Measurement Science and Technology

A customised dual-wire sensor, incorporating both a cold-wire and a hot-wire, has been developed to enable simultaneous measurement of temperature and velocity at the same wall-normal location within a thermal boundary layer. This study details the sensor's design, its in-situ calibration technique, and addresses potential sensor drift, offering guidance for its effective utilisation. While the results obtained from this sensor confirm the similarity between the wall-normal profiles for mean temperature and velocity, profiles of the temperature variance present a less obvious plateau in the logarithmic region when compared to the streamwise velocity variance. The simultaneously acquired instantaneous signals also reveal the markedly different characteristics of the streamwise velocity and temperature fluctuations at the edge of the boundary layer, highlighting the dissimilarity between velocity and temperature (with the latter exhibiting a unique one-sided distribution).

Time-gated (TG) Raman spectroscopy (RS) has been shown to be an effective technical solution for the major problem whereby sample-induced fluorescence masks the Raman signal during spectral detection. Technical methods of fluorescence rejection have come a long way since the early implementations of large and expensive laboratory equipment, such as the optical Kerr gate. Today, more affordable small sized options are available. These improvements are largely due to advances in the production of spectroscopic and electronic components, leading to the reduction of device complexity and costs. An integral part of TG Raman spectroscopy is the temporally precise synchronization (picosecond range) between the pulsed laser excitation source and the sensitive and fast detector. The detector is able to collect the Raman signal during the short laser pulses, while fluorescence emission, which has a longer delay, is rejected during the detector dead-time. TG Raman is also resistant against ambient light as well as thermal emissions, due to its short measurement duty cycle.

In recent years, the focus in the study of ultra-sensitive and fast detectors has been on gated and intensified charge coupled devices (ICCDs), or on CMOS single-photon avalanche diode (SPAD) arrays, which are also suitable for performing TG RS. SPAD arrays have the advantage of being even more sensitive, with better temporal resolution compared to gated CCDs, and without the requirement for excessive detector cooling. This review aims to provide an overview of TG Raman from early to recent developments, its applications and extensions.

Camera calibration is a key component of three-dimensional particle tracking velocimetry (PTV) experiments, and its proper implementation is key to the success of the method. In this paper, we review and compare four different camera calibration models used in PTV experiments without volumetric refinement. One of the calibration models is new and provides an analytical inversion of the Soloff polynomial. The other three calibration models are taken from three established open source PTV frameworks: OpenPTV, MyPTV and proPTV. In particular, we present a general formulation of calibration models that allows their rigorous comparison and evaluation with respect to their 3D-to-2D projection errors and 2D-to-3D reconstruction errors. We compare the models and the calibration errors in three different tasks, including extrapolation and interpolation of marker points, using a realistic calibration of an experimental camera setup. In the end, we conclude with the pros and cons of each method in order to be able to choose the most suitable one for individual needs.

High-resolution reconstruction of flow-field data from low-resolution and noisy measurements is of interest due to the prevalence of such problems in experimental fluid mechanics, where the measurement data are in general sparse, incomplete and noisy. Deep-learning approaches have been shown suitable for such super-resolution tasks. However, a high number of high-resolution examples is needed, which may not be available for many cases. Moreover, the obtained predictions may lack in complying with the physical principles, e.g. mass and momentum conservation. Physics-informed deep learning provides frameworks for integrating data and physical laws for learning. In this study, we apply physics-informed neural networks (PINNs) for super-resolution of flow-field data both in time and space from a limited set of noisy measurements without having any high-resolution reference data. Our objective is to obtain a continuous solution of the problem, providing a physically-consistent prediction at any point in the solution domain. We demonstrate the applicability of PINNs for the super-resolution of flow-field data in time and space through three canonical cases: Burgers' equation, two-dimensional vortex shedding behind a circular cylinder and the minimal turbulent channel flow. The robustness of the models is also investigated by adding synthetic Gaussian noise. Furthermore, we show the capabilities of PINNs to improve the resolution and reduce the noise in a real experimental dataset consisting of hot-wire-anemometry measurements. Our results show the adequate capabilities of PINNs in the context of data augmentation for experiments in fluid mechanics.

The growing complexity of machinery and the increasing demand for operational efficiency and safety have driven the development of advanced fault diagnosis techniques. Among these, convolutional neural networks (CNNs) have emerged as a powerful tool, offering robust and accurate fault detection and classification capabilities. This comprehensive review delves into the application of CNNs in machine fault diagnosis, covering its theoretical foundation, architectural variations, and practical implementations. The strengths and limitations of CNNs are analyzed in this domain, discussing their effectiveness in handling various fault types, data complexities, and operational environments. Furthermore, we explore the evolving landscape of CNN-based fault diagnosis, examining recent advancements in data augmentation, transfer learning, and hybrid architectures. Finally, the future research directions and potential challenges to further enhance the application of CNNs for reliable and proactive machine fault diagnosis are highlighted.

