Journal of Nondestructive Evaluation最新文献

Gearboxes are widely used in various industries such as aircrafts, automobiles, wind turbines, ship industries among others. Due its complex configuration, it is a challenging task to identify fault and failures patterns. Its internal components, such as bearings and gears, have different fault patterns, that can appear in one or in both components. The vibration signals were processed using the Empirical Mode Decomposition (EMD) and the Pearson Correlation Coefficient (PCC) to select the significant Intrinsic Mode Functions (IMFs) and then 18 features were extract from this IMFs. Four features ranking techniques [ReliefF, Chi-Square, Max Relevance Min Redundancy (mRMR) and Decision Tree] were used in a committee to select the best feature set, among the 10 with the highest rank, that appears at least in 3 of the 4 methods. The new feature set was used as an input to Support Vector Machine (SVM), Random Forest (RF) and Artificial Neural Networks (ANN) algorithms. The results showed that the use of the PCC value as a tool for selecting the significant IMFs, combined with the feature committee led to good results for this classification problem. In this case study, the ANN model outperformed the SVM and the RF algorithms, by using only 4 features to achieve 95.42% of accuracy and 6 features to achieve 100% of accuracy.

Air-coupled ultrasonic (ACU) testing has been used for several years to detect defects in plate-like structures. Especially, for automated testing procedures, ACU testing is advantageous in comparison to conventional testing. However, the evaluation of the measurement data is usually done in a manual manner, which is an obstruction to the application of ACU testing. The goal of this study is to automate and improve defect characterization and NDE 4.0 accordingly with deep learning. In conventional ACU testing the measurement data contains temporal (A-scans) and spatial (C-scans) information. Both data types are investigated in this study. For the A-scans, which represent time series data, neural network architectures tailored to such data types are applied. In addition, it is evaluated if further adaptions of the training procedure increase the performance. The C-scans are segmented by applying different U-net similar architectures and training strategies. In order to use spatial and temporal information, a further approach is taken. The prediction of the time series models is segmented with image models. The performance of all trained models and training strategies is compared with the F1-score and benchmarked against the conventional evaluation, which is thresholding of the C-scans. As specimens, artificial defects in acrylic and carbon fiber-reinforced polymer plates are investigated.

Modeling piezoelectric elements (piezos) using the finite element method with electro-mechanical coupling requires significant computational resources. The electro-mechanical interaction between piezo and structure in the interface will consume the most computational resources because it needs to be updated for each time step. If many piezos are involved, the wave-propagation-based analysis, including simulations of the wave motion, will be handicapped and might lead to the cancellation of the computation. Therefore, a simplified approach for modeling the piezoelectric response is presented, accounting for a ‘purely’ mechanical interaction between piezo and structure, where the electric potential is calculated analytically by multiplying the first two mechanical principal-strain components with the piezoelectric constants a posteriori. This way, the calculation of the equilibrium of the piezoelectric material is omitted which reduces the computational cost significantly without loss of accuracy in the piezoelectric response. An application case is demonstrated, where steel-ball impacts on an aluminum plate were successfully localized using a wave-propagation-based localization method (time reverse modeling), and the piezos were modeled with a simplified mechanical material behavior.

Automatic detection of welding flaws based on deep learning methods has aroused great interest in the non-destructive testing. However, few studies focus on the characteristics of welding flaws in the X-ray image. This study uses four deep learning models to train and test on a dataset containing 15,194 X-ray images. A hybrid prediction based on OR logic is proposed to avoid the miss detection as much as possible and reduce the miss detection rate to 0.61%, which is state of the art. Quantitative analysis of flaws’ characteristics, including the area, aspect ratio, mean, and variance, suggests the aspect ratios of miss detected flaws are smaller than 2, and the coefficient variances of miss detected flaws are smaller than 0.2. Tracking the critical pixels of X-ray images show that salt noises lead to false alarmed predictions. Error analysis indicates that when using the deep learning method for automatic welding flaws detection, the characteristics of flaws and the factors caused by inappropriate X-ray exposure techniques also should be noted.

The structural integrity of operated components can be assessed by non-destructive mechanical tests performed in-situ with portable instruments. Particularly promising in this context are small scale hardness tests supplemented by the mapping of the residual imprints left on metal surfaces. The data thus collected represent the input of inverse analysis procedures, which determine the material characteristics and their evolution over time. The reliability of these estimates depends on the accuracy of the geometry scans and on the robustness of the data filtering and interpretation methodologies. The objective of the present work is to evaluate the accuracy of the 3D reconstruction of the residual deformation produced on metals by hardness tests performed at a few hundred N load. The geometry data are acquired by portable optical microscopes with variable focal distance. The imperfections introduced by the imaging system, which may not be optimized for all ambient conditions when used in automatic mode, are analysed. Representative examples of the output produced by the scanning tool are examined, focusing attention on the experimental disturbances typical of onsite applications. Proper orthogonal decomposition and data reduction techniques are applied to the information returned by the instrumentation. The essential features of the collected datasets are extracted and the main noise is removed. The results of this investigation show that the accuracy achievable with the considered equipment and regularization procedures can support the development of reliable diagnostic analyses of metal components in existing structures and infrastructures.

