Showing papers in "Ndt & E International in 2019"

Temporal and spatial deep learning network for infrared thermal defect detection

Abstract: Most common types of defects for composite are debond and delamination. It is difficult to detect the inner defects on a complex shaped specimen by using conventional optical thermography nondestructive testing (NDT) methods. In this paper, a hybrid of spatial and temporal deep learning architecture for automatic thermography defects detection is proposed. The integration of cross network learning strategy has the capability to significantly minimize the uneven illumination and enhance the detection rate. The probability of detection (POD) has been derived to measure the detection results and this is coupled with comparison studies to verify the efficacy of the proposed method. The results show that visual geometry group-Unet (VGG-Unet) cross learning structure can significantly improve the contrast between the defective and non-defective regions. In addition, investigation of different feature extraction methods in which embedded in deep learning is conducted to optimize the learning structure. To investigate the efficacy and robustness of the proposed method, experimental studies have been carried out for inner debond defects on both regular and irregular shaped carbon fiber reinforced polymer (CFRP) specimens.

Approaches for improvement of the X-ray image defect detection of automobile casting aluminum parts based on deep learning

Abstract: Nondestructive testing (NDT) for casting aluminum parts is an essential quality management procedure. In order to avoid the effects of human fatigue and improve detection accuracy, intelligent visual inspection systems are adopted on production lines. Conventional methods of defect detection can require heavy image pre-processing and feature extraction. This paper proposes a defect detection system based on X-ray oriented deep learning, which focuses on approaches that improve the detection accuracy at both the algorithm and data augmentation levels. Feature Pyramid Network (FPN) was primarily adopted for algorithm modification, which proved to be better suited for detecting small defects than Faster R-CNN, with a 40.9% improvement of the mean of Average Precision (mAP) value. In the final regression and classification stage, RoIAlign indicated apparent accuracy improvement in bounding boxes location compared with RoI pooling, which could increase accuracy by 23.6% under Faster R-CNN. Furthermore, different data augmentation methods compensated for the lack of datasets in X-ray image defect detection. Experiments found that an optimal mAP value existed, instead of it continuously increasing with the number of datasets rising for each data augmentation method. Research indicated that the three proposed improvement approaches performed better than baseline Faster R-CNN in X-ray image defect detection of automobile aluminum casting parts.

New features for delamination depth evaluation in carbon fiber reinforced plastic materials using eddy current pulse-compression thermography

Abstract: The growing application of composite materials in aerospace leads to the urgent need of non-destructive testing and evaluation (NDT&E) techniques capable of detecting defects such as impact damage and delamination possibly existing in those materials. Eddy current pulsed thermography is an emerging non-destructive testing (NDT) technique capable of detecting such defects. However, characterization of delamination within composite materials is difficult to be achieved by a single pulse excitation, especially in carbon fiber reinforced plastic materials as the extraction of thermal diffusion in such multi-layered structures is challenging. To cope with this problem of signal-to-noise ratio, this paper proposes the eddy current pulse-compression thermography (ECPuCT), combining the Barker code modulated eddy current excitation and pulse-compression technique to enhance the capability of characterizing delamination on carbon fiber reinforced plastic materials. Additionally, a thermal pattern enhanced method based on kernel principal component analysis technique is used to locate the delaminated areas. Two features, including a newly proposed crossing point of impulse responses related to defective and non-defective areas and skewness of impulse responses are investigated for delamination depth evaluation. Results show that delamination can be detected within depths ranging from 0.46 mm to 2.30 mm and both the proposed features have a monotonic relationship with delamination depths.

Visual inspection and characterization of external corrosion in pipelines using deep neural network

Abstract: In this paper, we proposed a computer vision based approach to detect corrosion in water, oil and gas pipelines. For this, we created a dataset containing more than 140,000 optical images of pipelines with different levels of corrosion. A custom designed convolutional neural network (CNN) was applied to classify the images of pipelines based on their corrosion level. This in-house fabricated CNN has very few parameters to be learned in comparison with the existing CNN classifiers. However, it produced significantly higher classification accuracy (98.8%) with an ability to discriminate between images of corroded pipelines and images without corrosion but having patterns similar to corroded pipelines. The proposed network surpassed most of the state-of-the-art classifiers in its performance. In addition, we proposed a localisation algorithm based on a recursive region-based method, to selectively identify the corroded regions in a given image with higher precision. The proposed deep learning approach effectively wards off the need for manual inspection and other non-vision based non-destructive evaluation techniques for pipeline corrosion which are cost ineffective and interrupts the functioning of pipelines.

