Ndt & E International最新文献_第5页

Mode-switchable computed laminography: System design and imaging analysis for plate-like objects 模式可切换计算机层析术：板状物体的系统设计和成像分析

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-11-29 DOI: 10.1016/j.ndteint.2025.103614

Chuandong Tan , Zhiting Chen , Qi Li , Ao Wang , Fenglin Liu , Yufang Cai , Liming Duan

Computed laminography (CL) overcomes the dual limitations of X-ray energy constraints and mechanical structural restrictions inherent in computed tomography (CT), enabling high-resolution nondestructive imaging of plate-like objects. Nevertheless, existing CL imaging systems are constrained to single scanning modes, failing to meet diverse application requirements or support research on relevant CL theory and reconstruction algorithms. To address these challenges, we design a mode-switchable CL system (MSCL) that seamlessly transitions between multiple scanning configurations and can achieve various scanning trajectories, including circular, linear, and composite trajectories. This system achieves theoretical detail resolution at the micrometer (μm) level. Meanwhile, a software tool “CLRecTool” is created to process the projection data of different CL scanning modes collected using the TIGRE toolbox, achieving CL image reconstruction. Simulation and actual experiments evaluate the effects of scanning trajectories and reconstruction algorithms on imaging quality. This versatility and scalability establish MSCL as a critical experimental platform for advancing CL imaging theory and algorithm development, while accelerating CL technology's adoption in multi-scenario industrial applications.

计算机层析成像（CL）克服了x射线能量限制和计算机断层扫描（CT）固有的机械结构限制的双重限制，实现了板状物体的高分辨率无损成像。然而，现有的CL成像系统仅限于单一的扫描模式，无法满足多样化的应用需求，也无法支持相关CL理论和重建算法的研究。为了应对这些挑战，我们设计了一种模式切换CL系统（MSCL），该系统可以在多种扫描配置之间无缝转换，并可以实现各种扫描轨迹，包括圆形、线性和复合轨迹。该系统达到微米（μm）级的理论细节分辨率。同时，创建了CLRecTool软件工具，对TIGRE工具箱采集的不同CL扫描方式的投影数据进行处理，实现CL图像重建。仿真和实际实验评估了扫描轨迹和重建算法对成像质量的影响。这种多功能性和可扩展性使MSCL成为推进CL成像理论和算法开发的关键实验平台，同时加速了CL技术在多场景工业应用中的采用。

{"title":"Mode-switchable computed laminography: System design and imaging analysis for plate-like objects","authors":"Chuandong Tan , Zhiting Chen , Qi Li , Ao Wang , Fenglin Liu , Yufang Cai , Liming Duan","doi":"10.1016/j.ndteint.2025.103614","DOIUrl":"10.1016/j.ndteint.2025.103614","url":null,"abstract":"<div><div>Computed laminography (CL) overcomes the dual limitations of X-ray energy constraints and mechanical structural restrictions inherent in computed tomography (CT), enabling high-resolution nondestructive imaging of plate-like objects. Nevertheless, existing CL imaging systems are constrained to single scanning modes, failing to meet diverse application requirements or support research on relevant CL theory and reconstruction algorithms. To address these challenges, we design a mode-switchable CL system (MSCL) that seamlessly transitions between multiple scanning configurations and can achieve various scanning trajectories, including circular, linear, and composite trajectories. This system achieves theoretical detail resolution at the micrometer (μm) level. Meanwhile, a software tool “CLRecTool” is created to process the projection data of different CL scanning modes collected using the TIGRE toolbox, achieving CL image reconstruction. Simulation and actual experiments evaluate the effects of scanning trajectories and reconstruction algorithms on imaging quality. This versatility and scalability establish MSCL as a critical experimental platform for advancing CL imaging theory and algorithm development, while accelerating CL technology's adoption in multi-scenario industrial applications.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103614"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691336","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}

引用次数: 0

Evaluation of AI for Eddy Current testing according to the reliability framework for rail inspection 基于钢轨检测可靠性框架的涡流检测人工智能评价

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-03 DOI: 10.1016/j.ndteint.2025.103611

D. Kanzler , G. Olm , A. Friedrich , M. Selch

The requirements for using AI algorithms are highly stringent for safety-critical and high-risk applications, such as in the field of non-destructive testing (NDT). Furthermore, there is a regulatory need for conclusive metrics to evaluate AI for high-risk applications. This article introduces a process for evaluating AI used in NDT methods. Based on a commonly used AI evaluation metric adapted for the NDT field, the evaluation aligns with the well-known NDT reliability processes. This evaluation process is applied to analyzing eddy current (ET) data in rail inspection. The aim is to quantify the capability of AI-supported data evaluation against that of the testing setup itself. However, the available real-world dataset of rail inspection data was insufficient for training, validation and demonstrating the reliability of the ET process and AI. To address this, simulated ET data was used to generate a large dataset for evaluation purposes. Using this simulated data, a reference probability of detection (POD) curve was created to provide a benchmark for assessing the performance of the AI using a newly introduced metric called Reliability Metric Score (RESa). The AI model analyzed ET data for crack-like defects. The results were then evaluated and compared to the reference POD. This article explores the evaluation process, highlighting potential misinterpretations and situations where an operator’s judgment is necessary to determine the effectiveness of the AI model in specific cases. This process revealed different regions of interest, which are very useful for further assessment of the AI process and continued development.

