Abstract With the continuous improvement of the performance requirements for asphalt mixtures in the transportation construction industry and some special scenarios, how to efficiently and accurately obtain the Marshall parameters of asphalt mixtures has become a difficult problem in the field of transportation and roads. However, existing determination methods have problems such as poor accuracy and insufficient efficiency. Based on this, this article proposes a non-destructive determination model of Marshall Parameters of asphalt mixtures, which improves ultrasonic testing through multiple technical algorithms and integrates eXtreme Gradient Boosting. The results show that the determination coefficient and mean absolute error of stability of asphalt mixtures are 0.9374 kN and 0.124 kN. The accuracy of determining air voids is 96.11%, and the reuse rate of specimens is above 99.90%. At the same time, the temperature sensitivity coefficient is below 1.00%, the mean value of the sum of squared errors of surface state adaptability is 7.76%, and the average time is 2.11 min. The mean absolute percentage errors of Marshall Parameters under different pavement conditions are all lower than those of the comparison models. These results confirm the feasibility and practicality of the proposed model and provide a new non-destructive testing method for determining Marshall Parameters of asphalt mixtures in road engineering.
Abstract Structural health monitoring (SHM) using AI-assisted signal-processing models has demonstrated high efficiency and practicality for applications that are difficult to address using traditional signal-processing techniques. However, these models typically require large, annotated damage datasets that are time-consuming, expensive, and often impractical to obtain, making data scarcity a critical bottleneck in the field. A proposed framework presents a synthetic data generation framework for ultrasonic guided-wave signals in thin plates based on a conditional Wasserstein generative adversarial network with gradient penalty (WGAN-GP). The proposed framework generates realistic multichannel guided-wave signals using a limited number of experimentally acquired measurements. Guided-wave data were collected from an Aluminum plate instrumented with a sparse rectangular network of piezo transducers. In the first study, the model was trained using only 10% of clean guided-wave signals, while the remaining were synthetically generated under different defect scenarios, achieving agreement with experimental measurements, with correlation coefficients exceeding 0.95 and structural similarity index measure (SSIM) values above 0.97. In the second study, the framework generated signals for previously unseen subzones within the same inspection region, and in the third study, the learned wave propagation characteristics were evaluated on an unseen monitoring zone with different propagation characteristics. The proposed framework preserves statistically consistent spatial guided-wave characteristics, including temporal wave form evolution, and dispersive spectral energy behavior within the experimental configuration. Overall, the proposed approach substantially reduces the experimental effort while maintaining strong temporal and time–frequency consistency with experimentally measured guided-wave signals and offering a promising solution to data scarcity in SHM.
Abstract To meet the requirements of intelligent operation and maintenance for wind turbine main shaft systems, this study addresses the limitations of traditional methods, which exhibit low sensitivity to early defects and have difficulty predicting performance degradation. This study proposes a fault early warning and Remaining Useful Life (RUL) prediction model that integrates intelligent optimization and deep learning. In the fault early warning stage, an improved Particle Swarm Optimization algorithm dynamically optimizes the hyperparameters of the eXtreme Gradient Boosting. In the RUL prediction stage, a Multi-Scale Convolutional Neural Network extracts temporal features from degradation signals, and a Self-Attention mechanism highlights critical degradation stages. Experimental results indicate that the proposed model significantly reduces the Root Mean Square Error and Mean Absolute Error of the RUL prediction to 8.7 h and 6.3 h, respectively. Correlation coefficients between the predicted sequences and the true sequences are 0.978 and 0.971. The model maintains high prediction accuracy across the early-, middle-, and late-life stages. The proposed model is significant for improving the timeliness of fault early warning, the accuracy of prediction, as well as the stability in complicated conditions.