A novel machine learning model for eddy current testing with uncertainty

Abstract: A novel deep learning based eddy current inversion algorithm is proposed and investigated in this paper. Eddy current testing (ECT) for defects detection problem is adopted to demonstrated the proposed algorithms. The proposed model based on a Convolutional Neural Network (CNN) is developed to improve the defect detection performance with uncertainty information. The novelty of our work consists in combining characteristics of ECT data with general deep learning model to improve performance of deep learning in ECT field including a region of interest (ROI) method based on robust principle component analysis, a CNN classification model with weighted loss function and measurement of uncertainties. Experimental dataset obtained from eddy current inspection of heat exchanger tubes is utilized to validate the detection performance improvement. As a result, both the classification accuracy and the percentage of defects correctly identified have been increased to almost 100%.

Nonlinear guided wave technique for localized damage detection in plates with surface-bonded sensors to receive Lamb waves generated by shear-horizontal wave mixing

Abstract: Mutual wave interactions provide a very promising technique for nondestructive testing and evaluation due to their exceptional sensitivity to micro-scale damage growth in metallic materials. This article describes detection of localized fatigue damage in aluminum plates using polyvinylidene difluorine (PVDF) sensors to simultaneously receive shear-horizontal waves and the secondary Lamb waves that they generate. Finite element simulations explore several aspects of wave interaction and mode identification. Then laboratory experiments confirm secondary wave generation at the sum frequency in an aluminum plate using the PVDF sensor by conducting frequency-wave number domain and supporting analyses. The PVDF sensor enables computation of amplitude ratios even though the primary and secondary waves have different polarities. Finally, a simple guided wave technique based on electrically scanning the wave mixing zone around a plate is shown to detect early localized fatigue damage in an aluminum plate.

Automated defect classification of SS304 TIG welding process using visible spectrum camera and machine learning

Abstract: Tungsten Inert Gas welding is dependent on human supervision, it has an emphasis on visual assessment, and it is performed in a controlled environment, making it suitable for automation. This study designs a system for assessing the tungsten inert gas welding quality with the potential of application in real-time. The system uses images in the visible spectrum paired with the state-of-the-art approach for image classification. The welding images represent the weld pool in visible spectra balanced using high dynamic range technology to offset the powerful arc light. The study trains models on a new tungsten inert gas welding dataset, leveraging the state-of-the-art machine learning research, establishing a correlation between the aspect of the weld pool and surrounding area and the weld quality, similar to an operator's assessment.

Reliability assessment of pulsed thermography and ultrasonic testing for impact damage of CFRP panels

Abstract: In order to quantitatively compare the reliability of pulsed thermography and ultrasonic testing techniques, a set of thirty-five Carbon Fiber Reinforced Plastic (CFRP) composite panels with impact damages are inspected by pulsed thermography and ultrasonic C-scan. The comparative experimental results and Probability of Detection (PoD) analysis results are presented. The quantitative comparison shows that pulsed thermography testing has smaller defect size at 90% PoD with 95% confidence level, i.e. a90/95 values than ultrasonic testing for the parameters and setup used in the inspections of these thirty-five CFRP composite panels.

Automated defect classification in infrared thermography based on a neural network

Abstract: This paper reports on the use of a neural network in infrared thermography to classify defects, such as air, oil, and water, which can degrade material performance. A finite element method and experiment were adopted to simulate air, water, and oil ingress. Raw data, and thermographic signal reconstruction coefficients were used to train, and test the two multilayer, feed-forward NN models. Quantitative comparisons showed that the model using coefficients as features performed better than the one using raw data. It was more precise and had better test repeatability. This indicates the model is more generalizable.

