Journal of Nondestructive Evaluation最新文献

{"title":"A Novel Approach for Three-Dimensional Acoustic Source Localization in Orthotropic Medium","authors":"Jérôme S. Afoutou,&nbsp;Frédéric Lamy,&nbsp;Frédéric Dubois","doi":"10.1007/s10921-026-01358-2","DOIUrl":"10.1007/s10921-026-01358-2","url":null,"abstract":"<div><p>This work investigates the three-dimensional localization of acoustic emission sources in a cartesian orthotropic medium using a geometric approach. The proposed method minimizes a constrained nonlinear objective function formulated from arrival times, computed via the 3D Christoffel equation for wave propagation in orthotropic solids. Acoustic test points (ATPs) and emission times were randomly generated to evaluate the robustness of the localization procedure. Localization was performed under two scenarios: first, assuming known material parameters; and second, jointly estimating both the source coordinates and the material parameters. This increased the number of unknowns from four (source position and emission time) to ten, with the six additional variables corresponding to the 3D Hankinson parameters characterizing wave velocity anisotropy. The study examined the effects of both number of sensors (ranging from 4 to 14) and spatial configuration (eight distinct arrangements) on localization accuracy. Two global optimization algorithms (DIRECT and differential evolution) were employed to minimize the objective function. Results have revealed that localization performance is influenced more by sensor placement than by sensors quantity. The differential evolution algorithm exhibited greater robustness with and less sensitivity to sensor positioning. The findings confirmed the effectiveness of the proposed methodology for accurate acoustic source localization and concurrent estimation of material parameters in orthotropic media.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147607294","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Defect Localization in Composites Using AI-Augmented Thz Imaging","authors":"Battipally Jeevakranthi Goud,&nbsp;Bobba Krishna Sai,&nbsp;M. Muth Tamil Pooja,&nbsp;S. Vinson Joshua,&nbsp;A. Mercy Latha","doi":"10.1007/s10921-026-01359-1","DOIUrl":"10.1007/s10921-026-01359-1","url":null,"abstract":"<div><p>Nondestructive testing (NDT) of composite materials presents significant challenges due to their complex, anisotropic, and heterogeneous internal structures, which often limit the efficacy of conventional inspection techniques. In this study, a terahertz (THz) imaging system is employed to address these limitations by enabling identification of subsurface and bulk defects in composite materials. However, the THz images are highly noisy with low resolution and high background fluctuations. Hence, for automated defect localization, it is crucial to employ the state-of-the-art artificial intelligent algorithms yielding in AI augmentation of the THz images. Here, a dataset comprising over 600 THz images, categorized into eighteen distinct defect classes, has been employed for automated defect localization studies. Due to the poor THz image quality, it becomes extremely challenging to localize the defects in the composites manually or visually. Therefore, it is essential to develop an efficient AI-based framework capable of localizing defects with increased accuracy. Three object detection frameworks—Region-based Convolutional Neural Network (RCNN), YOLOv8l, and Faster R-CNN — have been evaluated in terms of their detection accuracy, localization precision, and computational efficiency. The RCNN model achieved a maximum precision and recall of 0.9738 and 0.9737, respectively, while YOLOv8l attained a precision and recall of 0.9550 and 0.9459, respectively. Faster R-CNN attained perfect classification scores, with precision and recall of 1.0000, and outperformed other models in terms of inference speed, requiring only 1.2 ms per image, which has not been reported elsewhere. Given its superior balance of classification performance, localization, and real-time inference capability, Faster R-CNN has been identified as the most effective model for THz-based defect detection in composites. However, YOLOv8l demonstrated superior localization accuracy with the highest mean IoU of 0.9872. These findings demonstrate the potential of AI augmented THz imaging in enabling robust, automated, and scalable NDT solutions for advanced composite inspections.