Abstract Good quality radiographic images are essential for accurate medical diagnosis and reliable defect detection in welds. As X-ray equipment ages, the radiographic spatial resolution decreases because the focal spot area (FSA) enlarges, increasing geometric unsharpness and degrading image quality. To compensate for this degradation, operators often increase the film exposure time based on personal experience. This paper introduces a method for the systematic assessment of a dimensionless aging correction factor for the standard exposure time. Key geometric, thermal, and physical parameters involved in the aging process are combined into a dimensionless thermo-radiographic coupling index derived using the Buckingham-Pi theorem. This index, which links the thermally-induced strain demand to radiographic degradation, is used to develop a logistic growth model describing the time-dependent evolution of the FSA. The model is used to estimate time-dependent aging factors, which are illustrated using nomograms and applied to aged industrial X-ray equipment. The corrected exposure improves crack detectability and visibility of thin wires in Image Quality Indicators (IQIs). The findings of this paper provide a more consistent and physics-based assessment of exposure time, enhancing the reliable operation of aged medical and industrial X-ray equipment.
This work uses staged X-ray computed tomography to track porosity evolution and fracture initiation in AISI 316L specimens produced by laser powder bed fusion under low global energy-density conditions, resulting in porous material dominated by lack-of-fusion (LOF) defects. Specimens were imaged in the undeformed state, after loading to peak force, and after fracture under interrupted monotonic tensile loading. Half of the specimens were machined to remove the as-built surface crust to isolate surface-condition effects. Global porosity statistics were similar between as-manufactured and machined groups, indicating that machining neither altered existing porosity nor introduced new detectable voids. Unsupervised K-means grouping of pore morphology separated gas-, LOF-, and keyhole-like populations and confirmed LOF-like porosity as the dominant contribution. Analysis of pore shapes indicates pronounced elongation and reorientation during loading, shifting from predominantly flat/transverse features in the undeformed state toward pores aligned with the loading direction at peak force and after fracture, while the largest pores grew markedly. In parallel, running-average effective cross-sectional area profiles highlight regions that later neck and/or contain large pore clusters. Crack-initiation mechanisms differed by surface condition: in as-manufactured specimens, cracks initiated from deep surface defects, while in machined specimens, cracks initiated when large subsurface LOF pores linked to the external surface. Overall, failure initiation is governed by the combined effects of surface condition, defect size, and defect-to-surface distance, underscoring that surface finishing and bulk defect mitigation must be addressed jointly.
The certification of adhesively bonded composite joints remains a significant challenge for aerospace applications, requiring reliable methods to assess bondline quality during manufacturing. This study validates in situ ultrasonic monitoring for real-time assessment of adhesive bondline thickness during autoclave cure. Three secondary-bonded laminates were fabricated using 26-ply unidirectional IM7G/8552 carbon fiber half-panels and FM209-1U unsupported film adhesive under varying bondline conditions: two adhesive plies without edge dam (Panel 1), four adhesive plies without edge dam (Panel 2), and four adhesive plies with edge dam (to reduce resin bleed at panel edges) (Panel 3). A custom ultrasonic scanning system (operational up to 180 degrees C) continuously monitored panels throughout autoclave cure, measuring time-of-flight (TOF) data at 1 mm & times; 1 mm resolution over 225 mm & times; 25 mm scan areas. A previously calibrated machine learning-based temperature correction factor compensated for temperature-dependent ultrasonic velocity changes through the composite adherends. Panel 1 exhibited approximately half the TOF of Panels 2 and 3, consistent with its reduced adhesive ply count. Panel 2 showed a characteristic dome-shaped thickness distribution resulting from resin bleed at panel edges, while Panel 3 maintained a uniform thickness distribution due to edge dam containment of the adhesive. Optical microscopy confirmed strong agreement between ultrasonic TOF measurements and post-cure bondline thicknesses. The study establishes that in situ ultrasonic monitoring, combined with appropriate temperature correction algorithms, provides a reliable quantitative assessment of adhesive bondline geometry during cure processes, offering potential for real-time quality control and manufacturing process optimization.