Thickness measurement of non-magnetic steel plates using a novel planar triple-coil sensor

Abstract: In the eddy current testing, the impedance tested by the sensor is influenced by the lift-off between the sensor and the sample plate. Consequently, the thickness inferred from the measured impedance is also affected by the lift-off. In this paper, a novel planar triple-coil air-core sensor is designed, in which three coils (two excitation coils and one receiving coil) with different radii are co-axially arranged in the same plane. Based on this sensor setup, an algorithm for the thickness measurement of the non-magnetic metallic plate is proposed. By using the transmitter-receiver combinations and the proposed algorithm, the measured thickness is almost immune to the lift-off variation. Moreover, previously the measurement of thickness requires a complex frequency feature – peak frequency for the real part of the impedance, which is obtained from the multi-frequency impedance or inductance spectra (frequency-sweeping mode), while with the method proposed in this paper, only single frequency measurement is needed, hence relaxing the instrument and measurement requirements. This method has a potential of on-line real-time measurement of the thickness for non-magnetic steel plates.

Comparative analysis of eddy current pulsed thermography and long pulse thermography for damage detection in metals and composites

Abstract: Eddy current pulsed thermography (ECPT) and long pulse thermography (LPT) are two emerging non-destructive evaluation techniques that have been recently used for sub-surface defect detection in both metallic and composite components. This paper provides a comparative analysis of these two methods by focussing on the detection of flat bottom holes (FBHs) in both aluminium and carbon fibre reinforced plastic samples. Principal component analysis was used for the post-processing of thermal data in order to enhance the detection of FBHs with various diameters and depths. Results showed that LPT had a better performance for detecting FBHs in low-thermally conductive materials such as carbon fibre-reinforced thermoplastic composites, whilst ECPT revealed a superior performance for detecting FBHs in materials with higher thermal and electrical conductivity such as aluminium.

An investigation of the stepped thermography technique for defects evaluation in GFRP materials

Abstract: The ability of thermography to detect defects in composite materials has been demonstrated and showed in various works and in many applications. In this regard, various NDT techniques are currently used for defect detection in composites such as Lock-in Thermography (LT), Pulsed (PT) Stepped Thermography (ST/SH), all of which have their own peculiarities and capabilities. A critical aspect concerns the overall lengthy time required for testing and analysis of thermographic data above all, for large structure where it is necessary a scanning approach. In this work, two algorithms based on the stepped thermography approach were investigated in quantitative way with the aim to optimizing the testing parameters and data analysis in terms of testing time and signal to noise ratio. In particular, several tests were carried out on a sample specimen with simulated defects and the well-established lock-in thermography technique has been used as comparison.

Automated defect detection for Fluorescent Penetrant Inspection using Random Forest

Abstract: Fluorescent Penetrant Inspection (FPI) is the most widely used NDT method in the aerospace industry. Inspection of FPI is currently done visually and difficulties arise distinguishing between penetrant associated with defects and that due to insufficient wash-off or geometrical indications. This, in addition to the nature of the inspection process, means inspection is largely influenced by human factors. The ability to perform automated inspection would provide increased consistency, reliability and productivity. The Random Forest algorithm was used to detect defects in a number of flat titanium plates which had been processed with FPI and photographed to produce digital images. This method has demonstrated the ability to correctly distinguish between defects and other non-relevant indications with accuracy comparable to a human inspector with a very small number of training examples. These results show the potential for the Random Forest algorithm to be used to detect defects in aerospace components, allowing the entire FPI line to become autonomous.

Augmented Ultrasonic Data for Machine Learning

Abstract: Flaw detection in non-destructive testing, especially for complex signals like ultrasonic data, has thus far relied heavily on the expertise and judgement of trained human inspectors. While automated systems have been used for a long time, these have mostly been limited to using simple decision automation, such as signal amplitude threshold. The recent advances in various machine learning algorithms have solved many similarly difficult classification problems, that have previously been considered intractable. For non-destructive testing, encouraging results have already been reported in the open literature, but the use of machine learning is still very limited in NDT applications in the field. Key issue hindering their use, is the limited availability of representative flawed data-sets to be used for training. In the present paper, we develop modern, deep convolutional network to detect flaws from phased-array ultrasonic data. We make extensive use of data augmentation to enhance the initially limited raw data and to aid learning. The data augmentation utilizes virtual flaws—a technique, that has successfully been used in training human inspectors and is soon to be used in nuclear inspection qualification. The results from the machine learning classifier are compared to human performance. We show, that using sophisticated data augmentation, modern deep learning networks can be trained to achieve human-level performance.