Advanced three-dimensional (3D) tracking methods are essential for studying particle dynamics across a wide range of complex systems, including multiphase flows, environmental and atmospheric sciences, colloidal science, biological and medical research, and industrial manufacturing processes. This review provides a comprehensive summary of 3D particle tracking and flow diagnostics using digital holography (DH). We begin by introducing the principles of DH, accompanied by a detailed discussion on numerical reconstruction. The review then explores various hardware setups used in DH, including inline, off-axis, and dual or multiple-view configurations, outlining their advantages and limitations. We also delve into different hologram processing methods, categorized into traditional multi-step, inverse, and machine learning (ML)-based approaches, providing in-depth insights into their applications for 3D particle tracking and flow diagnostics across multiple studies. The review concludes with a discussion on future prospects, emphasizing the significant role of ML in enabling accurate DH-based particle tracking and flow diagnostic techniques across diverse fields, such as manufacturing, environmental monitoring, and biological sciences.

In this paper, we focus on utilizing deep learning methods for the localization of near-field signals in an unknown colored noise, which has fundamental applications in array signal processing. Specifically, a new Vision Transformer-based near-field signal localization approach (VT-NSL) is proposed by taking advantage of the self-attention mechanism of the Transformer, which circumvents the limitation of local perceptual domains in traditional convolutional neural networks. This enables VT-NSL to possess outstanding capability in capturing global information and effectively enhance the learning of critical information in colored noise environments. In contrast to typical data-driven estimation approaches that treat source localization as a multi-label task, VT-NSL takes the sample covariance matrix of the array's received signal as input and directly outputs the localization parameters through a regression structure, which improves the training efficiency without affecting the estimation accuracy. Additionally, we explicitly derive the closed-form Cramér–Rao lower bound in the unknown colored noise environment. Extensive simulation results demonstrate the superior performance of VT-NSL compared to existing data-driven methods and traditional model-driven methods.

Accurate measurement of small forces on large objects under experimental laboratory conditions is challenging due to the limited sensitivity of available instruments and adaptable methods. For example, in wind tunnel testing for aerodynamic performance, the generation of very small lift or drag forces on the test object often proves difficult to quantify. Recent advancements in maglev technology have led to its application in wind tunnel studies, where it is used to mitigate the interference on measurements caused by support systems. However, the accuracy of measurements may be compromised due to systemic errors affecting the performance of force sensors in the wind tunnel, such as mechanical vibrations, misalignments, and contact deformations. To address these issues in the measurement of small forces during wind tunnel experiments, this paper proposes a contactless method for measuring small horizontal forces based on current variation in the levitation coil, which can be generally applied in maglev systems. To investigate and understand the changes in the coil with applied control system, a coil-to-permanent-magnet maglev model with applied horizontal force has been developed and studied. Both software simulations and experimental validations are used to ensure the validity of the proposed method.

Railway track structural health monitoring (SHM) plays a critical role in modern railway systems. Axle box acceleration (ABA) signals serve as a valuable information source for track SHM. However, the relationship between ABA frequency components and train–track dynamics, especially the muti-tone harmonics interaction, remains underexplored. This paper investigates the frequency components of ABA signals with time–frequency analysis methods, including short-time Fourier transform, continuous wavelet transform, and wavelet synchro squeezed transform. Through experimental field studies on a ballasted railway track and an embankment–bridge transition zone, the interactions between the sleeper passing frequency (${f_{{\text{sp}}}}$) and wheelset rotation frequency (${f_{{\text{wr}}}}$) are identified as the result of multi-tone harmonic coupling. The revealed empirical coupling relationship is shown by ${f_{\text{c}}} = \left( {1/2} \right){C_1}{f_{{\text{sp}}}} \pm \left( {1/2} \right)$ ${C_2}{f_{{\text{wr}}}}$, which remains consistent across different train speeds on the embankment and could also be observed on bridge sections. Additionally, the influence of differential settlement in the transition zone on the collected ABA signals from experimental campaigns has also been considered. The novelty of this paper lies in the identification of multi-tone harmonic interactions in ABA signals, and the empirical validation of frequency coupling relationships across varying track conditions using the time–frequency analysis methods. These contributions give insights into practical applications for ABA signal analysis and potentially offer a guideline for future practical applications.

Well defined multi-property measurements are crucial for the quantification of flow within all facilities. Conventionally, these measurements are collected using physically intrusive pressure and temperature probes which introduce undesired perturbation to flow and measurement. More recently, optical techniques have been employed to supplement probe measurements, with most applications requiring particles to be seeded into the flow. In this contribution, the advancements required to integrate a molecular filtered Rayleigh scattering (FRS) instrument into a full-scale flow facility are discussed, laying the foundation for accurate and precise measurements in the presence of high intensity particle and background scattering, the primary sources of bias in applied FRS. Measurements are demonstrated in the 432 mm diameter Annular Cascade Facility at Wright-Patterson Air Force Base, downstream of a distortion generating test article, resulting in swirl and total pressure distortion. Multi-property FRS measurements of time-averaged three-component velocity, temperature, and density in addition to derived static and total pressure are compared with standard 5-hole probe and particle image velocimetry measurements. Comparison between the FRS and reference measurements resulted in differences typically falling within pre-determined uncertainty bounds. Root-mean-square difference (RMSD) values are shown to be less than 8.0 m s⁻¹, 1.7 deg, and 4.7 deg for axial velocity, tangential flow angle, and radial flow angle respectively. Measurements of static temperature and density are found to show RMSD values of 11.1 K (4.0%) and 0.043 ${\text{kg}}\,{{\text{m}}^ - }^3$ (4.3%) respectively compared to the expected average. FRS-derived static and total pressures are compared with reference measurements with good agreement near the center of the duct with RMSD values of less than 3.9 kPa (4.8%) and 5.0 kPa (4.9%) respectively.