Magnetic adaptive testing (MAT), thanks to its relative simplicity from the point of view of both hardware and software, appears to be very promising for non-destructive analysis of various ferromagnetic constructional materials used in many industrial applications. In order to make the inspection of tested objects faster, even more straightforward and the processing of data easier, we concentrated our effort both to improve the experimental procedure, based on specific way of the measurement of magnetization curves at piece-wise linear (triangular) exciting field with a constant field rate of change (i.e., slope) as well as to create a universal software tool for MAT data analysis. On contrary to the original implementation of MAT, the hysteresis loops are measured with decreasing maximum field values, starting at sample saturation region; therefore, time consuming sample demagnetization can be skipped completely. In addition, an advanced tool for the processing of experimentally obtained magnetization curves is presented. Using this application allows to find proper non-traditional magnetic parameters (e.g., the differential permeability) being the most sensitive to various types of industrial load (e.g., thermal, mechanical and/or neutron irradiation) and, at the same time, sufficiently correlated with other, traditional magnetic as well as non-magnetic parameters used routinely for the assessment of possible structural changes associated with applied, often long-term, load even on a microscopic scale, prior to any damage manifests on a macroscopic, visible level. The software capabilities were demonstrated on the data representing the material, whose response to acting thermal load is being difficult to analyze, since the dependence of observed parameters (differential permeability) upon defined artificial ageing was rather complicated. Nevertheless, the sensitivity of differential permeability to such a load was found being more than 8 times larger than in case of traditional hysteretic parameter, namely the remanent flux density while the correlation between them was high.

This paper proposes a nondestructive testing (NDT) method for the inspection of corrosion in rivets used in an aircraft. The NDT system uses an ultrasonic sensor coupling with a membrane that allows the ultrasonic wave propagates through to the inspecting rivet. The measured signal is then analyzed by a spiking neural network (SNN), a neural network that mimics the biological neurons for efficient detection of the corrosion in rivet. Compared to the conventional deep neural network, SNN is low energy consumption and can be implemented on a compact SNN accelerator chip, making them better run on a compact NDT system and general edge computing applications. We have tested the proposed SNN model on different sizes of corrosion in rivets (i.e., 30–70% of cross-section area) and at different depths from the surface (i.e., 1.0–2.0 mm). The proposed SNN model achieves about 95.4% accuracy with a small number of rivet samples (i.e., four rivet with corrosion) for training.

Impact damage constitutes a major threat to the performance and safety of fiber-reinforced composites. In this regard, transmission air-coupled ultrasound inspection technology has been identified as an ideal method for detection of common structural defects in modern multilayer composites. However, traditional machine learning algorithms and ultrasonic signal analysis methods are limited in terms of efficiency and accuracy. To remedy the situation, four one-dimensional deep learning models based on A-scan signals obtained from air-coupled ultrasound, which can automatically detect the impact damage in fiber-reinforced polymer composites, are constructed in this paper. Remarkably, all four models have attained high accuracy and recall on the testing sets, even though the training data and test data correspond to different materials and even structures. Among the four models, the long short-term memory recurrent neural network outperforms the other three models, which demonstrates its robustness and effectiveness.

X-ray computed laminography is a depth-resolving non-destructive testing technique well suited for the non-destructive examination of large and flat structures where traditional computed tomography is impractical. This technique provides 3D radiographic imaging and characterization with depth information of welding imperfections in welded components, ensuring component quality meets the standard criteria and safety purposes. Furthermore, determining the welding imperfection’s location in fabrication, in-service and maintenance is crucial for welding repair, resulting in the areas where the repair work needs to be started. This work highlights the characterization of welding imperfections by experimental digital radiography with digital detector array (RT-D with DDA) and coplanar translational laminography (CTL) techniques applied to welded carbon steel plates. A test specimen was tested, specially prepared with artificial planar and volumetric flaws like lack of fusion, clustered porosities and slag inclusions with varying dimensions and the approaches were analyzed. Additionally, a test phantom was fabricated with known geometry features that access the CTL system’s optimal detection accuracy to demonstrate a broad functionality and acceptance of the CTL system for depth information in the plate-like structures. The coplanar translational laminography technique provides advantages for characterizing welding imperfections and testing phantom features with high contrast and acceptable image quality. The result is confirmed by the phased array ultrasonic testing and RT-D with DDA. The exposure conditions, image sensitivity, and quality are analyzed according to ISO 17636-2 to ensure compliance with industry standards in digital radiography.

This study investigated the reliability of a corroded drag beam of a helicopter landing gear. The drag beam made up of high-strength steel failed due to corrosion-initiated cracking. The fracture probability was calculated using a simple capacity and demand model. The strength distribution of the drag beam was obtained through the estimated thickness using radiography, and a model for thickness measurement was developed using a linear attenuation coefficient for both a base metal and a corrosion product. The thickness reduction by corrosion was estimated by comparing the photon intensities of the corroded region and the region of known thickness without any corrosion. Due to uncertainties in the model parameters—thickness of the good area, photon intensity with or without the corrosion product, and corrosion rate—Monte Carlo simulation was conducted. The load distribution was obtained using the flight load data from strain gauges attached to the drag beams, which mainly carried a compressive load and a much smaller torsional load. The results show that the currently operated drag beams have a sufficient margin of safety. Considering uncertainties, the inspection of the drag beam using radiography proved that it was structurally reliable.