High power phased EMAT arrays for nondestructive testing of as-cast steel

Abstract: A new high-power electromagnetic acoustic transducer (EMAT) solid state pulser system has been developed that is capable of driving up to 4 EMAT coils with programmable phase delays, allowing for focusing and steering of the acoustic field. Each channel is capable of supplying an excitation current of up to 1.75 kA for a pulse with a rise time of 1 μs. Finite element and experimental data are presented which demonstrate a signal enhancement by a factor of 3.5 (compared to a single EMAT coil) when using the system to transmit a longitudinal ultrasound pulse through a 22.5 cm thick as-cast steel slab sample. Further signal enhancement is demonstrated through the use of an array of detection EMATs, and a demonstration of artificial internal defect detection is presented on a thick steel sample. The design of this system is such that it has the potential to be employed at elevated temperatures for diagnostic measurements of steel during the continuous casting process.

Ultrasonic imaging of irregularly shaped notches based on elastic reverse time migration

Abstract: Ultrasonic techniques have been proved to be useful for detection and characterization of flaws in solid structures. However, it remains challenging to characterize flaws that do not have regular shapes. In this paper, an ultrasonic imaging technique based on elastic reverse time migration (ERTM) is developed for imaging notches with irregular shapes. In this method, the image is generated by cross-correlating the forward propagated wavefield from the source with the time-reversed backward propagated wavefield from the scatterer. Comparing to traditional ultrasonic imaging methods based on the travel time of ultrasonic signals, this method considers full waveforms which contain the information of mode conversions and multiple scattering, and therefore enables the possibility to image flaws with complex shapes. In this paper, the ERTM algorithm is applied to image a branched surface-breaking notch and an embedded stepped notch, showing excellent reconstruction results in both simulations and experiments.

A spectroscopic method based on support vector machine and artificial neural network for fiber laser welding defects detection and classification

Abstract: Diverse welding processes have been utilized in manufacturing industry for years. But up to date, welding quality still cannot be guaranteed, due to the lack of an efficient and on-line welding defects monitoring method, and this leads to increased manufacturing costs. In this paper, a method based on feature extraction and machine learning algorithm for on-line quality monitoring and defects classification was presented. Plasma radiation was captured by an optical fiber probe, and delivered by an optical fiber to the spectrometer. The captured spectral signal was processed by selecting sensitive emission lines and extracting features of spectral data's evolution, which realized spectral data compression with low computational cost. After selecting the proper training data set, the designed ANN and SVM allows automatic detection and classification of welding defects. The validity of proposed method was successfully approved by test data set in welding experiments. Welding experiments on galvanized steel sheets showed the corresponding relationship between the output of classifiers and welding defects. Finally, the two classifiers were compared. Experiments indicated the performance of ANN is slightly better than that of SVM, while both of them have its own advantages.

Fast reconstruction of 3-D defect profile from MFL signals using key physics-based parameters and SVM

Abstract: Fast reconstruction of three-dimensional (3-D) defect profile from three-axis magnetic flux leakage (MFL) signals is important to the pipeline inline inspection (ILI) in the oil and gas industry. Traditional methods require the processing of a large amount of raw input and output data, which poses significant challenges in balancing the inspection efficiency, i.e. sensing and data processing speed, and the ILI accuracy and robustness. Here, a novel fast reconstruction framework combining key physics-based parameters and data-driven machine learning algorithms is proposed. Geometric parameters based rational Bezier curve (RBC) model is proposed to generate the 3-D defect profile, while local and global feature parameters are determined using a nonlinear least square (NLS) approach from the three-axis MFL signals. These physics-based geometric and feature parameters are then correlated through a least-square support vector machine (LS-SVM). Meanwhile, a pipeline inspection gauge (PIG) is developed to measure the three-axis MFL signals for evaluating the reconstruction performance through field testing. Both simulation and experimental results demonstrate that the proposed method's accuracy, robustness and computation speed have been improved significantly comparing with other existing methods.

