excellent in laboratory testing is not satisfactory in real applications. Few works utilizing non-linear features as a damage sensitive feature for reference-free SHM have been proposed with varying degrees of success. The current work builds on this initial idea for utilizing non-linear features and pairs it up with instantaneous baseline approach to develop a technique for reference-free damage detection and localization. Through the utilization of appropriate time windows and the inherent directional sensitivity of the fibre Bragg grating (FBG), it has been shown that damage can be satisfactorily detected and localized. In order to ascertain the presence of damage, a novel metric based on the spectral energy is developed. The robustness of the metric is further enhanced through multi-frequency data fusion. The developed metric and the utilization of FBG sensors for measurement of non-linear features is the main novelty of this work. In addition, a systematic approach for the threshold determination which takes into consideration systematic non-linearities is proposed. The methodology is applied on two aluminium plates riveted together. Damage is simulated through removal of the particular rivet. Various damage scenarios have been studied and the performance of the technique is compared with existing state-of-the art methods.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103483"},"PeriodicalIF":4.1,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144656364","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"U-NET autoencoder for amplitude recovery on binarized Full Matrix Captures","authors":"Aurélien Thon ,&nbsp;Guillaume Painchaud-April ,&nbsp;Alain Le Duff ,&nbsp;Pierre Bélanger","doi":"10.1016/j.ndteint.2025.103481","DOIUrl":"10.1016/j.ndteint.2025.103481","url":null,"abstract":"<div><div>Full Matrix Capture (FMC) and the Total Focusing Method (TFM) are instrumental techniques in ultrasonic nondestructive testing (NDT) in industries such as aerospace, oil and gas, and manufacturing, and allow efficient defect detection by capturing all possible transmitter–receiver pairs and generating highly resolved images on a predefined pixel grid. The use of dense linear or matrix probes presents significant challenges in data storage and transfer but also in the complexity of the acquisition system’s electronics. In this context, binary acquisition steps in as an attractive alternative for simplifying acquisition equipment and reducing data size. However, binary formats carry the drawback of amplitude information loss. To address this, the present study explores the application of a U-NET autoencoder neural network to reconstruct amplitude data from binarized FMC signals. The autoencoder’s U-NET architecture is particularly suited for this task due to its effectiveness with limited datasets, a common issue in NDT. Finite element simulations were used to generate training and validation datasets. Experimental tests were then conducted on steel samples containing various defects, such as Electrical Discharge Machining (EDM) cracks, side-drilled holes (SDH), and a realistic fatigue crack in a steel bar. The reconstructed FMC data were evaluated using TFM images and Structural Similarity Index Measure (SSIM), showing that the neural network accurately reconstructed FMCs. Notwithstanding the presence of minor amplitude errors, the spatial positioning of defects remained precise, demonstrating the method’s viability for practical NDT applications.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103481"},"PeriodicalIF":4.1,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144623360","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Geometry complex structure inspection using stretchable eddy current array sensors","authors":"Lei Peng ,&nbsp;Na Zhang ,&nbsp;Xuhui Huang ,&nbsp;Zi Li ,&nbsp;Yiming Deng","doi":"10.1016/j.ndteint.2025.103482","DOIUrl":"10.1016/j.ndteint.2025.103482","url":null,"abstract":"<div><div>Eddy-current testing (ECT) is widely used for the inspection and evaluation of structures made from conductive materials. In recent years, significant efforts have been made to improve the reliability and sensitivity of ECT sensors, including the development of state-of-the-art flexible eddy current sensor arrays based on polyimide substrates. These probes can adapt to uneven surfaces due to their flexibility. However, there are still situations where these probes fail. In this work, we propose a stretchable EC sensor array probe with a soft substrate material, designed to deform and conform to complex surfaces. A 3D simulation model was developed to study the effect of the probe's stretchability on defect detection. A prototype of the probe was fabricated and tested on various samples. Experimental results demonstrate that the proposed probe successfully inspected pit defect with 2 mm radius and 0.5 mm depth on a bent pipe with bending radius of 6 inches, where the flexible probe had previously failed.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103482"},"PeriodicalIF":4.1,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144611602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Damage imaging using 2D teager-kaiser operator for early-time ultrasonic guided wavefields","authors":"Mohsen Barzegar,&nbsp;Muchao Zhang,&nbsp;Helena G. Ramos","doi":"10.1016/j.ndteint.2025.103485","DOIUrl":"10.1016/j.ndteint.2025.103485","url":null,"abstract":"<div><div>Ultrasonic wavefield imaging using wavefield measurement systems, such as scanning laser Doppler vibrometer systems, has proven to be a powerful non-destructive testing (NDT) technique. Yet, conventional energy-based imaging methods require bypassing the incident wave and reflection from borders to isolate trapped waves, which makes these approaches less effective in scenarios where excitation occurs within the inspection area and boundaries are closely situated, like an adhesive lap joint. This study introduces a damage imaging approach using the 2D Teager-Kaiser energy operator (TKO) applied to the wavefield and its square waveform. The proposed method exploits early wavefield propagation for direct damage imaging. This is particularly advantageous for disbond imaging, where the S0 mode is converted to A0 in the disbond region. In such cases, it enhances effectiveness in systems where measurements are more sensitive to the out-of-plane vibrations (A0) while the in-plane vibrations (S0) are weak. It is found that applying TKO to the wavefield achieves high contrast imaging compared to root-mean-square (RMS) calculations. The proposed method, applied to the square waveform of the wavefield, indirectly exploits its phase sensitivity to the disbond region and directly provides high contrast imaging, which provides an effective solution for high contrast disbond imaging using early wave propagation. Numerical simulation and experimental validation on different adhesive joints with a variety of disbond sizes demonstrate that the proposed technique effectively amplifies weak defect-induced interactions while reducing masking effects from direct wave propagation.