Accurate fault detection and diagnosis are critical components of any fault-tolerant control system, especially for Unmanned Aerial Vehicles (UAVs) where reliability is paramount. Traditionally, both model-based and data-driven approaches have been applied for fault diagnosis. However, the increasing complexity of high-dimensional UAV systems has shifted focus toward data-driven methods, which leverage advanced classification algorithms to enhance fault identification and isolation. This study builds on this evolution by developing a sophisticated condition-based monitoring (CBM) system specifically designed for multirotor UAVs. In contrast to earlier studies that primarily relied on raw data for classifier training, this work introduces advanced preprocessing techniques and multi-domain feature extraction, significantly improving the robustness and accuracy of fault detection. A comparative analysis is performed between feature selection methods, including Recursive Feature Elimination with Cross-Validation (RFECV) and Variational Autoencoder (VAE), to extract critical insights into UAV operational behavior. Through testing and evaluating various classification models on data from a hexarotor UAV under diverse actuator fault conditions, this research identifies optimal approaches for real-time fault detection and diagnosis. Results demonstrate notable improvements across all evaluation metrics, establishing this approach as a substantial advancement in UAV fault tolerance.
Shear-horizontal (SH) waves offer unique advantages for guided wave-based structural health monitoring applications. Their in-plane motion reduces sensitivity to environmental conditions. Additionally, the non-dispersive SH0 mode minimizes signal distortion. This article employs resonant metamaterials to focus SH0 waves before detection by a sensor. The objective is to amplify the signal and overcome attenuation of the mode as it propagates through a structure. The results show that flexural resonances of the unit cell modify SH modes and produce polariton-like dispersion behaviors. The dispersion curve for the SH mode through the metamaterial exhibits three distinct regions: non-dispersive, dispersive, and polariton. These regions are then used to design a frequency-selective metamaterial system for SH0 waves in a plate. The results represent the first demonstration of focusing SH0 modes using resonant metamaterials.
Structural integrity assessment of flawed pipes, as required by ASME B&PV Code Section XI, relies on failure stress calculated by the limit load criterion (LLC). The failure stress is derived from flow stress, which depends on yield and ultimate tensile strengths. Instrumented indentation technique (IIT) is a nondestructive alternative to conventional tensile testing. A large number of studies related to IIT reported that IIT is capable of assessing the yield strength and the ultimate tensile strength with 10% accuracy compared to the tensile testing. The benefit of IIT for the analysis of allowable flaw sizes is critically assessed in this study based on experimental data and code calculations. Intrinsic standard deviations of yield and ultimate tensile strengths for type 304 stainless steels are discussed in this article. As a result, the IIT method is capable of estimating the flow stress for type 304 stainless steel within 3.2% deviation from the result of conventional tensile tests. Then, the allowable circumferential flaw sizes for pipes subjected to tensile loading are calculated based on strength properties assessed by IIT. In addition, the allowable flaw sizes are compared to the allowable flaw sizes derived from the flow stress tabulated in the ASME B&PV Code Section II, Materials. The conclusion is drawn that IIT yields a sufficiently accurate estimate of the flow stress on austenitic stainless steel piping. Therefore, IIT is a beneficial method for LLC.
The material presented in this article summarizes the results from phase 2 of a larger Electric Power Research Institute (EPRI) study to determine if permanently mounted strain sensors can be used to monitor and quantify the presence and growth characteristics of flaw indications in primary loop piping in light water nuclear power reactors (LWRs). The samples utilized in this phase 2 work consisted of butt welded sections of austenitic 304 stainless steel pipe segments. Axial and circumferential notches, intended to simulate service flaws of varying depths and lengths, were generated into the inner diameter (ID) of the samples. Resistive strain gauges were applied in longitudinal (axial) and transverse (hoop) orientations, on the outer diameter (OD) in the vicinity of the notch locations. The axial notches perpendicularly intersected the weld while circumferential notches were generated along the weld fusion planes on one side (15 deg from the radial direction). The notch morphologies, strain gauge placement details, and a summary of the strains measured during pressurization tests are presented in this article. Based on the strain data obtained, transverse strain measurements may be sufficient to allow online monitoring (OLM) of axial and circumferential flaws in the vicinity of welds in the LWR pressurized primary coolant loop austenitic stainless steel piping and components.