Algorithm development for real-time thin asphalt concrete overlay compaction monitoring using ground-penetrating radar

Abstract: Ground-penetrating radar (GPR) is an effective, feasible, and non-destructive tool for real-time estimation of the density of asphalt concrete (AC) during compaction. When the AC pavement is thin, however, it is difficult to estimate the density due to the limited GPR signal resolution. In addition, the surface moisture, applied during compaction, has an effect on the GPR signal and hence impacts density prediction accuracy. In this study, a nonlinear optimization approach, based on gradient descent, was used to recover the AC pavement surface reflection. A “modified reference scan” approach was developed to remove the effect of surface moisture on GPR signals during density monitoring. The “modified reference scan” approach was first validated in a laboratory experiment setting, then validated in a field study near Chicago, IL, where GPR antennas were installed on the roller and data were collected during compaction. The results show that GPR is a prospective tool for real-time AC compaction monitoring.

Interaction of SH guided waves with wall thinning

Abstract: This paper investigates through experiment and finite element modelling, the interaction and mode conversion phenomenon of SH0 and SH1 guided wave modes on a metal plate with machined wall thinning. Quantitative analysis was performed by calculating the reflection and transmission coefficients at the leading and trailing linearly tapered edges, for incident SH0 and SH1 modes. Several geometries were evaluated by varying the taper length and depth. Experiments were performed with periodic permanent magnet array EMATs as transmitters and receivers, generating a single SH mode, whilst both SH0 and SH1 are received. Experimental and numerical data show good agreement, revealing that the interaction of SH guided waves with such defects is complex when mode conversion arises. The values of the reflection and transmission coefficients are non-monotonic along the thinning depth and edge angle ranges. The quantitative results provide insight into the capabilities and limitations of guided SH wave measurements for simple corrosion type defects, indicating that with current capabilities, inspection of real defects will be limited to screening type measurements rather than detailed quantification of the defect region.

Non-destructive hardness prediction for 18CrNiMo7-6 steel based on feature selection and fusion of Magnetic Barkhausen Noise

Abstract: Different features of Magnetic Barkhausen Noise (MBN) represent distinctive characteristics of material stress states, microstructures and domain wall (DW) behaviors. However, a group of features could be selected and fused for the evaluation of specific characters such as hardness and stress. This paper analyzes multiple features of MBN and their reflections on different material and health status influences and proposes a new feature selection and fusion method for hardness evaluation through multiple feature relationships and different material influence. Principal component analysis (PCA) algorithm combined with feature correlation analysis method is utilized for feature selection by reducing redundant features and multiple parameter influences. Finally, we apply multivariate linear regression (MLR) with selected features to build a statistical linear model of MBN feature fusion for various material hardness prediction. The effectiveness of this feature selection is validated by various MLR models with different MBN features.

Automated visualization of concrete bridge deck condition from GPR data

Abstract: Ground-penetrating radar (GPR) is one of the most commonly used technologies for condition assessment of concrete bridge decks. However, there have been no fully automated algorithms to visualize the data collected with this technique. In such context, the current paper presents a method for a full automation of GPR data visualization and analysis. Based on the background removal, depth correction, synthetic aperture focusing technique (SAFT), and interpolation algorithms, this automated method produces a plan view map of amplitude of GPR signals. In the obtained map, two types of information are observed at the same time. First, as the strongest reflectors of electromagnetic energy, rebars will appear as the most visible. Second, the areas of corrosive environment and, thus, likely corrosion, will be detected as having low amplitude rebar reflections. As a proof of concept, the proposed method was implemented for two bare concrete bridge decks and two concrete bridge decks with asphalt overlays. In all cases, the results obtained were excellent where the maps pinpointed the areas affected by corrosion. These areas were confirmed by other methods of evaluation, such as electrical resistivity (ER), half-cell potential (HCP), chloride analysis of core samples, or visual inspection. With the demonstrated performance, the proposed method is expected to be an excellent alternative to the available methods of GPR data evaluation and visualization. In the future, it should be improved to provide an indication of corrosion severity/probability at each deck location.