In recent years, Doppler through-wall radar has been widely used in target localization. However, time–frequency aliasing occurs, which poses a great challenge to the estimation of target parameters. To solve this problem, we propose a novel time–frequency enhancement algorithm. In this paper, we roughly identify the time-frequency features of the interested target component from the time–frequency distribution of the local maximum synchrosqueezing transform and construct a series of discrete demodulation operators that approximate the true properties. Then an optimized window function with adaptive parameters (window width) is designed according to its time–frequency features, which can help effectively compress the echo energy from the selected target component. Finally, an instantaneous frequency (IF) estimator with a local maximum extraction criterion is designed to reduce the noise interference. The experimental results demonstrate that the presented algorithm successfully mitigates time–frequency aliasing, significantly enhancing the accuracy of IF estimation and target localization.

Three-dimensional (3D) localization and single-particle tracking (SPT) are powerful techniques for characterizing shape, motion, and deformation of materials. However, these methods often require complex optical setups that demand expert use, limiting their accessibility to the wider scientific community. This paper presents a new technique called individualized defocusing particle tracking (IDPT), which uses intrinsic aberrations in ordinary lab microscopes to perform 3D surface measurements at camera frame rates. The IDPT technique comprises a simple in-situ calibration procedure and 3D localization algorithm that leverages particles' unique defocusing patterns to enhance measurement sensitivity while compensating for optics-induced bias errors. Our empirical approach implicitly transforms systematic optical effects—including those caused by misalignment or defects of optical elements—into valuable sources of positional information, allowing our method to work with common lab microscopes. We validate the IDPT technique through synthetic and benchtop experiments involving rigid body motion of a planar substrate and the dynamic deformation of elastic discs, demonstrating that our algorithm surpasses comparable SPT algorithms in accuracy and resolution. The IDPT technique is simple yet robust, offering broad applicability for dynamic surface profilometry and deformation analysis.

GNSS and accelerometer technologies are widely used in deformation monitoring, such as earthquakes, landslides and bridges. However, the positioning accuracy of traditional GNSS and accelerometer Kalman filters may be decreased due to the continuous changes of ground tilt during deformation growth. To address this issue, we propose an improved GNSS/accelerometer deformation monitoring algorithm based on denoised tilt angle information. Firstly, since the calculation of the tilt angle by the accelerometer is highly susceptible to external motions such as ground vibrations caused by landslides, this paper employs the Savitzky–Golay filter to enhance the accuracy of the tilt angle calculated by the accelerometer. Then, an angle cosine matrix is constructed based on the obtained tilt angle to correct the original accelerometer signals. In this way, the positioning results can be improved through the fusion of GNSS and cleaner acceleration data. The simulated deformation experiments were conducted to validate the proposed algorithm.The results showed that the root mean squared error (RMSE) of pitch and roll angles were 0.185° and 0.169° in low dynamic conditions, respectively, which were improved by 76.36% and 76.13%, respectively, compared with the traditional method. Furthermore, the positioning RMSE of East–North–Up (E, N and U) directions presented 0.011 m, 0.009 m, and 0.006 m, respectively, which were 88.78%, 95.93% and 91.78% over that of the traditional Kalman filter. Furthermore, the algorithm has been effectively verified in a real landslide environment. Due to the acquisition of more 'pure' acceleration data after inclination correction, the positioning accuracies of the proposed algorithm in the E, N and U directions have been increased by 16.22%, 37.20%, and 83.73%, respectively, compared with the traditional Kalman filtering algorithm.

Bearings are critical components in equipment, and it is important to predict their remaining useful life (RUL) so that early intervention can be applied before their failure. A challenge in predicting the RUL is that the target bearing's working condition and/or failure mode is usually unknown. Although various methods have been proposed to address this challenge, few of them consider two practical issues. (1) When various datasets are available and used to train a single deep learning model, training datasets related to bearings of quite different statuses can affect prediction accuracy. (2) Transfer learning can be used to alleviate issue (1), but this requires data collection from the target bearing and fine-tuning, and the additional time required by this process may delay RUL prediction and intervention. To address these issues in industrial practice, we propose stacking-based RUL (StRUL) prediction for bearings of unknown working conditions and failure modes. StRUL is based on the stacking of transformers with novel designs: a modified amplitude spectrum comparison approach, a similarity-based attention mechanism, and a distribution-based attention mechanism. First, StRUL pre-trains each transformer with a specific dataset so that each transformer can generate an encoding for the input data from the target bearing. Second, it applies a modified amplitude spectrum comparison approach to calculate the similarity value between the input data and each transformer's training dataset. StRUL then uses a similarity-based attention mechanism to prioritize transformer encodings with relatively large similarity values in prediction. Third, it includes an additional transformer trained with all the training datasets and uses a distribution-based attention mechanism to determine how much the additional transformer contributes to the predicted RUL when the distribution of the similarity values is nearly uniform. Case studies performed using the XJTU-SY and PRONOSTIA data, each containing more than 10 datasets, demonstrate that StRUL can efficiently use all the training datasets without additional data collection or fine-tuning to achieve high prediction accuracy and speed useful for practical deployment.