对于安全关键和高风险应用，例如无损检测（NDT）领域，使用人工智能算法的要求非常严格。此外，监管机构还需要制定结论性指标来评估人工智能在高风险应用中的应用。本文介绍了评估无损检测方法中使用的人工智能的过程。基于一种适用于无损检测领域的常用人工智能评估指标，该评估与众所周知的无损检测可靠性过程保持一致。将该评价过程应用于钢轨检测中涡流数据的分析。目的是将人工智能支持的数据评估能力与测试设置本身的能力进行量化。然而，现有的铁路检测数据的真实数据集不足以用于训练、验证和证明ET过程和人工智能的可靠性。为了解决这个问题，模拟的ET数据被用来生成一个用于评估目的的大型数据集。利用这些模拟数据，创建了一个参考检测概率（POD）曲线，为使用新引入的称为可靠性度量分数（RESa）的指标评估人工智能的性能提供基准。AI模型对ET数据进行裂纹类缺陷分析。然后对结果进行评估并与参考POD进行比较。本文探讨了评估过程，强调了潜在的误解和操作员判断在特定情况下确定人工智能模型有效性所必需的情况。这个过程揭示了不同的兴趣区域，这对进一步评估人工智能过程和持续发展非常有用。

{"title":"Evaluation of AI for Eddy Current testing according to the reliability framework for rail inspection","authors":"D. Kanzler , G. Olm , A. Friedrich , M. Selch","doi":"10.1016/j.ndteint.2025.103611","DOIUrl":"10.1016/j.ndteint.2025.103611","url":null,"abstract":"<div><div>The requirements for using AI algorithms are highly stringent for safety-critical and high-risk applications, such as in the field of non-destructive testing (NDT). Furthermore, there is a regulatory need for conclusive metrics to evaluate AI for high-risk applications. This article introduces a process for evaluating AI used in NDT methods. Based on a commonly used AI evaluation metric adapted for the NDT field, the evaluation aligns with the well-known NDT reliability processes. This evaluation process is applied to analyzing eddy current (ET) data in rail inspection. The aim is to quantify the capability of AI-supported data evaluation against that of the testing setup itself. However, the available real-world dataset of rail inspection data was insufficient for training, validation and demonstrating the reliability of the ET process and AI. To address this, simulated ET data was used to generate a large dataset for evaluation purposes. Using this simulated data, a reference probability of detection (POD) curve was created to provide a benchmark for assessing the performance of the AI using a newly introduced metric called Reliability Metric Score (RESa). The AI model analyzed ET data for crack-like defects. The results were then evaluated and compared to the reference POD. This article explores the evaluation process, highlighting potential misinterpretations and situations where an operator’s judgment is necessary to determine the effectiveness of the AI model in specific cases. This process revealed different regions of interest, which are very useful for further assessment of the AI process and continued development.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103611"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691396","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}

引用次数: 0

Non-uniform scan angle selection for detecting cells via CT CT检测细胞的非均匀扫描角度选择

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-01 DOI: 10.1016/j.ndteint.2025.103616

Han Yan , Long Chao , Liu Zefang , Tan Chuandong , Zhou Jianing , Tan Hui , Duan Liming

Sparse-angle scanning can achieve rapid detection of laminated cells by the computed tomography (CT). Typically, sparse-angle scanning uses equiangular sparse angle to scan the object, which may overlook information-rich regions of the internal structure, leading to partial structural loss of the reconstructed images. This paper proposes a non-uniform sparse scanning angle selection method by analysing the discrepancies between the full sinograms and the equiangular sparse sinograms. Firstly, the equiangular sparse sinograms are obtained by an equiangular sparse operator applied to the full sinograms of the laminated cells. Then, the sinograms error curve is obtained by comparing the full sinograms with the equiangular sparse sinograms. Finally, a scanning angle selection model is designed, which can select scanning angles. The optimal experiment results show that our method increases PSNR by 3.7053 and improves SSIM by 0.1797. These results demonstrate that our method improves the quality of CT images while using the same number of scanning angles. Our method offers a novel idea for obtaining clearer CT reconstruction image under CT rapid scanning.

