Journal of Nondestructive Evaluation最新文献

As a promising reference technique for non-destructive evaluation of both electrowelded and butt-fused polyethylene (PE) assemblies, Phased Array Ultrasonic Testing (PAUT) is still studied extensively in several laboratories worldwide and is supported by the technical standard ISO TS 16943. During the last 10 years, several joint projects have been completed aiming at evaluating the acuity of PAUT applied to both pipes and electrofused assemblies either exhumed from the field or prepared in laboratory. More recently, a focus has been made on fixing some acceptance criteria combining PAUT data and long term resistance of the laboratory joints. This paper presents the updated data obtained on electrofused assemblies—63 mm saddles and 110 mm sockets—containing different types of defects such as: insufficient heating time, pipe under-penetration in the socket, excessive localized scraping, pollutants and calibrated thin strips, in both mass and cross configuration, put at the interface pipe-saddle. PAUT scanning on the different specimens, both during the welding phase and after cooling, confirms the capability of the technique to visualize and size the Heat Affected Zone (HAZ), which can be revealed and compared afterwards on sample sections. Moreover, most of the defects are detected and sized, confirming the fairly good performance of PAUT, except for the smallest strips which are located in non accessible zones, due to the particular design of the saddle. Long term resistance of the welds is then evaluated by Hydrostatic Pressure Tests (HPT) followed by a decohesion test after rupture, according to the requirements of both the ISO 13956 and NF EN 1555 standards. Under such test conditions, every joints comply with the requirements of the standards (rupture time greater than 1000 h at 80 °C and 5 MPa), even those violating the critical proportions of non-welded zones.

The determination of void fraction in various two-phase flows holds great significance across a range of industries, including gas, oil, chemical, and petrochemical sectors. Scientists have proposed a wide array of methods for measuring void fractions. In comparison to other methods, capacitive-based sensors stand out as a good choice due to their affordability, nondestructively, robustness, and reliability. However, one of the factors that can affect the accuracy of these sensors is changes in the fluid composition. For instance, even a minor alteration in the fluid within the pipe can result in a significant void fraction measurement error. To address this issue, regular calibration is necessary, which can be a laborious task. In this paper, an Artificial Neural Network (ANN) is employed in order to make sensor measurements independent of fluid changes, which allows for more reliable and precise measurements without the need for frequent calibration. Our focus is on studying stratified two-phase flow. In this research, four different combinations of electrodes of a four-concave sensor are utilized as the input of an ANN. As a result, the ANN’s output accurately quantifies the void fraction. COMSOL Multiphysics software is utilized to simulate the behavior and measure the capacitance value of different combinations of this sensor. Additionally, a Multilayer Perceptron (MLP) neural network in MATLAB is designed and implemented, which can forecast the gas percentage within a two-phase fluid containing different liquids, achieving a remarkable mean absolute error of only 0.0031.

Long-term operation of tunnels is influenced by factors such as material degradation and variations in service loads, resulting in varying degrees of lining cracks and water leakage. Addressing the challenge of low target contrast and significant background interference in thermal imaging detection of tunnel lining cracks, this study established an experimental system capable of simulating tunnel operational humidity and temperature environments. The lining cracks were localized and the temperature field distribution on the cracked lining surface was characterized by computer vision techniques. Additionally, numerical simulations using thermal-flow coupling were employed to validate and extend the results of indoor experiments. Based on the distribution patterns of the temperature field, a gain transformation function for thermal images was established. The findings indicate that the processed images were 2–6 times better than the conventional algorithm in terms of SCRG (Signal Clutter Ratio Gain) and BSF (Background Suppression Factor) metrics. This research provides valuable insights and references for the practical application of thermal imaging detection in tunnel water leakage scenarios.

