Journal of Nondestructive Evaluation最新文献

This study presents a Decision Support System (DSS) designed for Non-Destructive Online Evaluation in welding. Based on the Multi-Scale Dense Cross Block Network (MDCBNet), it is able to detect, classify, and recommend remedial actions to prevent surface defects in welding. The performance of the network architecture is enhanced with synthetic defect samples generated through image augmentation techniques. By employing gradient attribution and t-SNE plot methods, we gained insights into the network’s predictions and comprehensively analyzed decision-making process. Comparative evaluations against pre-trained deep learning techniques revealed that our proposed model exhibits significant improvements, ranging from 2 to 10% across various performance metrics. Extensive comparisons with state-of-the-art methods underscored the effectiveness of our approach in detecting and classifying weld defects. Notably, our network, initially trained on Gas Tungsten Arc Welding images, demonstrated remarkable adaptability and versatility by effectively classifying images from Gas Metal Arc Welding processes. These findings emphasize the potential of the MDCBNet-based DSS to enhance welding practices, thereby contributing to producing high-quality weldments. The successful implementation of our DSS recommendations further supports its capacity to optimize the welding process and facilitate improved weld quality.

The ice formation over the aerofoil structure of the aircraft wing has been an obstruction as they abrupt the airflow, acting as drag. The investigation will intend to determine ice accumulation on carbon fiber-reinforced polymer (CFRP), approximated as ice build-up on aircraft wings. The observation is carried out over quasi-isotropic composite laminates using ultrasonic-guided waves with a central working frequency regime of 100 kHz. The three-dimensional (3D) finite element (FE) simulations are performed to observe the scattering effect to explore the reflection site in the far field. This effect was quite prominent for different thicknesses of Glaze ice (G-Ice) and was found to be strongly linked with the wave propagation and dispersion effect. The scattering results for the reflection of Lamb mode, when it interacted with the G-Ice interface, were quite noteworthy along the angular region rather than on the center line, indicating that the scattering was more prominent due to the presence of a 45° or (− 45)-degree fiber orientation in that laminate. A similar but complex scattering phenomenon was observed for different stacking sequences where the wave propagation angle and its amplitude at the receiver nodes are found to be closely bound with the exponential decay in group/phase velocity for the ice thicknesses studied. The FE approach is verified, and the results are validated analytically. Analytically, we have investigated a much-closed approximation with the detectability obtained from three-dimensional studies. Where the dispersion study performed has also contributed to verifying the present investigation in the long wavelength limits. This study can reveal the various optimized locations for placing the sensor for ice detection and quantification, which can be further helpful for practical guided wave inspection in ice detection and its removal.

Computer-aided weld defect recognition is transforming the field of Non-Destructive Testing by addressing the shortcomings of slow and error-prone manual inspections. This technology provides a reliable solution for detecting changes in pipeline conditions and structural damage. While conventional neural networks fall short in precise fault localization, deep learning-based object detection techniques step in to fill the gap. Addressing a real-industrial problem, particularly visually inspecting an X-ray welding database, without relying on a pre-existing benchmark presents a significant challenge in this field. Additionally, the poor quality of our welding data, which is riddled with small, sticky porosity in each image, poses several issues related to selecting the appropriate deep neural network object detector. This is yet another challenge that needs to be tackled. To direct these challenges, we introduced a novel approach based on the renowned Faster RCNN architecture to develop a model specifically designed for weld defect detection and recognition. This study dives deep into the inner workings of this newly adopted methodology. In our research, we have thoroughly parameterized, trained, tested, and validated this model. Our approach stands out through a comparative analysis with YOLO and DCNN models, highlighting the superiority of our Faster RCNN-based system. By evaluating its robustness and efficiency, our study reveals that the Faster RCNN model outperforms its counterparts in weld defect detection and localization for this specific small and sticky porosity defect type. This stands as a testament to effectively setting a new standard in this area.

Lead-based water pipelines pose a significant public health risk in the US. The challenge lies in locating these pipelines, as current identification technologies have limitations. This study discusses potential and challenges of identifying the water Service Line (SL) material through a stress wave propagation methodology. Since buried service lines are surrounded by soil and contain water, the stress wave propagation is non trivial. This work presents numerical simulations to investigate the applicability of the proposed method. First, authors consider wave propagation properties that could be used in a stress wave approach to identify buried lead based pipelines. For instance, dispersion curves are quite different for steel, copper, Lead, and PVC pipes filled with water. While the soil surrounding pipes causes a decrease in wave propagation energy due to the energy leakage into the soil medium, this phenomenon can enable the detection of leaked waves with sufficiently sensitive sensors installed near the soil surface. The received signals vary for different types of pipe materials, allowing to differentiate among service line materials. This study’s numerical simulations and lab experiments suggest that stress wave propagation could become a valuable tool for identifying buried lead-based water SL materials.