稀疏角度扫描可以实现计算机断层扫描（CT）对层压细胞的快速检测。稀疏角扫描通常采用等角稀疏角对目标进行扫描，这可能会忽略内部结构中信息丰富的区域，导致重建图像的部分结构丢失。通过分析全角稀疏图与等角稀疏图的差异，提出了一种非均匀稀疏扫描角选择方法。首先，利用等角稀疏算子对层合单元的完整正弦图进行稀疏处理，得到等角稀疏正弦图；然后，将完整的正弦图与等角稀疏的正弦图进行对比，得到正弦图误差曲线。最后，设计了扫描角度选择模型，实现了扫描角度的选择。优化后的实验结果表明，该方法的PSNR提高了3.7053，SSIM提高了0.1797。结果表明，在相同扫描角度的情况下，我们的方法提高了CT图像的质量。该方法为在CT快速扫描下获得更清晰的CT重建图像提供了一种新的思路。

{"title":"Non-uniform scan angle selection for detecting cells via CT","authors":"Han Yan , Long Chao , Liu Zefang , Tan Chuandong , Zhou Jianing , Tan Hui , Duan Liming","doi":"10.1016/j.ndteint.2025.103616","DOIUrl":"10.1016/j.ndteint.2025.103616","url":null,"abstract":"<div><div>Sparse-angle scanning can achieve rapid detection of laminated cells by the computed tomography (CT). Typically, sparse-angle scanning uses equiangular sparse angle to scan the object, which may overlook information-rich regions of the internal structure, leading to partial structural loss of the reconstructed images. This paper proposes a non-uniform sparse scanning angle selection method by analysing the discrepancies between the full sinograms and the equiangular sparse sinograms. Firstly, the equiangular sparse sinograms are obtained by an equiangular sparse operator applied to the full sinograms of the laminated cells. Then, the sinograms error curve is obtained by comparing the full sinograms with the equiangular sparse sinograms. Finally, a scanning angle selection model is designed, which can select scanning angles. The optimal experiment results show that our method increases PSNR by 3.7053 and improves SSIM by 0.1797. These results demonstrate that our method improves the quality of CT images while using the same number of scanning angles. Our method offers a novel idea for obtaining clearer CT reconstruction image under CT rapid scanning.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103616"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691397","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}

引用次数: 0

Spherical helical trajectory CL: An improved strategy for circular trajectory cone-beam CL imaging 球面螺旋轨迹CL：一种改进的圆轨迹锥束CL成像策略

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-02 DOI: 10.1016/j.ndteint.2025.103615

Yanmin Sun , Yu Han , Lei Li , Xiaoqi Xi , Xuejing Lu , Siyu Tan , Linlin Zhu , Yuan Zhang , Bin Yan

Cone-beam computed laminography (CL) is an X-ray three-dimensional imaging technique designed for large plate-shaped objects. However, the commonly used circular-trajectory CL, featuring a tilted rotation axis, results in missing imaging data and leads to aliasing artifacts in reconstructed images. This study proposes spherical helical computed laminography (SHCL), a novel imaging approach that addresses CL data deficiencies and enhances image reconstruction quality. SHCL is achieved by progressively reducing the tilt angle of the rotation axis during circular trajectory rotation, allowing supplementary data acquisition from small tilt angles and significantly mitigating CL data loss. In commonly used CL imaging configurations with tilt angles ranging from 10° to 45°, SHCL can recover 43 %–69 % of the missing data of circular trajectory CL. This study develops an improved FDK reconstruction algorithm to accommodate the spherical helical trajectory of SHCL. Experimental results demonstrate that SHCL effectively reduces aliasing artifacts in CL imaging. The SHCL trajectory is simple to implement and does not introduce additional scanning workload, making it practical for real-world applications.

锥束计算机层析成像（CL）是一种针对大型板状物体设计的x射线三维成像技术。然而，通常使用的圆轨迹CL，由于其旋转轴倾斜，导致成像数据缺失，并导致重建图像中的混叠伪影。本研究提出了球面螺旋计算机层析成像（SHCL），这是一种新的成像方法，可以解决CL数据不足并提高图像重建质量。SHCL是通过在圆轨迹旋转过程中逐渐减小旋转轴的倾斜角来实现的，允许从小倾斜角获取补充数据，并显着减少CL数据丢失。在倾角为10°~ 45°的常用CL成像配置中，SHCL可以恢复圆轨迹CL丢失数据的43% ~ 69%。本文提出了一种改进的FDK重建算法，以适应SHCL的球面螺旋轨迹。实验结果表明，SHCL有效地降低了CL成像中的混叠伪影。SHCL轨迹易于实现，并且不会引入额外的扫描工作负载，使其适用于实际应用程序。