Accurate diagnosis of outer race defects and their interaction with secondary faults remain a critical challenge in bearing condition monitoring. This study presents a physics-based diagnostic approach that integrates extended Hamilton's principle (EHP) with an improved one-against-all multiclass support vector machine (OAA-MCSVM) for identifying and classifying complex bearing faults. A dynamic model of the rotor-bearing system is developed using EHP to capture the influence of outer race defects under combined fault scenarios, including misalignment, unbalance, and radial clearance variation. Vibration responses are acquired from a controlled test rig under isolated and coupled fault conditions. Fault signatures are extracted through time-frequency analysis and mapped to system dynamics derived from the variational formulation. The extracted features are classified using the improved OAA-MCSVM framework, which enhances boundary discrimination between closely interacting faults. Experimental validation shows that the proposed method achieves high classification accuracy across all tested fault conditions, with improved sensitivity to outer race-related compound faults. The integration of physics-based modeling with machine learning enables a more interpretable and reliable fault diagnosis scheme suitable for real-time application in rotating machinery.
Cables are a key component of power systems. Local defects such as insulation aging and mechanical damage can easily cause serious faults. Impedance spectroscopy analysis has become a research hotspot in defect detection due to its noninvasive nature and sensitivity. However, traditional Fourier transforms have problems such as weak noise suppression, low characteristic resolution, spectral leakage, and insufficient low-frequency resolution, which limit the ability to detect minor defects. To address this limitation and achieve high-precision detection and location of local defects in cables, this article adopts an improved method based on the inverse fast Fourier transform (IFFT). The equivalent time component (2l/v) is extracted from the reflection coefficient spectrum at the beginning of the cable, and the time-frequency resolution is optimized by combining the adaptive window function. The defect characteristic frequency band is enhanced through the frequency-domain weighting algorithm, and the location function DF(x) is constructed to highlight the defect differences. Meanwhile, the signal propagation speed is simulated and measured by comsol software to provide key positioning parameters. Experiments show that this method is not sensitive to the strength of the injected signal (as long as the minimum power is met), but sensitive to the severity of defects. It can accurately locate multiposition defects (such as defects at 10 m, 20 m, and 40 m on a 60 m XLPE cable) and is not affected by cable curls. It is suitable for high-voltage cables and radio frequency (RF) cables with semiconductive layers.
This work investigates strategies to enhance full waveform inversion (FWI) of ultrasonic signals for high-resolution imaging and defect detection of composites. Various misfit measures-including L-2-norm, L-1-norm, cross-correlation, and envelope measures-are compared to assess their effectiveness in improving FWI outcomes. Additionally, different regularization methods, such as Tikhonov (L-2 norm), LASSO (L-1 norm), and total variation regularization, are examined for stabilizing FWI and enhancing its reconstruction. Parameterization techniques, like transforming model parameters to Sigmoid space, are explored to improve reconstruction accuracy and convergence. The methodology is demonstrated through two case studies: reconstructing a delamination defect in a sandwich composite and reconstructing a high-resolution image of a highly heterogeneous reinforced concrete beam. While the delamination defect was not fully reconstructed, the results indicate that combining cross-correlation misfit measures with total variation regularization and Sigmoid parameter transformation results in significantly better reconstructed images of composites and identifying sharp discontinuities in multilayered and heterogeneous structures.
The crude oil pretreatment system is a critical step in crude oil processing, and it is responsible for removing impurities to ensure that the quality meets the standard. Its fault can lead to equipment damage, decreased product quality, production interruptions, and environmental pollution. Therefore, it is crucial to promptly identify the cause of failures and analyze the current reliability of the system. To address this urgent issue, this article designs a fault diagnosis model based on object-oriented Bayesian networks and a system reliability assessment model based on dynamic Bayesian networks. By utilizing a fault diagnosis Bayesian network, both single and multiple faults in the system are successfully identified. Subsequently, the output results of the fault diagnosis network are used as key information input into a reliability assessment Bayesian network, enabling an in-depth and comprehensive analysis of the system's reliability status during fault occurrences. This integrated method not only possesses fault diagnosis and reliability assessment capabilities but also precisely analyzes the weak links in the system, providing valuable and practical suggestions and guidance to on-site operators and maintenance teams.