The whispering-gallery mode microresonator self-injection locking lasers for atomic comagnetometers have excellent characteristics, and their stability needs to be improved. In this study, through the analysis and simulations of optical injection rate equations with thermal dynamics of the microresonator, a frequency adjustment strategy is presented that ensures the stability of the self-injection locking laser output wavelength over time. The normalized Allan deviation of the output wavelength is 1.7 × 10⁻¹⁰, and both the 20.6 $\text{Hz}/\sqrt{\text{Hz}}$ laser frequency noise and the 375 Hz linewidth are maintained within one hour. These results demonstrate exceptional self-injection locking stability with the 10⁹ quality factor microresonator.

Traditional physical models face significant challenges in parameter determination and adapting to complex systems, whereas data-driven models are constrained by data quality and quantity, making them susceptible to overfitting or underfitting problems. These limitations lead to deficiencies in robustness, physical interpretability, and generalization capabilities of existing models. In recent years, the fusion of physical mechanism-based and data-driven approaches has effectively addressed the shortcomings of both types of models, attracting widespread attention. However, there is no systematic review specifically on bearing prognosis and health management under hybrid physical mechanism and data-driven methods. To fill this gap, this paper comprehensively analyzes the research advancements in bearing prognosis and health management based on hybrid physical mechanism and data-driven methods. From the perspective of fusion strategy, the paper categorizes bearing prognosis and health management methods based on the fusion of physical mechanism and data-driven model into three levels: data level, network level, and model level, and further subdivides the research methods at each fusion level. In each subdivision field, this paper discusses the application of each research method in three main aspects: condition monitoring, fault diagnosis, and remaining useful life prediction, summarizing the research methods employed by current scholars. Finally, this paper evaluates the advantages and disadvantages of each analytical method in practical applications, identifies current research challenges, and proposes future research directions. The aim is to provide guidance and in-depth insights for researchers and engineers in the field of bearing prognosis and health management.

Rotating machinery holds a pivotal role in industrial and daily life. Incorporating robust health monitoring or fault diagnosis methodologies is imperative to safeguard such equipment's reliability and operational availability. Digital twin (DT) technology, which mirrors the real-time condition of a physical entity through a digital counterpart, has garnered considerable attention, particularly in the realm of fault diagnosis for rotating machinery. Despite its growing significance, there is a notable lack of comprehensive literature reviews pertaining to this domain. Consequently, this paper thoroughly reviews current and emerging DT technology-driven rotating machinery fault diagnosis trends. Initially, the origin and evolution of DT, alongside its application in fault diagnosis across diverse areas, are elucidated. Subsequently, the pivotal technologies and extant solutions within the DT-driven fault diagnosis purview are inspected. Furthermore, the literature on DT-driven rotating machinery for real-time condition monitoring, fault diagnosis, and life prediction is examined through varying tasks and application imperatives. Ultimately, the extant challenges of besetting DT-driven rotating machinery fault diagnosis techniques are defined, and their prospective evolutionary trajectories are appraised.

To improve the accuracy of bearing state classification in multi-source monitoring environments, a novel decision fusion method based on probabilistic Jensen–Shannon (PJS) divergence and D-S evidence theory is proposed. Firstly, the initial state probability of each bearing is determined by the bearing state classification score value, and the basic probability assignment (BPA) value of each bearing state is set by obtaining the global credibility of each sensor based on the pre-network test results. Secondly, the PJS divergence is proposed on the basis of the Kullback Leibler (KL) divergence to measure the conflict between the data collected by each sensor, and the weight value of the BPA value of each sensor is reset according to the PJS divergence value, which improves the accuracy of the BPA value. Finally, the D-S evidence theory is used to fuse the BPA value of each sensor and obtain the final bearing state classification result. The experimental part shows the effectiveness using the bearing signals collected in the laboratory, and the generality of this paper's method is verified by using it in conjunction with other networks.

Advanced three-dimensional (3D) tracking methods are essential for studying particle dynamics across a wide range of complex systems, including multiphase flows, environmental and atmospheric sciences, colloidal science, biological and medical research, and industrial manufacturing processes. This review provides a comprehensive summary of 3D particle tracking and flow diagnostics using digital holography (DH). We begin by introducing the principles of DH, accompanied by a detailed discussion on numerical reconstruction. The review then explores various hardware setups used in DH, including inline, off-axis, and dual or multiple-view configurations, outlining their advantages and limitations. We also delve into different hologram processing methods, categorized into traditional multi-step, inverse, and machine learning (ML)-based approaches, providing in-depth insights into their applications for 3D particle tracking and flow diagnostics across multiple studies. The review concludes with a discussion on future prospects, emphasizing the significant role of ML in enabling accurate DH-based particle tracking and flow diagnostic techniques across diverse fields, such as manufacturing, environmental monitoring, and biological sciences.