{"title":"Spherical helical trajectory CL: An improved strategy for circular trajectory cone-beam CL imaging","authors":"Yanmin Sun , Yu Han , Lei Li , Xiaoqi Xi , Xuejing Lu , Siyu Tan , Linlin Zhu , Yuan Zhang , Bin Yan","doi":"10.1016/j.ndteint.2025.103615","DOIUrl":"10.1016/j.ndteint.2025.103615","url":null,"abstract":"<div><div>Cone-beam computed laminography (CL) is an X-ray three-dimensional imaging technique designed for large plate-shaped objects. However, the commonly used circular-trajectory CL, featuring a tilted rotation axis, results in missing imaging data and leads to aliasing artifacts in reconstructed images. This study proposes spherical helical computed laminography (SHCL), a novel imaging approach that addresses CL data deficiencies and enhances image reconstruction quality. SHCL is achieved by progressively reducing the tilt angle of the rotation axis during circular trajectory rotation, allowing supplementary data acquisition from small tilt angles and significantly mitigating CL data loss. In commonly used CL imaging configurations with tilt angles ranging from 10° to 45°, SHCL can recover 43 %–69 % of the missing data of circular trajectory CL. This study develops an improved FDK reconstruction algorithm to accommodate the spherical helical trajectory of SHCL. Experimental results demonstrate that SHCL effectively reduces aliasing artifacts in CL imaging. The SHCL trajectory is simple to implement and does not introduce additional scanning workload, making it practical for real-world applications.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103615"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691400","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}

引用次数: 0

Non-destructive adhesive cure assessment in carbon fiber reinforced composites using dielectric spectroscopy 用介电光谱法评价碳纤维增强复合材料的无损粘接固化

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-01 DOI: 10.1016/j.ndteint.2025.103621

Minhazur Rahman , Monjur Morshed Rabby , Vamsee Vadlamudi , Rassel Raihan

Detecting the cure quality of adhesive in an adhesively bonded composite joint is crucial, as improper curing of adhesive in critical structural joints may compromise performance, durability, and service life. Since the composite adherend materials being bonded have poor thermal conductivity, ensuring that the adhesive reaches a proper cure temperature throughout the bonded region is challenging. While a large body of work exists on in-situ cure monitoring using thermocouples, dielectric, and other sensing elements, utilizing these sensing systems without compromising the adhesive layer is often challenging. Here, a frequency domain, non-destructive, post-cure assessment method has been explored using Broadband Dielectric Spectroscopy (BbDS) over 0.1Hz to 0.1 MHz range, which can identify whether the adhesive has appropriately been cured or not without exposing the bond. Bonded adhesive composite samples in lap-shear configuration with different levels of cure were manufactured by controlling the exposure time to the cure temperature. The stages of cure were verified using a Differential Scanning Calorimeter (DSC) and Fourier Transform Infrared Spectroscopy (FTIR). Dielectric spectroscopy revealed significant differences in the Dielectric Relaxation Strength (DRS) with different levels of cure. Mechanical testing of bonds was carried out and a proportionate correlation was observed with degree of cure of the adhesive. The kinetics of cure mechanisms were also studied in a temperature-frequency dependent Dielectric Spectroscopy.

检测粘接复合材料接头中胶粘剂的固化质量至关重要，因为关键结构接头中胶粘剂的不当固化可能会影响其性能、耐久性和使用寿命。由于被粘合的复合粘附材料导热性差，因此确保粘合剂在整个粘合区域达到适当的固化温度是具有挑战性的。虽然在使用热电偶、电介质和其他传感元件进行现场固化监测方面存在大量工作，但在不损害粘合剂层的情况下利用这些传感系统通常具有挑战性。在这里，我们探索了一种频域、非破坏性的固化后评估方法，该方法使用0.1 hz至0.1 MHz范围内的宽带介电光谱（BbDS），可以在不暴露粘合剂的情况下识别粘合剂是否已适当固化。通过控制固化温度下的暴露时间，制备了不同固化程度的搭剪型粘结复合材料样品。用差示扫描量热仪（DSC）和傅里叶变换红外光谱（FTIR）验证了固化阶段。电介质谱显示，不同固化水平的介质弛豫强度（DRS）有显著差异。进行了键的力学测试，并观察到与粘合剂固化程度成比例的相关性。在温度-频率相关的介电光谱中也研究了固化机理的动力学。