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147607343","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Pipeline Defect Size Prediction Method Integrating Simulation-Driven, Data Augmentation, and Physics-Guided Approaches","authors":"Xulai Zhu,&nbsp;Yuxiang Zhang,&nbsp;Qiansheng Fang,&nbsp;Jin Jiang,&nbsp;Gongquan Zhang","doi":"10.1007/s10921-026-01355-5","DOIUrl":"10.1007/s10921-026-01355-5","url":null,"abstract":"<div><p>Magnetic Flux Leakage (MFL) testing is a vital technique for the quantitative assessment of internal defect dimensions in oil and gas pipelines. However, existing inversion methods face dual challenges: traditional physical forward computational models are excessively time-consuming, while purely data-driven deep learning models (i.e., “black-box” models) typically lack physical interpretability and generalization capability. To address these issues, this paper proposes a novel pipeline defect size prediction framework that integrates simulation-driven modeling, data augmentation, and physics-guided mechanisms. First, a defect dataset containing dual-channel MFL signals (axial and radial components) is constructed via finite element simulation. Subsequently, a Conditional Variational Autoencoder (CVAE) is employed to augment the dataset, thereby overcoming the bottleneck of scarce high-fidelity defect samples. Finally, a dual-channel Convolutional Neural Network (CNN) is constructed, into which a Physical Proxy Model (PPM) is introduced. The PPM embeds the “dimension-signal” physical mapping mechanism into the network as a loss constraint, enforcing the model to strictly adhere to underlying physical laws during prediction. Experimental results demonstrate that the proposed method achieves superior performance in predicting defect length, width, and depth, yielding an overall Mean Absolute Error (MAE) of 0.7525 mm and a Root Mean Square Error (RMSE) of 1.2128 mm. Compared to the benchmark LSTM model, the MAE is significantly reduced by 14.15%. The proposed framework realizes an effective fusion of data-driven and physics-driven mechanisms, offering a novel paradigm for defect prediction in complex scenarios where complete physical equations are unavailable.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147607281","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhanced EfficientNet-B0 with Dual Attention Mechanisms for Food Category Classification in X-ray Images","authors":"Jianfeng Yao,&nbsp;Hengyuan Liu,&nbsp;Junchao Ye","doi":"10.1007/s10921-026-01348-4","DOIUrl":"10.1007/s10921-026-01348-4","url":null,"abstract":"<div><p>To improve the operational efficiency of food foreign object inspection systems, this study proposes a food category recognition model for X-ray images. 6800 X-ray images containing 34 types of food were collected. Image labels were automatically assigned according to the folder corresponding to each food category. The dataset was divided into training, validation, and testing sets in an 8:1:1 ratio. The model was based on EfficientNet-B0. To enhance the representation capability of key image regions, Convolutional Block Attention Module (CBAM) and Efficient Channel Attention Module (ECA) have been added to EfficientNet-B0. To improve the robustness of the model to noise and image variability, the images were enhanced through various methods such as random scaling, horizontal flipping, rotation, occlusion and so on. The proposed model achieved a test accuracy of 96.60%, which is 2.03% higher than ResNet-50, 1.56% higher than the baseline EfficientNet-B0, and 2.14% higher than MobileNet-V2. The test results indicate that the proposed method can provide accurate, efficient, and automated technical solutions for food classification in X-ray images.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147562050","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dynamic Laser Speckle Imaging for Quantifying Particle Size During Intermittent Stages of Natural Rubber Latex Coagulation","authors":"S. H. Keerthana,&nbsp;A. K. Sooraj Viswam,&nbsp;P. Radhakrishnan,&nbsp;A. Mujeeb","doi":"10.1007/s10921-026-01349-3","DOIUrl":"10.1007/s10921-026-01349-3","url":null,"abstract":"<div><p>Dynamic laser speckle imaging is a growing optical imaging technique used for exploring the dynamics of colloidal dispersions. High resolution in the spatial and temporal domain is the major attraction of this technique for analyzing complex dynamic processes of the materials. This imaging technique is both non-destructive and cost-effective, making it a valuable choice for various applications. Conventional image processing methods incorporated with the technique include Time history of speckle patterns and Co-occurrence matrices. These methods gather information from certain regions of the speckle patterns only. To avoid this limitation, frequency domain techniques are employed which isolate the areas of different activities in the complex materials. Hence, in this work, we employed Wavelet based entropy for the analysis of speckle images captured for the intermittent stages of the coagulation process of natural rubber latex. Wavelet based entropy is more effective for non-stationary processes when compared to the Fourier transform, which is applicable only to stationary processes. The results were also compared with traditional methods in the temporal domain such as Time history of speckle patterns, Co-occurrence matrices and Second-order intensity Autocorrelation function and showed agreement. Second-order intensity Autocorrelation function was employed to quantify particle size distribution of natural rubber latex during different stages of the coagulation process.