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103485"},"PeriodicalIF":4.1,"publicationDate":"2025-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144595804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Active compensation method for restraining high-speed MFL signals distortion based on atypical synthesized magnetic field","authors":"Yuan Zhong,&nbsp;Gang Wang,&nbsp;Liwei Wang,&nbsp;Ce Liang,&nbsp;Xue Yu,&nbsp;Huaguang Zhang","doi":"10.1016/j.ndteint.2025.103476","DOIUrl":"10.1016/j.ndteint.2025.103476","url":null,"abstract":"<div><div>Magnetic flux leakage (MFL) detection is attaching importance in pipeline non-destructive detection. However, the MFL signals will be distorted due to the speed effect, resulting in a low detection accuracy. Reducing the detection speed and algorithmic compensation for the offline data are the two main current methods, but these methods reduce the detection efficiency and require a large amount of training data. In this article, an active compensation method for restraining high-speed MFL signals distortion based on atypical synthesized magnetic field (ASMF) is proposed to solve above problems. ASMF is a magnetic field with the asymmetric double peaks, whose shape and intensity can be complementary to the distorted MFL signals. First, the dynamic circuit theory is introduced to elaborate the speed effect mechanism. Second, the ASMF method is proposed to compensate the distorted MFL signals according to the speed. Furthermore, the target and non-target magnetic fields are corrected to satisfy the MFL detection under different lift-off. The relative optimum position between the main peak and side peak is studied to further reduce the MFL signals distortion. Third, the active compensation circuit system is developed, which can measure the speed and lift-off in real time to adjust ASMF and compensate the distorted MFL signals. Finally, the MFL detection experiments under different speed, lift-off, and defect are performed to verify the effectiveness of the proposed method. Results show that the minimum relative error of the defect depth detection using the compensated MFL signals can reach -2.33%, and Type A uncertainty <span><math><msub><mrow><mi>U</mi></mrow><mrow><mi>A</mi><mi>m</mi></mrow></msub></math></span> can be kept within 0.0346 mm.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103476"},"PeriodicalIF":4.1,"publicationDate":"2025-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144571863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing accuracy of an efficient analytical-finite element framework for modelling guided wave scattering from corrosion defects in pipes","authors":"Abdul Mateen Qadri ,&nbsp;Peter Huthwaite ,&nbsp;Michael Lowe ,&nbsp;Thomas Vogt","doi":"10.1016/j.ndteint.2025.103471","DOIUrl":"10.1016/j.ndteint.2025.103471","url":null,"abstract":"<div><div>Modelling guided ultrasound in pipelines is crucial for testing new approaches, understanding physical behaviour, and validating methods in non-destructive evaluation. Fully numerical methods like the finite element method accurately simulate ultrasonic inspections for realistic geometries but are limited by computational memory and processing needs, especially for large structures like long pipes. This work examines the accuracy of a methodology integrating analytical dispersion curves with finite element modelling. Dispersion curves model guided wave propagation over long, uniform sections of the pipe, whilst finite element modelling handles geometrical complexities such as defects. This integrated model maximises numerical modelling potential without overwhelming computational resources. The method was applied to analyse torsional T(0,1) mode scattering from corrosion defects using a GPU-based FEM solver and Fast-Fourier Transform decomposition of scattered wavefields. The effect of mesh refinement on performance of the integrated model is analysed, considering individual variations in axial, circumferential, and radial element sizes. These findings can significantly reduce the spatial domain needing numerical calculation, enhancing computational efficiency.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103471"},"PeriodicalIF":4.1,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144563624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Time domain extraction method of ultrasonic nonlinear coefficient","authors":"Bingsheng Yan ,&nbsp;Dongwen Pang ,&nbsp;Zhiyuan Huang","doi":"10.1016/j.ndteint.2025.103478","DOIUrl":"10.1016/j.ndteint.2025.103478","url":null,"abstract":"<div><div>During the extraction of the fundamental and second harmonic components, factors such as noise, spectral leakage, and the picket fence effect can reduce the accuracy of the ultrasonic nonlinearity coefficient <em>β</em> calculated using the Fast Fourier Transform (FFT) method. To address this issue, two time-domain extraction methods, Supplementary Zero Least Squares (SZ-LS) and Reduction of Noise Preprocessing Prony (RN-Prony) are proposed. The SZ-LS method employs supplementary zero on the truncated signal, ensuring that the number of data points satisfies an integer multiple relationship between the frequency resolution and the fundamental frequency, thereby reducing the impact of picket fence effect on the accuracy of the least squares calculation. The RN-Prony method utilizes The Fast Empirical Wavelet Transform (FEWT) to reduce noise in the signal, followed by Prony's method to calculate the amplitudes of the fundamental and second harmonic components. Simulation studies demonstrate that SZ-LS offers higher accuracy than FFT, but its results still fail to accurately obtain <em>β</em> within the error limits when the number of signal cycles is large. RN-Prony effectively reduces the influence of noise, improving the SNR of the processed signal by average of 54 % compared to the original signal, and achieves higher accuracy than the other two methods. The SZ-LS, RN-Prony and FFT methods are then applied to a set of martensitic stainless steel fatigue specimens and 6061-T6 aluminum alloy tensile specimens where the changes in the ultrasonic nonlinearity coefficient <em>β</em> with the degree of failure are determined. The <em>β</em> parameters processed with the RN-Prony method produce trends that make better physical sense than those processed with the SZ-LS method and FFT.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"156 ","pages":"Article 103478"},"PeriodicalIF":4.1,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144548480","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}