{"title":"Non-destructive adhesive cure assessment in carbon fiber reinforced composites using dielectric spectroscopy","authors":"Minhazur Rahman , Monjur Morshed Rabby , Vamsee Vadlamudi , Rassel Raihan","doi":"10.1016/j.ndteint.2025.103621","DOIUrl":"10.1016/j.ndteint.2025.103621","url":null,"abstract":"<div><div>Detecting the cure quality of adhesive in an adhesively bonded composite joint is crucial, as improper curing of adhesive in critical structural joints may compromise performance, durability, and service life. Since the composite adherend materials being bonded have poor thermal conductivity, ensuring that the adhesive reaches a proper cure temperature throughout the bonded region is challenging. While a large body of work exists on in-situ cure monitoring using thermocouples, dielectric, and other sensing elements, utilizing these sensing systems without compromising the adhesive layer is often challenging. Here, a frequency domain, non-destructive, post-cure assessment method has been explored using Broadband Dielectric Spectroscopy (BbDS) over 0.1Hz to 0.1 MHz range, which can identify whether the adhesive has appropriately been cured or not without exposing the bond. Bonded adhesive composite samples in lap-shear configuration with different levels of cure were manufactured by controlling the exposure time to the cure temperature. The stages of cure were verified using a Differential Scanning Calorimeter (DSC) and Fourier Transform Infrared Spectroscopy (FTIR). Dielectric spectroscopy revealed significant differences in the Dielectric Relaxation Strength (DRS) with different levels of cure. Mechanical testing of bonds was carried out and a proportionate correlation was observed with degree of cure of the adhesive. The kinetics of cure mechanisms were also studied in a temperature-frequency dependent Dielectric Spectroscopy.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103621"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691401","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}

引用次数: 0

On acoustic fields of Lamb wave scattering in plates based on Convolutional Neural Network-Transformer 基于卷积神经网络-变压器的板材Lamb波散射声场研究

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-23 DOI: 10.1016/j.ndteint.2025.103628

Zhen Zhang , Linfeng Wang , Jian Li , Zhoumo Zeng , Yang Liu

This paper presents a hybrid deep learning model that combines Convolutional Neural Network (CNN) and Transformer to enable efficient prediction of far-field scattered signals of S0 mode Lamb waves from defects of thin plates. The proposed model combines CNN for local spatial feature extraction with Transformer to model global temporal dependencies, enhancing the ability to predict scattering from irregularly-shaped defects beyond the limitations of traditional methods. A three-dimensional (3D) finite element model of an aluminum plate with irregularly-shaped defects was developed to generate scattering fields with diverse morphologies and parameters for model training and testing. CNN-Transformer model successfully predicted the scattering behavior of S0 mode Lamb wave, demonstrating high accuracy in scenarios with irregularly-shaped defects. The model's performance was further validated through laser Doppler experiments, demonstrating strong consistency with the predicted scattering characteristics. Furthermore, the model was extended to solve the scattering matrix, enabling accurate prediction of scattered signals across multiple incident angles. This study introduces a new approach to defect scattering in ultrasonic guided wave detection. It provides both theoretical insights and practical support for engineering applications.

本文提出了一种结合卷积神经网络（CNN）和变压器（Transformer）的混合深度学习模型，能够有效预测薄板缺陷的50模兰姆波远场散射信号。该模型结合CNN局部空间特征提取和Transformer建模全局时间依赖性，提高了不规则形状缺陷散射预测能力，超越了传统方法的局限性。建立了不规则缺陷铝板的三维有限元模型，生成了具有多种形态和参数的散射场，用于模型训练和测试。CNN-Transformer模型成功地预测了S0模式Lamb波的散射行为，在不规则形状缺陷情况下显示出较高的精度。通过激光多普勒实验进一步验证了模型的性能，与预测的散射特性有较强的一致性。此外，将模型扩展到求解散射矩阵，实现了对多个入射角散射信号的准确预测。介绍了超声导波检测中缺陷散射的一种新方法。它为工程应用提供了理论见解和实践支持。

{"title":"On acoustic fields of Lamb wave scattering in plates based on Convolutional Neural Network-Transformer","authors":"Zhen Zhang , Linfeng Wang , Jian Li , Zhoumo Zeng , Yang Liu","doi":"10.1016/j.ndteint.2025.103628","DOIUrl":"10.1016/j.ndteint.2025.103628","url":null,"abstract":"<div><div>This paper presents a hybrid deep learning model that combines Convolutional Neural Network (CNN) and Transformer to enable efficient prediction of far-field scattered signals of S0 mode Lamb waves from defects of thin plates. The proposed model combines CNN for local spatial feature extraction with Transformer to model global temporal dependencies, enhancing the ability to predict scattering from irregularly-shaped defects beyond the limitations of traditional methods. A three-dimensional (3D) finite element model of an aluminum plate with irregularly-shaped defects was developed to generate scattering fields with diverse morphologies and parameters for model training and testing. CNN-Transformer model successfully predicted the scattering behavior of S0 mode Lamb wave, demonstrating high accuracy in scenarios with irregularly-shaped defects. The model's performance was further validated through laser Doppler experiments, demonstrating strong consistency with the predicted scattering characteristics. Furthermore, the model was extended to solve the scattering matrix, enabling accurate prediction of scattered signals across multiple incident angles. This study introduces a new approach to defect scattering in ultrasonic guided wave detection. It provides both theoretical insights and practical support for engineering applications.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103628"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145839675","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}