Large volumes of light sources and their characteristics control devices are challenges to out-of-laboratory applications of quantum precision measurement devices, such as spin-exchange relaxation-free (SERF) inertial and extremely weak magnetic field measurement devices. High-Q whispering gallery mode (WGM) resonators have excellent optical field confinement capability, extremely narrow linewidth, and small mode volume, which can easily realize compact low noise and high-frequency stability light sources, having a great potential to realize more compact quantum precision measurement devices. This will broaden the use of quantum precision measurement devices in areas such as transportable SERF gyroscopes, wearable magnetometers, etc. In this review, the fundamentals and characteristics of WGM optical cavities are firstly introduced, as well as its fabrication techniques and built-up materials. Then, the laser stabilizing applications of WGM optical cavities including wavelength tuning, self-injection locking, and thermal stabilization are discussed. Finally, the application prospects of quantum precision measurement are discussed.

Ground-airborne frequency domain electromagnetic (GAFDEM) is an important tool for mineral resource exploration and geological surveys in complex mountainous areas. The rapid and high-precision calculation of background resistivity is the foundation for high-resolution imaging. Due to the influence of attitude noise, the current accuracy of background resistivity calculations is not high. Furthermore, when the measurement area is large and the terrain is complex, the need to establish a terrain model for the measurement area leads to a decrease in efficiency. To solve the above problems, this paper proposes a method to calculate the background resistivity using the total magnetic intensity. Firstly, the magnetic field component is analyzed for its ability to identify anomalies, then the effects of transmitting frequency and source-receiver distance parameters are analyzed, and finally, simulation experiments and field experiments are conducted. Simulation experiments show that the method of calculating background resistivity using total magnetic intensity can accurately calculate the background resistivity of the double anomaly model and the model with topographic relief compared with the current method. Field experiments show that compared with the current method, the total magnetic intensity method of calculating background resistivity is not affected by attitude noise, and the results of background resistivity calculation are consistent with the real results with higher accuracy. The method in this paper supports GAFDEM to realize high-resolution and fast exploration of underground structures.

Conductive heterostructures are broadly adopted in engineering fields involving energy, aerospace, etc., in virtue of their superior performance by integrating the favourable properties of different metals. However, due to hostile and particularly electrochemical environments, corrosion would occur within the body of the conductive heterostructure. Such interlaminar corrosion severely undermines the integrity and thus the performance of heterostructures. In light of this, it is imperative to non-intrusively detect and evaluate the interlaminar corrosion in conductive heterostructures via non-destructive evaluation methods before it progresses further. Following the proposition of pulse-modulation eddy current (PMEC) testing, in this paper the incident-field self-differential mechanism is intensively investigated, and exploited to achieve higher testing sensitivity to interlaminar corrosions. Based on the structure of the conventional pancake probe, an opposing excitation coil is introduced for cancelling out the localized incident field at the sensor position. A semi-analytical model is established for assessment regarding the PMEC response to a conductive heterostructure subject to the interlaminar corrosion. Closed-form solutions for prediction of testing signals are formulated and subsequently utilised for investigation of the probe performance. In a bid to further analyse the performance of the proposed probe, experiments are carried out, in comparison with the conventional pancake probe. The results from the theoretical simulations and experiments infer that in virtue of the incident-field self-differential mechanism the proposed PMEC probe is superior in high-sensitivity recognition of interlaminar corrosions in conductive heterostructures.

Remaining useful life (RUL) prediction is critical for ensuring the safe and efficient operation of rotating machinery, which plays a vital role in industrial production. However, the traditional transformer model, relying on a simple linear output in its final layer, is limited in capturing nonlinear relationships. The nonlinear Wiener process (WP) model, despite its strengths in characterizing stochastic degradation, fails to fully capture the intricate time-series features in equipment operation data. To address these challenges, this paper introduces an improved method for RUL prediction which combines transformer and Kolmogorov-Arnold network (KAN) and utilizes WP for uncertainty quantification. The transformer can efficiently capture the long-term dependencies in equipment degradation data through its powerful sequential modeling capability and global attention mechanism, and its combination with the KAN further enhances the flexibility and accuracy of feature extraction. In this paper, the transformer-KAN model is proposed as a drift function for WP. Meanwhile, this paper utilizes the first hit time (FHT) to derive an approximate expression for the RUL probability density function (PDF) and estimate the drift and diffusion coefficients of the transformer-KAN-WP model. The method is validated on the bearing dataset and the tool holder power head degradation data and compared and analyzed with other commonly used methods, which proves the effectiveness of the method in improving the prediction accuracy.