In the field of welding inspection, radiographic non-destructive evaluation (NDE) is a widely used technique for detecting defects in welds. However, this technique requires professionally qualified workers to manually judge radiographs to determine the presence, type, and location of defects. Recently, deep learning techniques have been developed to automate this process by using image segmentation. Despite its effectiveness, small-size targets in segmentation can have blurred boundaries, making it difficult to accurately annotate them at the pixel level. In this study, we propose an automated approach using the Point-REND Res-UNet model to improve the accuracy of detecting welding defects. Our method uses the improved Point-Rend algorithm to iteratively refine coarse segmentation results, allowing for more accurate defect detection. We evaluate our approach on a set of X-ray data and demonstrate that it achieves an improvement in model dice of 6.22%. Our proposed approach can potentially save labor time and costs while enhancing the accuracy and efficiency of welding defect detection.

Accurate quantification of the jawbone-implant interface plays a pivotal role in assessing the mechanical stability of dental implant structures. This study proposes a methodology integrating the electro-mechanical impedance (EMI)-based technique with a deep learning algorithm to autonomously monitor the bone-implant interface during the osseointegration process. We develop a 1D convolutional neural network (1D CNN) model, which automatically processes raw EMI data and extracts optimal features for predicting osseointegration ratios. To validate our approach, we conduct predictive 3D numerical modelling of the PZT-implant-bone system. This model simulates the implant’s EMI response under varying degrees of osseointegration. Next, we employ traditional statistical metrics to monitor osseointegration and discuss their limitations. Finally, we apply the proposed 1D CNN model to predict bone-implant osseointegration rate. We train and test the network using the simulated EMI data with added noise to account for real-world conditions. The results show that the trained model achieves a minimal testing error of just 2.4%. Even when 60% of testing cases are not trained, the model maintains a prediction accuracy exceeding 94%.

The surface spectral absorptivity of surface-whitened mortars due to the occurrence of efflorescence (i.e., mortars whose surface was covered with calcium carbonate) was measured, and the relationship between the spectral absorptivity and inspection capability of active thermography inspection was investigated. The spectral absorptivity of mortars increased significantly at a wavelength of approximately 3000 nm regardless of the presence/absence of the discoloration. Experiments for mortar specimens using optical lights with wavelengths in the visible, short wavelength, and medium/long wavelength ranges showed that the heating efficiency and defect detection capability of active thermography inspection were correlated with the surface spectral absorptivity, and were higher when long wavelength light was used as a heater. Defects in the surface-whitened mortar specimen were detected more efficiently when the specimen was heated using a CO₂ laser, whose wavelength is in the long wavelength range, than when using an optical light having a wavelength in the visible/short wavelength range.

Defect inspection of IC devices is getting more challenging with the increase of package density. Scanning acoustic microscopy (SAM) is widely used in electronic industry. The detection resolution is, however, limited by the penetration depth of ultrasound. It is necessary to find a way to improve the resolution and accuracy. A new strategy of multi-scale decomposition and fusion based on the wavelet transform was proposed to enhance the image resolution in SAM detection. The original SAM image was subjected to wavelet decomposition at different scales. Two recombined images A and B were decomposed into low frequency band (cAd1 and cAd2) and high frequency bands (cHd1, cVd1, cDd1, and cHd2, cVd2, cDd2), which were then merged respectively based on the local area energy. A high resolution SAM image was reconstructed by using the new coefficients. Back propagation network modified with genetic algorithm (GA-BP) was utilized to classify the solder joints. The proposed scheme achieved highest recognition accuracy (97.16%) compared with other methods. The new strategy provides an effective way to enhance the image quality and recognition accuracy in SAM detection of micro defect.

The challenges inherent in effective nondestructive evaluation of backside defects in steel, such as cracks, arise from the limited penetration of eddy currents (EC) due to the high permeability of steel. While the magnetic flux leakage (MFL) technique is able to detect deep defects, it lacks detailed geometry information. In this study, a hybrid approach is proposed, involving the simultaneous analysis of MFL and EC signals using a custom-designed magnetic probe. The probe is developed based on Finite Element Method simulations, followed by validation on 2 mm carbon steel plates containing artificial slits. The simulation results showed that the spatial and intensity responses of MFL and EC signals within the slits can be utilized for characterizing the slits. Furthermore, validation with fabricated backside slits confirms the correlation between slit depth, length and the intensity of the measured signals, particularly when an optimized excitation frequency is employed. The proposed method enables the prediction of slit depth and identification of slit shapes, thereby resulting in an enhancement of backside defect detection capabilities. Through this proposed hybrid technique, a connection is established between MFL and EC methods to enable a versatile tool for the precise assessment of cracks.

The field dependences of the signal of induction transducer U^~(H), proportional to the magnetic incremental permeability, have been measured on nickel and iron subjected to both elastic compression and tension. The inflections and additional maxima on the U^~(H) curves are observed for nickel under tension and for iron under compression. The appearance of the features on the U^~(H) curves is traceable to the induction of magnetic texture of the “easy-plane” type caused by elastic deformation of nickel and iron samples. These features appear only if the signs of magnetostriction and applied load are opposite. Only in this case, there is the possibility of estimation of applied mechanical stresses and residual stresses (after annealing). This can be useful for technical diagnostics of objects made of ferromagnetic materials with different signs of magnetostriction. The proposed technique is makes it possible to estimate the “easy-axis” magnetostriction constant.