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147561375","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generalized Ultrasonic TOFD Techniques for Detecting Axial Defects in Pipelines","authors":"Shijie Jin,&nbsp;Shuyao Huangfu,&nbsp;Junyi Peng,&nbsp;Zhongbing Luo","doi":"10.1007/s10921-026-01350-w","DOIUrl":"10.1007/s10921-026-01350-w","url":null,"abstract":"<div><p>The time-of-flight diffraction (TOFD) technique in ultrasonic testing is limited in detecting axial defects in pipelines due to the increased quantitative errors induced by pipeline curvature. The generalized TOFD techniques involving diffracted longitudinal wave, mode-converted wave, and LS-L wave (L-longitudinal and S-shear waves) are developed to detect and quantify axial defects by establishing the three-dimensional relationship between eccentric distance, defect depth, and the corresponding wave propagation time. The axial cracks beyond and within dead zone under different pipeline parameters (e.g., curvature radius and material) and detection conditions (e.g., probe center separation (PCS) and eccentric distance) are detected quantitatively by combining the obtained theoretical three-dimensional relationship. The simulated and experimental measurement errors are no more than 4.7% and 4.9%, respectively. Finally, the quantitative errors under different eccentric distances using the previous and newly proposed generalized TOFD techniques are compared to indicate the latter's advantages, and another generalized TOFD technique is investigated to detect axial defects.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147561174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Elastic Registration for Non-Destructive Testing of Additively Manufactured TPMS Lattice Structures","authors":"Ammar Alsheghri,&nbsp;Ahmed Malaekah,&nbsp;Abdulrahman Alhendawi,&nbsp;Mohd Tauheed,&nbsp;Jana El Bsat,&nbsp;Hristo Valchanov,&nbsp;Vladimir Brailovski,&nbsp;Francois Guibault","doi":"10.1007/s10921-026-01342-w","DOIUrl":"10.1007/s10921-026-01342-w","url":null,"abstract":"<div><p>In additive manufacturing, defects lead to deviations between the as-designed versus the as-manufactured product. This challenge is pronounced for intricate lattice structures such as triply periodic minimal surface (TPMS) lattices; where hidden features complicate the quality assurance procedure. Micro computed tomography based on x-rays is the best method for the non-destructive testing of such lattice structures. However, this method is costly and time-consuming. Here, we propose a framework to accelerate the non-destructive testing of such lattices by reducing the time needed from a few hours for scanning to approximately 100-200 seconds. We propose an elastic registration framework that will use a few computed tomography (CT) projections to deform the prior available computer aided design (CAD) model such that an as-manufactured mesh is obtained. This mesh can then be used for quality control and deviation analysis. In addition, our work proposes the use of surface determination method to reconstruct the mesh for validation, geometry x-ray simulator to reduce the time and cost, and a learning-based approach for elastic registration. To demonstrate our algorithm, we conduct a case study on using TPMS lattice structures for bone-replacement applications such as scaffolds and implants. We compare three different gyroid-based geometries (i.e. sheet-based, strut-based, and hybrid) with trabecular bone structure. We also demonstrate that a hybrid lattice structure made of a combination of sheets and struts presents better manufacturability and is more suitable to construct scaffolds for bone replacement applications. The proposed framework contributes to industry 5.0 where accelerated and accurate quality assurance aid human-centric decisions.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147561059","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on Magnetic Memory Detection of Stress States in Bridge Steel Structure Welds under Bending Stress","authors":"Zhenfeng He,&nbsp;Shike Zhang,&nbsp;Chongwei Ran,&nbsp;Tianyu Hu,&nbsp;Hong Zhang","doi":"10.1007/s10921-026-01346-6","DOIUrl":"10.1007/s10921-026-01346-6","url":null,"abstract":"<div><p>The accurate detection of stress states in bridge steel welds is of critical significance for ensuring structural safety and durability. In this study, metal magnetic memory testing (MMMT) technology is innovatively introduced to detect and evaluate stress states in steel welds. By conducting magnetic detection experiments on two categories of welds (comprising eight specimens) under three-point bending, the effects of variables such as detection paths and specimen differences (structure type and initial magnetic state) on the distribution of the magnetic field were analyzed, and the evolutionary patterns of magnetic field under