In-plane local defect resonances for efficient vibrothermography of impacted carbon fiber-reinforced polymers (CFRP)

Abstract: It is well known that the efficiency of the vibrothermographic non-destructive testing (NDT) technique can be enhanced by taking advantage of local defect resonance (LDR) frequencies. Recently, the classical out-of-plane local defect resonance was extended towards in-plane LDR for enhanced efficiency of vibrometric NDT. This paper further couples the concept of this in-plane LDR to vibrothermography, on the basis of the promising potential of in-plane LDRs to enhance the rubbing (tangential) interaction and viscoelastic damping of defects. Carbon fiber-reinforced polymers (CFRPs) with barely visible impact damage (BVID) are inspected and the significant contribution of in-plane LDRs in vibrational heating is demonstrated. Moreover, it is shown that the defect thermal contrast induced by in-plane LDRs is so high that it allows for easy detection of BVID by live monitoring of infrared thermal images during a single broadband sweep excitation. Thermal and vibrational spectra of the inspected surface are studied and the dominant contribution of in-plane LDR in vibration-induced heating is demonstrated.

Quantitative evaluation of crack depths and angles for pulsed eddy current non-destructive testing

Abstract: Cracks with inclination angles may potentially cause damage to a larger region in the tested structures. Their characterization, in terms of depth and angle, is therefore paramount for ensuring the integrity of the specimen under test. This study extracts features from Pulsed eddy current (PEC) signals obtained in a linear scan, perpendicular to the simulated surface cracks. The novel features extracted, termed skewness, LLS and LSmax, are capable of defining crack depth and inclination angles simultaneously. Multiple linear regression (MLR) was built to perform depth prediction, and the pre-determined depths were used in the hierarchical linear model (HLM) for angle prediction. The results were then compared with depth and angle prediction using artificial neural network (ANN). Better reliability of the ANN model with recorded RMSE of 0.198 mm and 2.903° in depth and angle prediction are highlighted. ANN is favourable in handling simultaneous prediction of crack depth and inclination angles, when using interdependent features. Meanwhile, HLM is still approved as a technique to provide a preliminary understanding of the crack parameters.

Performance assessment method for crack repair in concrete using PZT-based electromechanical impedance technique

Abstract: Concrete cracks are considered as an important indicator for potential damage and durability and are generally monitored in structural maintenance. A typical repair method in practice for concrete cracks involving with durability issues is to use alternative materials to fill the cracks. As the recovery quality of the repair is difficult to determine, an effective assessment approach is necessary. The electromechanical impedance (EMI) technique is a widely favored method; however, the determination of the frequency range and corresponding damage indicator is a governing factor for practical applications. Furthermore, its application to the evaluation of the crack repair performance has not been reported in the literature to date. This study presents an applicability of the PZT-based EMI technique for the quality assessment of crack repair in concrete. To determine an optimal frequency range that strongly reflects the structural condition, the trends of four damage indicators are investigated with multiple damage severities in each sub-frequency range. Subsequently, the performance of the crack repair material is evaluated using the predefined optimal frequency range with the damage indicators. The proposed approach is shown to effectively assess the recovery quality of the crack repair material over time.

Studies on influences of initial magnetization state on metal magnetic memory signal

Abstract: Metal magnetic memory testing is a novel testing method which can early test stress concentration level of ferromagnetic components. In theory, the influences of the initial magnetization on the magnetic memory signal were studied. In the experimental aspect, take Q235 steel specimen for example, the relationship between initial magnetization state and metal magnetic memory signal was researched by static tensile test under three different initial magnetization conditions. Both the experimental studies and theoretical analysis come to similar conclusion that the variation trend of magnetization with the stress is that the initial magnetization status tends to shift towards the non-hysteresis magnetization state. With the increasing of stress, the magnetizing state gradually shifts towards the anhysteretic magnetization state, and the variation rate is directly proportional to the difference between the initial magnetization and the non-hysteresis magnetization. Our study indicates that the initial magnetization state has an important influence on the magnetic memory signal, and the research results can provide reference for quantitative test of metal magnetic memory testing technology.