引用次数: 0

A contribution by gradient explainability method for 1D-CNNs on ultrasonic data 梯度可解释性方法对超声数据一维cnn的贡献

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-11-19 DOI: 10.1016/j.ndteint.2025.103592

Yuyang Liu, Sergio Cantero-Chinchilla, Anthony J. Croxford

This paper proposes the Contribution by Gradients (C-grad) method for interpreting neural network models applied to regression problems on ultrasonic data. As machine learning applications continue to expand in Non-Destructive Testing (NDT), the “black-box” nature of neural networks raises concerns about the consistency and interpretability of AI-generated solutions in industrial applications. The C-grad method addresses these challenges by quantifying input contributions through their rate of change across layers, providing detailed layer-by-layer insights into model decisions. Unlike traditional gradient algorithms focused solely on final outputs, C-grad backward propagates activation changes from intermediate layers to input arrays, enabling a more detailed breakdown of feature importance at each layer and enhancing interpretability. Applied to 1-dimensional convolutional neural networks (1D-CNNs) for ultrasonic A-scan time-traces data in corrosion profiling, the method shows superior stability and consistency over the traditional Gradient-based Class Activation Map method (Grad-CAM). By combining explanation infidelity testing with a peak-to-peak distance metric that correlates model explanations with physical echo features, C-grad offers a comprehensive framework for assessing model robustness to corrupted inputs. C-grad has proven successful in identifying crucial physical features in the A-scan time-traces for the 1D-CNNs trained on two corrosion profiling metrics: mean material thickness and roughness. These case studies under C-grad investigation provide detailed insights into model trustworthiness and guide architecture optimisation in ML-driven NDT.

本文提出了梯度贡献（C-grad）方法来解释应用于超声数据回归问题的神经网络模型。随着机器学习应用在无损检测（NDT）领域的不断扩展，神经网络的“黑箱”性质引发了人们对工业应用中人工智能生成解决方案的一致性和可解释性的担忧。C-grad方法通过通过各层的变化率量化输入贡献来解决这些挑战，为模型决策提供详细的逐层洞察。与传统的梯度算法只关注最终输出不同，C-grad将激活变化从中间层向后传播到输入数组，从而能够更详细地分解每一层的特征重要性，并增强可解释性。将该方法应用于一维卷积神经网络（1d - cnn）的腐蚀剖面超声a扫描时间轨迹数据中，与传统的基于梯度的类别激活图方法（Grad-CAM）相比，该方法具有更好的稳定性和一致性。通过将解释不忠测试与峰对峰距离度量（将模型解释与物理回波特征相关联）相结合，C-grad为评估模型对损坏输入的鲁棒性提供了一个全面的框架。C-grad已经被证明成功地识别了d - cnn在两个腐蚀剖面指标（平均材料厚度和粗糙度）上训练的a扫描时间轨迹中的关键物理特征。C-grad调查下的这些案例研究提供了对模型可信度的详细见解，并指导了机器学习驱动无损检测的架构优化。

{"title":"A contribution by gradient explainability method for 1D-CNNs on ultrasonic data","authors":"Yuyang Liu, Sergio Cantero-Chinchilla, Anthony J. Croxford","doi":"10.1016/j.ndteint.2025.103592","DOIUrl":"10.1016/j.ndteint.2025.103592","url":null,"abstract":"<div><div>This paper proposes the Contribution by Gradients (C-grad) method for interpreting neural network models applied to regression problems on ultrasonic data. As machine learning applications continue to expand in Non-Destructive Testing (NDT), the “black-box” nature of neural networks raises concerns about the consistency and interpretability of AI-generated solutions in industrial applications. The C-grad method addresses these challenges by quantifying input contributions through their rate of change across layers, providing detailed layer-by-layer insights into model decisions. Unlike traditional gradient algorithms focused solely on final outputs, C-grad backward propagates activation changes from intermediate layers to input arrays, enabling a more detailed breakdown of feature importance at each layer and enhancing interpretability. Applied to 1-dimensional convolutional neural networks (1D-CNNs) for ultrasonic A-scan time-traces data in corrosion profiling, the method shows superior stability and consistency over the traditional Gradient-based Class Activation Map method (Grad-CAM). By combining explanation infidelity testing with a peak-to-peak distance metric that correlates model explanations with physical echo features, C-grad offers a comprehensive framework for assessing model robustness to corrupted inputs. C-grad has proven successful in identifying crucial physical features in the A-scan time-traces for the 1D-CNNs trained on two corrosion profiling metrics: mean material thickness and roughness. These case studies under C-grad investigation provide detailed insights into model trustworthiness and guide architecture optimisation in ML-driven NDT.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103592"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145555258","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}