Hydraulic systems are inherently complex and nonlinear, often prone to subtle, concurrent faults. These characteristics pose challenges for fault diagnosis, especially when time-domain data are limited. This paper studies multi-fault diagnosis problem in hydraulic systems, and proposes a novel multi-fault diagnostic framework, namely, Multi-Channel Multi-Modal Attention Fault Diagnosis Network (MC-MM-AFDN), to improve the diganosis accuracy with limited data. MC-MM-AFDN employs a parallel multi-channel architecture to extract and fuse sensor data features across different sampling frequencies. Specifically, each channel in MC-MM-AFDN adopts a dual-branch structure, with one processes temporal data using Temporal Convolutional Network (TCN) blocks, while the other converts such data into 2D images via Gram Angle Sum Fields (GASF) for spatial features extraction using 2D Convolutional Neural Network (CNN) blocks. Features from both branches are then fused via a multi-head cross-attention mechanism for complementary spatiotemporal information integration. To further improve multi-channel fusion efficiency, fusion weights for each channel are also optimized using an improved snow ablation optimizer. Experiments on public datasets are conducted to validate the proposed method. Results show that the MC-MM-AFDN achieves fault diagnosis accuracy exceeding 99.55% on hydraulic system datasets, maintaining robust performance even with limited sample sizes and under noisy conditions.

Controlled Active Seismic Sources (CASS) have advantages of environmental friendli-ness, low-cost deployment, high waveform repeatability and precise control, and have been successfully applied in geophysical exploration and seismic time-lapse monitoring. Since the large rotational inertia of the eccentric rotation structure in CASS, combined with undetermined disturbances, waveform distortion often occurs during operation, thereby reducing the accuracy of the output seismic waveforms. To address this issue and improve waveform accuracy, a control method based on disturbance observation and feedforward compensation is proposed. First, dynamic model of the CASS system is derived, and a gravity disturbance equation is established to estimate gravity-induced perturbations. Next, to account for undetermined disturbances and modelling errors, a disturbance observer is introduced. Subsequently, a seismic source controller is de-signed, incorporating feedforward compensation for disturbances along with closed-loop feedback to enhance the control accuracy of the seismic waveform. Finally, simula-tions and physical experiments are conducted to assess the system's performance. The experiments, which involve the excitation of linear frequency-modulated signals using the CASS and signal extraction via the Wigner-Hough transform, demonstrate the effec-tiveness of the proposed method.

Bearings are critical components in equipment, and it is important to predict their remaining useful life (RUL) so that early intervention can be applied before their failure. A challenge in predicting the RUL is that the target bearing's working condition and/or failure mode is usually unknown. Although various methods have been proposed to address this challenge, few of them consider two practical issues. (1) When various datasets are available and used to train a single deep learning model, training datasets related to bearings of quite different statuses can affect prediction accuracy. (2) Transfer learning can be used to alleviate issue (1), but this requires data collection from the target bearing and fine-tuning, and the additional time required by this process may delay RUL prediction and intervention. To address these issues in industrial practice, we propose stacking-based RUL (StRUL) prediction for bearings of unknown working conditions and failure modes. StRUL is based on the stacking of transformers with novel designs: a modified amplitude spectrum comparison approach, a similarity-based attention mechanism, and a distribution-based attention mechanism. First, StRUL pre-trains each transformer with a specific dataset so that each transformer can generate an encoding for the input data from the target bearing. Second, it applies a modified amplitude spectrum comparison approach to calculate the similarity value between the input data and each transformer's training dataset. StRUL then uses a similarity-based attention mechanism to prioritize transformer encodings with relatively large similarity values in prediction. Third, it includes an additional transformer trained with all the training datasets and uses a distribution-based attention mechanism to determine how much the additional transformer contributes to the predicted RUL when the distribution of the similarity values is nearly uniform. Case studies performed using the XJTU-SY and PRONOSTIA data, each containing more than 10 datasets, demonstrate that StRUL can efficiently use all the training datasets without additional data collection or fine-tuning to achieve high prediction accuracy and speed useful for practical deployment.

The objectivity of three-dimensional (3D) imaging has made x-ray computed tomography (CT) an indispensable diagnostic method in clinical and industrial applications. Despite all the advances in x-ray CT imaging, high-resolution 3D reconstruction of defects and details in objects with large in-plane dimensions has been a fundamental challenge with the existing commercially available x-ray CT capabilities. An alternative scanning geometry, where an x-ray source and a detector move synchronously, irradiating a specimen at an angle from the rotation axis, has been proposed to address this challenge. To improve the reconstruction of radiography data obtained under limited angular conditions, new designs of prior (or penalty) functions are developed. The new Gaussian prior is aimed at reducing shape distortion artifacts, common in laminographic reconstructions, as well as improving edge contrast and defect detection. Prior function performance is demonstrated on a computational phantom that contains worst-case scenario defects and simulated using limited angular data CT geometry. The Gaussian prior's superior quantitative detection of defects is shown by the line profile plots as compared to the hyperbola prior. The performance of the reconstruction algorithm with the novel prior function is further verified under real CT scan conditions using a novel microfocus x-ray CT scanning device that rotates the x-ray source and moves the detector synchronously while the x-ray source and detector stay on the same opposite side of the test specimen (inclined CT). This device allows the scanning of critical areas up to 3 m wide and potentially unlimited-length specimens with the high resolution required for the identification of manufacturing and structural defects typical of modern composite structural elements.