Variational mode decomposition linked wavelet method for EMAT denoise with large lift-off effect

Abstract: Electromagnetic acoustic transducer (EMAT) is an emerging non-destructive testing technique which is widely used in the industry. EMAT has advantages of a non-coupling agent with the capability of big lift-off detection. However, EMAT has an issue of low efficiency in conversion and it is susceptible to noise. This paper proposes a modified variational mode decomposition (VMD) linked wavelet method for EMAT denoising. It enables to suppress both high-frequency narrowband noise and normal noise in EMAT signals with a large lift-off detection condition. In particular, a new ultrasonic echo signal model with a large lift-off influence is proposed for interpretation of denoising mechanism. To investigate the efficacy and the robustness of the proposed method, experimental studies have been carried out for different test samples. A comparative analysis has been undertaken to confirm that the proposed method not only removes the noise but also preserves the information of defect. The Matlab demo code can be linked: http://faculty.uestc.edu.cn/gaobin/zh_CN/lwcg/153392/list/index.htm .

Corrosion detection in steel bar: A time-frequency approach

Abstract: This paper presents a new technique for discerning corrosion in steel bars using guided ultrasonic waves with an improved signal processing technique. A mild steel bar has been subjected to accelerated corrosion. Information concerning variation in spectral traits associated with the spread of corrosion has been discerned. Dispersion curves for the steel bar have been deduced to understand and correlate the observed change in time-frequency spectra with shift in predominant modes with spread of corrosion. Theoretical and measured mass loss in the steel rebar is used as physical parameters to evaluate corrosion. Tensile tests of corroded specimen have also been undertaken to measure the residual strength of the corroded bars. The superiority of the proposed signal processing technique in clearly discerning the state of corrosion has been demonstrated.

Direct-write piezoelectric ultrasonic transducers for pipe structural health monitoring

Abstract: Novel direct-write ultrasonic transducers comprising piezoelectric poly(vinylidenefluoride/trifluoroethylene) [P(VDF/TrFE)] polymer coatings which were in-situ deposited, crystallized and patterned on pipe for structural health monitoring purpose. Lamb ultrasonic wave signals, generated and measured by the direct-write transducers and propagating along the pipe structure, were used to monitor the integrity of the pipe structure. The experimental measurements of the axial Lamb wave on pipe structure showed the substantial reduction in the ultrasonic signal by the defects. In addition, pipe thickness was accurately determined with the direct-write transducers to generate and detect the ultrasonic wave in the pipe thickness direction using pulse-echo mode. Our result and analyses suggest that implementation of the unique direct-write ultrasonic transducer technology is promising for realizing structural health monitoring for pipeline structures with improved consistency and reliability.

Deep neural networks based approach for welded joint detection of oil pipelines in radiographic images with Double Wall Double Image exposure

Abstract: This paper describes a method to support the field of Nondestructive Testing, especially, in radiographic inspection activities. It aims at detecting welded joints of oil pipelines in radiographs with Double Wall Double Image exposure. The proposed approach extracts information (windows of pixels) from the pipeline region in the radiographic image and then applies Deep Neural Network (DNN) models to identify which windows correspond to welded joints. We use pre-trained DNNs to map the knowledge from ImageNet Large Scale Visual Recognition Challenge to the welded joint context. The experiments consider 13 DNN models and 3 DNN input settings: stretched, proportional V and proportional H. Since, occasionally, radiographic images may be corrupted by some types of noise (e.g. white, impulsive), we also include experiments considering its influence on the DNNs behavior and its related results. The best combination provided an F-score average of 96.00% in the welded joint detection. The main contributions of this work are the proposed window extraction technique and a robust analysis of the noise influence on welded joint detection using different DNN models, input settings, exposure techniques and radiographic acquisition sources.