引用次数: 0

A method for conductivity measurement through eddy current testing and closed-loop feedforward control 一种通过涡流测试和闭环前馈控制的电导率测量方法

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-11-29 DOI: 10.1016/j.ndteint.2025.103612

Yuedong Xie , Xiaoxuan Geng , Pu Huang , Yong Yan

Metallic plates are widely used in aerospace, transportation, and the chemical industry. Conductivity is a significant physical property of metal materials, and material aging and corrosion can be detected in timely to ensure the safe and stable operation of industrial equipment and systems through monitoring variation of conductivity. In order to achieve high-precision conductivity measurement, this paper innovatively investigated an eddy current testing (ECT) method incorporating a feedforward Proportion-Integration-Differentiation (PID) controller. Specifically, a closed-loop PID controller is introduced to act on classical Dodd-Deeds analytical model. Conductivity can be accurately inverted by continuously reducing the deviation between the output of the analytical model and actual measurements. Considering the lift-off fluctuation affects the conductivity measurement accuracy, a feedforward controller is designed to reduce the lift-off distance variation, and the transfer function of the feedforward controller can be obtained by theoretical deduction. Experiments are also conducted to verify the proposed method. Results indicate the maximum relative error of conductivity measurement remains merely 1.64 % across the 3 mm lift-off range, demonstrating the efficiency and reliability of the proposed method.

金属板广泛应用于航空航天、交通运输和化学工业。电导率是金属材料的重要物理性能，通过监测电导率的变化，可以及时发现材料的老化和腐蚀，保证工业设备和系统的安全稳定运行。为了实现高精度电导率测量，本文创新性地研究了一种结合前馈比例-积分-微分（PID）控制器的涡流测试（ECT）方法。具体来说，在经典的Dodd-Deeds分析模型上引入了闭环PID控制器。通过不断减小分析模型输出与实际测量值之间的偏差，可以准确地反演电导率。考虑到升力波动对电导率测量精度的影响，设计了前馈控制器以减小升力距离变化，并通过理论推导得到了前馈控制器的传递函数。实验也验证了所提出的方法。结果表明，在3 mm上升范围内，电导率测量的最大相对误差仅为1.64%，证明了该方法的有效性和可靠性。

{"title":"A method for conductivity measurement through eddy current testing and closed-loop feedforward control","authors":"Yuedong Xie , Xiaoxuan Geng , Pu Huang , Yong Yan","doi":"10.1016/j.ndteint.2025.103612","DOIUrl":"10.1016/j.ndteint.2025.103612","url":null,"abstract":"<div><div>Metallic plates are widely used in aerospace, transportation, and the chemical industry. Conductivity is a significant physical property of metal materials, and material aging and corrosion can be detected in timely to ensure the safe and stable operation of industrial equipment and systems through monitoring variation of conductivity. In order to achieve high-precision conductivity measurement, this paper innovatively investigated an eddy current testing (ECT) method incorporating a feedforward Proportion-Integration-Differentiation (PID) controller. Specifically, a closed-loop PID controller is introduced to act on classical Dodd-Deeds analytical model. Conductivity can be accurately inverted by continuously reducing the deviation between the output of the analytical model and actual measurements. Considering the lift-off fluctuation affects the conductivity measurement accuracy, a feedforward controller is designed to reduce the lift-off distance variation, and the transfer function of the feedforward controller can be obtained by theoretical deduction. Experiments are also conducted to verify the proposed method. Results indicate the maximum relative error of conductivity measurement remains merely 1.64 % across the 3 mm lift-off range, demonstrating the efficiency and reliability of the proposed method.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103612"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145691398","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}

引用次数: 0

Inductance-based method evaluation of steel fiber distribution and orientation in hybrid-reinforced segment for shield tunnel linings 盾构隧道衬砌混合增强管片中钢纤维分布和取向的电感法评价

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-11-25 DOI: 10.1016/j.ndteint.2025.103610

Quan Zhang, Xiaohui Zhang, Hule Li, Shunhua Zhou, Naichen Shi, Chao He, Honggui Di