The calibration of NH₃-sensing instruments has been an ongoing challenge to atmospheric chemists. Here, the potential utility of ammonium nitrate (NH₄NO₃), -sulfate ((NH₄)₂SO₄), -bisulfate (NH₄HSO₄), -carbonate ((NH₄)₂CO₃), and -bicarbonate (NH₄HCO₃) to generate gas streams containing NH₃ or aerosol NH₄⁺ calibrated by measuring the co-emitted acids was investigated. Head space vapors and aerosols were analyzed using a commercial total nitrogen (N_t) instrument modified to quantify N_t mixing ratios in both the gas- and particle phases, scanning mobility particle sizing, thermal-dissociation cavity ring-down spectroscopy, and Fourier transform infrared spectroscopy. The best results were obtained with NH₄HCO₃. Using a simple gas delivery setup consisting of two dilution stages and a line heater, the NH₃ output from NH₄HCO₃ was calibrated by quantifying the stoichiometric co-emission of CO₂ with a relatively inexpensive non-dispersive infrared spectrometer. The new approach constitutes a viable alternative to conventional NH₃ cylinder calibration setups.

Controlled Active Seismic Sources (CASS) have advantages of environmental friendli-ness, low-cost deployment, high waveform repeatability and precise control, and have been successfully applied in geophysical exploration and seismic time-lapse monitoring. Since the large rotational inertia of the eccentric rotation structure in CASS, combined with undetermined disturbances, waveform distortion often occurs during operation, thereby reducing the accuracy of the output seismic waveforms. To address this issue and improve waveform accuracy, a control method based on disturbance observation and feedforward compensation is proposed. First, dynamic model of the CASS system is derived, and a gravity disturbance equation is established to estimate gravity-induced perturbations. Next, to account for undetermined disturbances and modelling errors, a disturbance observer is introduced. Subsequently, a seismic source controller is de-signed, incorporating feedforward compensation for disturbances along with closed-loop feedback to enhance the control accuracy of the seismic waveform. Finally, simula-tions and physical experiments are conducted to assess the system's performance. The experiments, which involve the excitation of linear frequency-modulated signals using the CASS and signal extraction via the Wigner-Hough transform, demonstrate the effec-tiveness of the proposed method.

Photoelasticity is a popular tool to measure the internal stresses of many transparent materials, with numerous industrial, medical and research applications. Importantly for this work, photoelastic techniques can be used to provide quantitative diagnostics for experimental studies of two-dimensional granular flows of circular discs. In this paper, we introduce a novel photoelastic testing apparatus and present the first experimental validation of photoelastic techniques being used to quantify frictional forces acting on cylindrical particles. In this case, the forces acting on the photoelastic particles have both normal and tangential components. Additionally, we perform a quantitative error analysis and obtain bounds on the spatial imaging resolution required to produce accurate photoelastic results from experimental data. Our results verify that photoelasticity can be used reliably to calculate and quantify normal and tangential forces in experimental granular systems, and also give insight into the range of forcing conditions that photoelastic techniques are most accurate. This information is of critical importance to researchers who are designing, running and analysing photoelastic experiments. During our own experiments, we discovered a mathematical error in the widely used open-source Photoelastic Grain Solver (PeGS) code. Much of the current photoelastic research depends on the effectiveness of PeGS, and we found that the error leads to incorrect output from experimental data whenever there are tangential components to any of the inter-particle forces. In this work, we outline the error, quantify its effect and provide a fix to ensure accurate output from the code. The photoelastic community has corroborated and validated our updated version of the software, and we have amended the open-source version of the code to include the modifications outlined in this article.

Reliable Remaining Useful Life (RUL) prediction contributes to fault analysis and preventive maintenance of rotating machinery. Existing artificial intelligence methodologies, however, are challenged by inaccurate feature extraction and uncertainty involved in the RUL prediction process. To this end, this paper proposes a reliable fault prognosis method for rotating machinery using neural networks with multi-scale vibration feature learning and uncertainty quantification. Specifically, the proposed fault prognosis framework starts with constructing a multi-scale semantic embedding module to identify the semantic information in mechanical vibrations. A neural network with local and global feature extraction capabilities is then created to capture information from each scale for RUL prediction. By quantifying the uncertainty of predictions, the framework provides a confidence level for each prediction, and therefore a confidence-based RUL decision fusion method is proposed to achieve the reliable RUL estimation. The feasibility, reliability, and superiority of the framework over state-of-the-art methods are validated by datasets from machinery. Overall, the proposed framework contributes to the safe operation and maintenance of rotating machinery systems.