The structural integrity of canisters that store industrial chemicals, fuels, gases, and spent nuclear fuel can be monitored reliably and remotely, either periodically or on demand, with helically propagating guided ultrasonic waves (HGUWs). Specifically, by generating and capturing HGUWs along a network of paths within the structure, damage areas can be identified from a tomographic map. Researchers have used this method to map corrosive damage, such as wall thinning, on canister shells. Although these studies have shown the effectiveness of this technique for long-term canister monitoring, it has not yet been tested on a critical part of the structure-namely, the weld zones. Welded regions in canisters are critical and prone to defects because factors like residual stresses, material heterogeneity, and geometric discontinuities increase their susceptibility to stress corrosion cracking and fatigue failure. Therefore, this work explores the feasibility of using HGUW-based tomography to detect damage within canister welds. Experiments were performed on a 5.6-in.-tall, 18-in.-diameter canister section to locate three different types of damage in the circumferential weld. The first damage simulated weld erosion, the second mimicked a shallow surface flaw, and the third represented an early-stage defect such as a crack. Using two arrays of circumferentially distributed piezoelectric (PZT) discs, HGUWs were generated and recorded across a dense network of paths. The wave data from all paths were then used to create tomographic maps with the reconstruction algorithm for probabilistic inspection of damage (RAPID) algorithm. These maps successfully pinpointed each damage, thus demonstrating the method's effectiveness in monitoring canister weld integrity. Ultimately, through these tests, this study establishes HGUW-based tomography as a viable tool for comprehensive monitoring of real-world canisters.
This paper proposes a novel Machine Vision (MV) - based technique to assess surface roughness (R-a) as an Online Diagnostic and Monitoring system tool. Such a tool helps reduce operational downtime and minimise human intervention. The proposed method involves image acquisition in a controlled environment, detection of surface features through a parameter called "Edge frequency" using the Edge Detection Algorithms (EDAs), and developing a correlation between the estimated Edge frequency and the experimentally measured Ra. Several well-known EDAs are evaluated on different samples, including emery papers with various grit values and turbomachinery blades. The superiority of LoG (Laplacian of Gaussian)-based EDA in terms of its resilience to noise and computational benefit is demonstrated. The Edge frequency-R-a correlation is subsequently used to predict the R(a )value of different sets to demonstrate prediction accuracy. Compared to contact-based measurements, the predictions on the emery samples are within 4.8%, while those on the blade samples are within 5.9%. Finally, the established correlation is used for the online diagnosis of a Legacy axial compressor under cleaned and fouled conditions. The predicted Ra values for the samples and the trends agree with the historical data reported by the Original Equipment Manufacturers (OEMs), demonstrating its ability as an On-board Diagnostic (OBD) tool.
Accurate fault detection and diagnosis are critical components of any fault-tolerant control system, especially for unmanned aerial vehicles (UAVs) where reliability is paramount. Traditionally, both model-based and data-driven approaches have been applied for fault diagnosis. However, the increasing complexity of high-dimensional UAV systems has shifted focus toward data-driven methods, which leverage advanced classification algorithms to enhance fault identification and isolation. This study builds on this evolution by developing a sophisticated condition-based monitoring (CBM) system specifically designed for multirotor UAVs. In contrast to earlier studies that primarily relied on raw data for classifier training, this work introduces advanced preprocessing techniques and multidomain feature extraction, significantly improving the robustness and accuracy of fault detection. A comparative analysis is performed between feature-selection methods, including recursive feature elimination with cross-validation (RFECV) and variational autoencoder (VAE), to extract critical insights into UAV operational behavior. Through testing and evaluating various classification models on data from a hexarotor UAV under diverse actuator fault conditions, this research identifies optimal approaches for real-time fault detection and diagnosis. Results demonstrate notable improvements across all evaluation metrics, establishing this approach as a substantial advancement in UAV fault tolerance.