Rebar and steel fiber hybrid reinforced concrete (RC-SFRC) segments are gaining prominence in large-cross-section subway shield tunnel construction due to superior durability and crack resistance. However, conventional methods for evaluating steel fiber distribution and orientation face destructiveness and complexity, limiting large-volume concrete segment quality control applications. This study develops inductance-based techniques to enable nondestructive assessment of fiber distribution and orientation characteristics. On this basis, the research analyzes inductive method measurement-related engineering affecting factors, investigates fiber distribution characteristics and their effect on crack resistance in RC-SFRC segments. Experimental results demonstrate that inductance techniques exhibit robustness and efficiency for steel fiber characterization in tunnel segment applications. RC-SFRC segments' representative specimen dimensions of 50 mm radius for cylindrical and 90 mm edge length for cubic samples effectively reduce fiber content inductance-testing dispersion. Steel fiber distribution and orientation in the RC-SFRC segment show preferential distributions, featuring two significantly thickness-dependent dose distributions while preferentially orienting in the central angle direction, thereby reducing outer arc layer crack resistance capacity. Inductive-based orientation characterization can effectively communicate Barcelona testing fracture surfaces and post-cracking performance, significantly advancing steel fiber distribution characteristics and structural mechanical properties evaluation in tunnel segments.

钢筋和钢纤维混合钢筋混凝土（RC-SFRC）管段由于其优异的耐久性和抗裂性，在大断面地铁盾构隧道施工中越来越受到重视。然而，传统的评价钢纤维分布和取向的方法具有破坏性和复杂性，限制了大体积混凝土管片质量控制的应用。本研究开发了基于电感的技术，使光纤分布和取向特性的无损评估成为可能。在此基础上，分析了电感法测量相关工程影响因素，研究了RC-SFRC段纤维分布特性及其对抗裂性能的影响。实验结果表明，电感技术在隧道段钢纤维表征中具有鲁棒性和有效性。RC-SFRC节段的代表性试样尺寸为圆柱形试样半径为50mm，立方试样边长为90mm，有效降低了纤维含量的电感测试色散。钢纤维在RC-SFRC段的分布和取向呈现优先分布，具有两种明显的厚度依赖剂量分布，同时优先向中心角方向取向，从而降低了外弧层抗裂能力。基于感应的取向表征可以有效地传达巴塞罗那测试断口表面和开裂后的性能，显著提高了钢纤维在隧道管段中的分布特征和结构力学性能评估。

{"title":"Inductance-based method evaluation of steel fiber distribution and orientation in hybrid-reinforced segment for shield tunnel linings","authors":"Quan Zhang, Xiaohui Zhang, Hule Li, Shunhua Zhou, Naichen Shi, Chao He, Honggui Di","doi":"10.1016/j.ndteint.2025.103610","DOIUrl":"10.1016/j.ndteint.2025.103610","url":null,"abstract":"<div><div>Rebar and steel fiber hybrid reinforced concrete (RC-SFRC) segments are gaining prominence in large-cross-section subway shield tunnel construction due to superior durability and crack resistance. However, conventional methods for evaluating steel fiber distribution and orientation face destructiveness and complexity, limiting large-volume concrete segment quality control applications. This study develops inductance-based techniques to enable nondestructive assessment of fiber distribution and orientation characteristics. On this basis, the research analyzes inductive method measurement-related engineering affecting factors, investigates fiber distribution characteristics and their effect on crack resistance in RC-SFRC segments. Experimental results demonstrate that inductance techniques exhibit robustness and efficiency for steel fiber characterization in tunnel segment applications. RC-SFRC segments' representative specimen dimensions of 50 mm radius for cylindrical and 90 mm edge length for cubic samples effectively reduce fiber content inductance-testing dispersion. Steel fiber distribution and orientation in the RC-SFRC segment show preferential distributions, featuring two significantly thickness-dependent dose distributions while preferentially orienting in the central angle direction, thereby reducing outer arc layer crack resistance capacity. Inductive-based orientation characterization can effectively communicate Barcelona testing fracture surfaces and post-cracking performance, significantly advancing steel fiber distribution characteristics and structural mechanical properties evaluation in tunnel segments.</div></div>","PeriodicalId":18868,"journal":{"name":"Ndt & E International","volume":"159 ","pages":"Article 103610"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145622286","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}

引用次数: 0

Corrigendum to “Multiporous-Cascaded coil based high lift-off and dynamic electromagnetic thermography of rail defects inspection” [NDT E Int 159C (2025) 103599] “基于多孔级联线圈的高升力和动态电磁热成像轨道缺陷检测”的勘误表[NDT E Int 159C (2025) 103599]

IF 4.5 2区材料科学 Q1 MATERIALS SCIENCE, CHARACTERIZATION & TESTING

Ndt & E International

Pub Date : 2026-04-01 Epub Date: 2025-12-08 DOI: 10.1016/j.ndteint.2025.103619

Geng Yang, Haoran Li, Bin Gao, Xiaolong Lu, Junhong Qi, Dong Liu, Guiyun Tian, Xiaojie Xue, Xingcai Liu

引用次数: 0