The implementation of strain-sensing structural materials for structural health monitoring in civil engineering has the potential to enable self-diagnosing buildings, including masonry and concrete structures. However, a major bottleneck of such technologies is the high cost of the measurement hardware. Dealing with smart sensing systems that require expensive laboratory equipment is impractical for the construction industry. Researchers typically use costly laboratory power sources, signal generators, and acquisition systems for measuring structural responses using smart self-sensing materials, due to performance and reliability reasons. In particular, promising sensing composites evaluated in this paper are piezoresistive nanocomposite or microcomposite concretes and bricks that allow measuring mechanical strain using their piezoresistive response. To address the issues above for such kind of materials, the newly proposed Smart construction Materials Electrometer (SME) device offers a practical alternative to off-the-shelf hardware. The SME can generate sinusoidal, triangular, and biphasic signals with $\pm$10 V and program signal frequencies ranging from 1 Hz to 10 Hz. What is more, it can also accurately measure the electrical outputs of smart materials under an applied mechanical load or change in environmental conditions. The SME provides significant advantages over traditional instrumentation: it integrates multiple functionalities for state-of-the-art measurements in self-sensing materials such as smart concretes and smart bricks, allows easy mobility for field testing, is about 50 times cheaper than commercial devices, and offers versatile and accurate measurements. These benefits are confirmed through experimental tests on commercial resistors, comparisons with off-the-shelf high-cost equipment, and field measurements on real structures. The proposed SME is a crucial step toward the practical implementation of strain-sensing structural materials and their widespread use in real-world structural health monitoring applications.

Deep learning (DL) has shown great promise in electrocardiogram (ECG) analysis, revolutionizing cardiovascular medicine by enabling precise and efficient diagnosis. This study explores the integration of expert knowledge into DL models, creating a flexible structure that adapts during training to refine this knowledge and guides feature separation in higher-dimensional space, thereby improving multi-label ECG classification. By leveraging domain expertise on the relationships between specific ECG abnormalities and their corresponding changes across lead dimensions, a lead-wise prior knowledge framework (LPKF) was introduced to enhance the learning efficiency of DL models. The effectiveness of this framework was validated by the higher classification performance of LPKF-enhanced models compared to their original versions. In addition, the LPKF-enhanced Inception model outperforms recent state-of-the-art DL methods, underscoring the benefits of integrating learnable expert knowledge. The study also demonstrated the interpretability of the LPKF-enhanced Inception model using gradient-weighted class activation mapping, revealing its capability to identify crucial diagnostic symptoms from ECG signals that align with clinical criteria.

Correlative imaging, integrating diverse observational methods, has become increasingly vital for comprehensive sample analysis by linking distinct datasets that are accessible only through individual techniques. This study combines cryogenic scanning electron microscopy (cryo-SEM) and cryogenic Raman micro-spectroscopy (cryo-Raman), which together offer extensive insight into sample analysis. Cryo-SEM effectively investigates the surface, structure, and morphology, while cryo-Raman excels in detecting and identifying chemical composition. To integrate these techniques, we developed a novel assembly compatible with commercial cryo-SEM sample holders. The assembly is designed for precise sample positioning and observation at cryogenic temperatures of liquid nitrogen. This setup, controlled by LabVIEW software, includes a Dewar vessel containing liquid nitrogen, a 3D-printed stage for mounting sample holders, and a piezoelectric stage for XY motion, with z axis adjustments connected to the Raman spectrometer table. A plexiglass cover minimizes contamination and preserves sample integrity during analysis. The assembly was validated by first analyzing polystyrene at cryogenic temperatures to ensure its functionality. Subsequent tests on Cupriavidus necator H16—bacteria capable of producing polyhydroxyalkanoates (PHAs), a type of biodegradable polyester—demonstrated the system's efficiency. PHAs, which are of interest as a sustainable alternative to petrochemical plastics, can be produced from industrial waste streams. Our results show that the combination of cryo-SEM and cryo-Raman is suitable for studying these microorganisms and permits a deeper understanding of the properties of polymer granules in microbial cells. This integrated approach represents a significant advance in semi-correlative imaging, providing more efficient and detailed analyses for future research in microbial physiology production and related fields.

Recent advances in volumetric camera calibration techniques have established non-invasive methods for multi-camera setups. Laser-beam methods, particularly those introduced by Hardege et al (2022 Proc. Int. Symp. on the Application of Laser and Imaging Techniques to Fluid Mechanics pp 1–15, 2023 Exp. Fluids64 193) and enhanced by Gunady et al (2024 Meas. Sci. Technol.35 105901), show significant promise. A critical component of these calibration procedures is the precise detection of laser beam intersections in 2D images. Current methodologies, including Hough-Line-Transformation, Template Matching, RANSAC, and 2D Gaussian beam fitting, exhibit limitations in both stability and accuracy, especially when encountering reflections, refractions, and other image artifacts. This paper introduces a deterministic approach for identifying straight light beams and accurately determining their intersections in grayscale 2D images. The proposed method exploits the Gaussian intensity distribution of laser beams, implements adaptive beam width detection, and eliminates the need for binarization, probabilistic sampling, or manual intervention such as template creation or parameter initialization. Extensive evaluation using both synthetic and real-world images demonstrates superior accuracy and robustness compared to existing methods. The presented approach reaches subpixel accuracy in noisy conditions at angles as low as 10^∘ which further increases for higher intersection angles.