varying stress conditions were elucidated. Drawing upon the shared characteristics observed in the magnetic field variations of both weld categories, the relative entropy and the gradient Euclidean norm of magnetic memory signals were extracted as key magnetic characteristic parameters in the study. The results show that relative entropy changes gradually during the elastic stage but increases sharply upon transition into plastic deformation, with its rate of rise markedly exceeding that in the elastic phase. This finding indicates that relative entropy may serve as a reliable indicator for accurately identifying the stress states of the structure. Furthermore, there is a high linear correlation between the Euclidean norm of the magnetic signal gradient and elastic stress. By employing optimized feature parameters, structural stress can be effectively quantified, with most measurement errors maintained within 20%. These findings provide a viable technical approach and methodological reference for the stress detection of bridge steel welds.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 2","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147561060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"D3-Net: Dynamic Defect Detection Network for Identifying Defects in Multifaceted Material of Automotive Seats","authors":"Rui Sun,&nbsp;Dejin Zhao,&nbsp;Xiaolong Yuan,&nbsp;Yunjie Ma,&nbsp;Jingzhe Zhang,&nbsp;Xu Zhu,&nbsp;Yi Xing,&nbsp;Jialin Li,&nbsp;Bo Li,&nbsp;Dexin Kong,&nbsp;Jianhai Zhang,&nbsp;Xiangliang Kong","doi":"10.1007/s10921-026-01351-9","DOIUrl":"10.1007/s10921-026-01351-9","url":null,"abstract":"<div><p>Automobile seats not only fulfill the traditional functions of comfort and safety but also increasingly incorporate various intelligent and structural features, which make their structure more complicated. This increasing structural complexity leads to the emergence of various defects during the manufacturing process, thereby necessitating the implementation of precise defect detection methods within automated production systems. In response to this challenge, we have developed an intelligent detection system aimed at identifying defects on seat surfaces and proposed the Dynamic Defect Detection (D³-Net) model as the algorithm for this system. The D³-Net model integrates a Dynamic Position-Aware Network (DPAN) to effectively capture multi-scale global and local information. Additionally, the Multi-Dimensional Adaptive Interaction Module (MDAIM) provides depth supervision aimed at minimizing background interference. To mitigate the effects of multifaceted materials on defect extraction, the Dual-Stream Dynamic Attention Fusion Module (D2AFM) consolidates feature information across various hierarchical levels. Utilizing our self-constructed dataset for leather seating surface defects, we demonstrate that D³-Net surpasses the Transformer-based DDQ-DETR model by 3.1% in Average Precision (AP). Further validation with a fabric seating surface defect dataset reveals a 3.2% improvement in AP over the Transformer-based DINO model, thereby confirming D³-Net's robust generalization capability. This model effectively extracts defect features and minimizes the influence of multifaceted backgrounds.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 1","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147559989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Global Attention-Guided Self-Calibrating Tire Defect Detection Using Deep Learning","authors":"Yue Chen,&nbsp;Nian Wang,&nbsp;Weiang Li,&nbsp;Liyang Zhang,&nbsp;Mengxin Chen,&nbsp;Jinghong Tian","doi":"10.1007/s10921-026-01345-7","DOIUrl":"10.1007/s10921-026-01345-7","url":null,"abstract":"<div><p>Tire X-ray defect detection is vital for ensuring product safety and quality. However, manual inspection methods are prone to errors and inefficiencies. This study aims to explore a new tire defect detection model algorithm YOLO-GUAC with deep learning techniques to address these challenges. To address these challenges, we introduce and adapt the Global Attention Mechanism to enhance feature extraction, particularly for small defects against complex backgrounds. We further incorporate a Self-Calibrated Convolution module into the backbone and neck networks to optimize spatial and channel information flow while reducing redundancy. Additionally, an Auxiliary Detection Head is integrated to provide intermediate supervision, minimizing information loss during training. The experimental results show that the mAP@0.5 accuracy of YOLO-GUAC reaches 95% and the accuracy of mAP@0.5–0.95 is improved by 3.7%. Our model can effectively detect tire defects better than current mainstream methods, thus highlighting its potential in practical applications.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":"45 1","pages":""},"PeriodicalIF":2.4,"publicationDate":"2026-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147441558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}