Accelerated ageing of cross-linked polyethylene (XLPE) insulated aerial bundled cables (ABC) installed at coastal regions for electric power distribution is a topic of major concern. Sudden failures caused by rapid insulation deterioration initiate unexpected power shutdowns. Accurate prognosis of incipient failures in cables will enable maintenance agencies to plan repair or replacement activities well on time thus improving the reliability and availability of the transmission grid. In this research work, historical non-destructive evaluation (NDE) data was generated using infrared thermography of ABCs to ascertain cable degradation parameter. Historical NDE data embodies the progressive degradation experienced by the ABCs installed at harsh marine environment. The degradation growth in cable insulation, when subjected to rough environmental conditions, is non-linear coupled with non-Gaussian / multimodal noise distributions. Therefore, a renowned nonlinear Bayesian estimator namely Particle Filter (PF) is applied on the historical database to determine degradation growth evolution over time. The proposed framework is further complimented with f-step prediction scheme in future time to quantify the prediction accuracy of cable degradation growth in future where measurement data is not available. The so predicted results are then compared to the actual degradation in future. The prediction accuracy demonstrates the efficacy of proposed technique for prognosis of ABCs installed at different locations. Accurate state prediction in different life phases of ABCs further displays robustness of proposed technique in estimating actual degradation growth during the different life stages.

Mining wire rope (MWR) is an important part of mine hoisting equipment and plays a key role in mining operations. Damage to these ropes can significantly reduce production efficiency and pose serious safety risks to workers. Therefore, quantitatively identifying damage in MWR is of great importance. Traditional methods for damage signal identification rely on manual feature extraction (MFE), which depends heavily on experience and lacks stability and flexibility. This paper proposes an end-to-end (E2E) quantitative identification model for MWR damage based on time series classification (TSC) and deep learning (DL). Unlike traditional methods, the E2E model learns features directly from the one-dimensional raw signals of MWR damage and does not require MFE. In order to test its validity and versatility, experiments were conducted on three different datasets. The results show that the E2E method performs well in quantitatively identifying MWR damage compared to other methods and this method meets the requirements of the mining industry in terms of precision and efficiency to ensure safe and reliable operation of mining work.

The rich and complex texture information of industrial products, and the fact that the number of normal products often far exceeds the number of defective products in industrial scenarios, poses a great challenge to product quality inspection. In order to solve this challenge, a weakly supervised defect detection model based on nested U-Net was proposed: the nested U-Net was used as the main body of the model, and an attention module was introduced, which could obtain the relationship between local features and between feature channels. Furthermore, A weakly supervised training strategy was employed: the defect mask from the Berlin noise, external data and normal samples are used to synthesize the defects, and a few real defect samples are randomly inserted into the synthetic defect samples to train the detection model. Experimental validation was carried out on the public datasets MVTec AD, DAGM, MT and custom CT (computed tomography) composite material dataset, and the evaluation indicators included image-level AUC (area under the receiver’s operating characteristic curve), pixel-level AUC and AP (average accuracy). Experimental results show that the proposed method achieves excellent performance of 99.9%/98.7%/84.1%, 99.1%/95.3%/76.1%, 100%/98.1%/86.7% and 73.6%/69.1%/36.0% on three types of metrics on four datasets, respectively, which is better than the current advanced model.

Near-field microwave imaging shows considerable promise for non-destructive evaluation of internal defects in high silica/phenolic composites, which are commonly used as thermal protection systems (TPS) for rocket/missile solid motor nozzles and space re-entry vehicles. However, effectively identifying defect features using post-processing algorithms remains challenging. To address this challenge, this paper proposes a microwave defect characterization algorithm based on Convolutional Neural Networks (CNN). A defect dataset derived from reflection microwave signals, was manually compiled by detecting samples with critical defects. The CNN framework was utilized for precise classification of microwave signals, employing a classification encoding strategy to extract two-dimensional defect information and achieve automatic localization and imaging of defects. Multiple deep learning models were compared in both simulations and experiments, revealing that the proposed CNN exhibited significant advantages in feature extraction, enabling highly effective identification of internal defects even with a limited dataset. Compared with traditional algorithms, the detection accuracy of the proposed 1D-SENet has been improved by 53.35% and 50.66%, respectively, and can achieve detection of defects with a minimum size of Φ6mm. These validate the effectiveness of algorithm in intelligent and automated microwave characterization of delamination defects within composite materials.