In this article, a novel method for modal response estimation and response reconstruction is proposed. The method is based on a newly developed modal response estimation method and transformation equations derived from the modal information of the target. The modal response estimation method is designed to acquire optimal modal responses using responses from only one sensor. Based on the derived optimal modal responses of convenient locations, the modal responses of critical locations can be extrapolated using the transformation equations. To demonstrate the overall reconstruction procedure, two numerical examples are presented, including cases of well-separated modes and closely spaced modes. The effects of mode numbers, sensor numbers, sensor locations, and noise levels are investigated in detail. Following this, a realistic turbine blade structure is used to validate the effectiveness and accuracy of the method in practical applications. The results indicate that the proposed method is accurate and reliable for response reconstruction, offering a viable alternative for structural health monitoring of various structures.
High-density polyethylene (HDPE) is a semicrystalline polymer used in several critical applications, ranging from cooling water pipelines in nuclear power plants and distribution pipelines for natural gas and hydrogen to biomedical implants. Embedded crack-like flaws form within HDPE during fabrication or operations, which may grow over time and can cause catastrophic failure if undetected. Large structures such as HDPE pipelines, where the location of a flaw is not known, require a fast, nondestructive evaluation (NDE) method where the sensor can move rapidly across the structure with a very short data collection window of microseconds at each location. This is only possible if the flaw is evaluated in HDPE and other polymeric structures using a microsecond time signal. Ultrasonic A-scan (time signal) allows for the rapid scan of large structures, whereas B-scan ultrasounds are limited, as they are slow and depend on postprocessing algorithms, where subtle information can be lost. We propose a methodology for training a convolutional neural network (CNN) using computer simulations of ultrasound on HDPE and applying the trained CNN to real-life experiments to decipher crack characteristics in HDPE or other polymer structures using ultrasound time (A-scan) signals. We show that a fully finite element simulation-trained CNN can accurately predict crack lengths (mean absolute percent errors (MAPE) 3.2%) and positions (MAPE 3.8%) in HDPE from experimentally measured ultrasound A-scan microsecond signals. The success of a 100% simulation-trained CNN without exposure to any prior experimental data in accurately predicting crack sizes from experimental time signal data underscores a promising path for next-generation NDE methodologies.
Threaded pipe structures are critical components in equipment used in various fields. The threaded sections are prone to induce defects due to stress concentration, threatening the safe operation of the equipment. Consequently, the inspection of these structures is essential. While the ultrasonic guided wave (UGW) method has been applied to inspect threaded pipes, the influence of thread parameters on UGW propagation characteristics is yet to be examined. This study first analyzed the dispersion characteristics of UGWs. It reveals that the group velocity dispersion curve of the L(0,2) mode initially increases and then decreases over the frequency range of 60–140 kHz. The group velocity is higher in trapezoidal threads than in rectangular threads. Dispersion curves for different thread heights exhibit a crossover near 85 kHz. Dispersion curves for different pitches intersect around 83 kHz. Second, the effect of thread parameter variations on the reflection characteristics was investigated. It was found that the trapezoidal thread exhibits a higher reflection coefficient than the rectangular thread. The reflection coefficient increases with the thread height and decreases with pitch. Third, the influence of thread parameters on defect detection sensitivity was examined. Results demonstrate that the L(0,2) mode offers high sensitivity for defect detection in threaded pipes featuring trapezoidal threads, a thread height of 1 mm, and a pitch of 4.5 mm. Finally, the effectiveness of the L(0,2) mode in detecting defects of varying depths within threaded pipes was validated. This research provides a novel method for the inspection of threaded pipe structures.