Improving detection accuracy without reducing detection depth has always been a challenge in ultrasound imaging. For multi-layer anisotropic carbon fiber reinforced polymers (CFRPs), the propagation path of ultrasonic waves becomes complex with severe attenuation. To improve the imaging accuracy of defects in CFRPs, a beam multiply and sum method suitable for full matrix data, FM-BMAS, has been proposed. This method introduces the spatial coherence of ultrasonic array signals through cross-multiplication operation. A new harmonic component is generated and high-quality imaging of side-drilled holes (SDHs) is achieved without affecting the detection depth. The FM-BMAS method can detect two SDHs with a diameter of 1 mm at a depth of 8 mm at 2.5 MHz detection frequency. In contrast, these defects cannot be visualized by the classical total focusing method due to ultrasonic attenuation even with 5 MHz detection frequency. Furthermore, FM-BMAS can achieve higher image resolution and superior noise suppression capabilities in CFRPs compared to other methods.

With increasing attention to safety, accurate detection of defects in diverse materials is crucial. Non-Destructive Evaluation (NDE) techniques play a key role in this effort. However, many approaches overlook preliminary signal analysis, which is vital for understanding fundamental sensor signal characteristics and improving advanced techniques. This study aims to investigate the fundamental signal properties from coplanar capacitive sensors (CCS) with square electrodes to gain insights into the relationship between signal properties and actual linear defect size in both conducting and non-conducting materials. The first-order derivatives (1st O.DE) of raw signals obtained from CCS through both simulations and experiments were further analysed, focusing on a circular surface defect. Nine sensor configurations were tested to examine the spreading effect. Experiment and simulation results were in good agreement, showing that the Full Width at Zero (FWZ) of the raw signals of both materials is greater than the actual defect diameter (spreading effect) whereas the signals processed using the 1st O.DE exhibited peak-trough widths greater than the actual defect diameter in non-conducting materials (spreading effect) and less than the actual diameter in conducting materials (shrinking effect). These results underscore that the spreading and shrinking effects are intrinsic characteristics of the CCS, attributed to the behavior of the CCS’s electric field and sensitivity distribution field (SDF) when interacting with different materials. By incorporating these insights into novel and advanced methods—such as imaging algorithms, machine learning approaches, and data fusion techniques—future developments can be effectively guided to enhance the accuracy, reliability, and advancement of defect detection, imaging, and sizing in coplanar capacitive sensing for NDE.

The range of application of eddy current testing (ECT) has been recently extended to non-electrically conductive materials, in which case it is called electromagnetic induction testing (EIT) given that EIT detects changes in the electromagnetic field of the displacement current. Although EIT has been reported for non-destructive characterization of non-electrically conductive materials, its detection principle is still unclear. We used finite element analysis (FEA) and experiments to evaluate the effect of cracks on the displacement current field in non-electrically conductive materials for crack detection using EIT. FEA was performed on electrically and non-electrically conductive materials with slits simulating cracks. The FEA results showed that crack detection differed between materials, as eddy currents bypassed the crack, while displacement currents passed through and formed a current path. Furthermore, the electric field intensity and displacement current induced in the non-electrically conductive materials varied significantly in the cracked area compared with the uncracked area. Experiments were conducted to detect cracks in carbon fiber reinforced thermoplastics (CFRTPs) and glass fiber reinforced plastics (GFRPs), which have isotropic electrical properties in the in-plane direction. In the CFRTP, the electromagnetic field varied significantly even at locations far from the crack, whereas it changed only slightly near the crack in the GFRP. This result demonstrates that for non-electrically conductive materials, EIT can identify cracks by detecting localized changes in the displacement current flowing through the cracks. Our findings can help clarify the principle of crack detection in non-electrically conductive materials, thereby extending the application of EIT to these materials.