Table of contents

Civil structures e.g. bridges, tunnels, and dams are essential to human societies. Currently, these complex engineered structures are challenged by aging issues. It is crucial to monitor the conditions of such structures in realtime to ensure their protection and conduct sufficient maintenance and rehabilitation when they begin to show omens of degradation or damage. Observation of Rayleigh scattering spectra from optical fibers using fiber Rayleigh reflectometry enables distributed sensing of static and dynamic strain in structural health monitoring for civil structures. Its key performance indices are the spatial resolution, the strain dynamic range, the measurement range, and the refresh rate. This article reviews tunable-wavelength optical time-domain reflectometry and coherent optical frequency-domain reflectometry and discusses the performance indices of each method in terms of the performance indices listed above. After analytical derivation, we have found that signal-to-noise ratios of both schemes are the same, which is a valuable discovery. In addition, we enumerate and review recent major industrial developments of both schemes.

Optical fluorescence microscopy provides molecular specificity and high contrast, which are powerful aspects in biomedical researches. Moreover, super-resolution microscopy techniques have broken through the diffraction-limited resolution, which had hindered optical microscopy. Among various techniques, stimulated emission depletion (STED) microscopy quasi-instantaneously reduces the size of the effective focal spot by suppressing the peripheral fluorescence of the excited spot with an additional depletion laser, while also providing optical sectioning. With these advantages, the usage of STED microscopy is increasing in the various field of research. Nevertheless, STED microscopy has been continuously improved to answer more biological questions. This review summarises the recent advancements and new techniques implemented for STED microscopy, including microscopy architectures, multicolour ability, deep-tissue imaging, aberration correction, three-dimensional super-resolution, fast measurement, photostability, and multimodality. It is expected that STED microscopy will further evolve and become an more widely useful tool for life sciences.

High sensitive triaxial accelerometers are used in several space missions to measure the non-gravitational accelerations acting on the spacecraft. Among these, the capacitive accelerometers developed for ESA missions Jupiter Icy moons Explorer (JUICE) and BepiColombo were designed to measure accelerations of the order of 3 · 10⁻⁶ m s⁻² with an accuracy level of 300 ppm in the frequency range (3 · 10⁻⁵–0.1) Hz. Despite the signal to be measured is of the same order of magnitude of the seismic noise on the earth, an accurate on-ground calibration is needed. The facility set-up at INRIM to this purpose is based on a simple principle: the base of the accelerometer is dynamically tilted by an angle α so that the sensor undergoes a component of the gravitational acceleration g proportional to angle α. In practice, several issues have to be addressed by the calibration facility, such as the seismic noise limiting the signal-to-noise ratio, the generation and the accurate measurement of the tilt angle. Furthermore, the calibration was performed taking into account different on-flight conditions such as different operating temperatures and possible deformation of the accelerometers during the launch. The experimental set-up and the calibration procedure are described in the paper. The measurement results and the uncertainty budget show that a relative accuracy of 240 ppm has been achieved.

The GEMIMEG-II project is intended to pave the way for digitalization in metrology. The central element of this digitalization initiative is the digital calibration certificate (DCC). It contains all calibration information in full digital form. This means, that it is machine readable and machine understandable without human interaction. This enables its utilization by being securely machine interpretable and machine actionable in the entire chain of truly digital workflows and information technology (IT) environments in Industry 4.0. Therefore, the DCC is created automatically in the calibration process in a standardized form based on a digital document schema. This systematic schema enables to safely transfer, process, and interpret all data in the DCC automatically in all subsequent IT based processes. This paper reflects the project status of GEMIMEG-II in its final phase and shares some insights on the concepts developed and solutions implemented as the results will be demonstrated in five Realbeds. Furthermore, the concept of quality of sensing and quality of data will be introduced as it is implemented in the GEMIMEG-II project to convey supplementary information on the measurement, environmental and/or surrounding modalities, and data quality. Finally, a brief outlook will be given on next steps and actions planned in the project related to other digitalization initiatives for the fab of the future.

Online wall temperatures were measured with fiber-coupled phosphor thermometry in a full-scale gas turbine combustor. The combustor was operated with natural gas and up to 100 vol% hydrogen at engine-relevant conditions. Two phosphors were tested for this application, namely YAG:Dy and YAG:Tm;Li. Although YAG:Tm;Li seemed to be the most promising phosphor for this application, it turned out to be incompatible with the used setup due to a strong interfering signal generated by the laser in the used fiber setup. A strategy to compensate for interferences from flame emissions during natural gas operation was developed. With this strategy it was possible to obtain single-shot temperature measurements at 15 Hz and a precision of 2–7 K for a 1 s average.

Since the current infrared thermography (IRT) is not effective in detecting deep and invisible cavities in the silicone structural adhesive of hidden frame glass curtain walls (HFGCW), a reference-free IRT with subsurface heating for the deep cavity is proposed. A near-infrared linear laser with high energy density and high transmission is chosen as the subsurface heating source to directly heat the silicone structural adhesive through the glass. Temporal sequence reconstruction and image enhancement based on reference-free calibration are proposed to reduce thermal inhomogenety and thermal noise and ensure comparable results for damage detection under different environments. The effects of traditional surface heating and subsurface heating are compared and analyzed through numerical simulations. And an evaluated feature, which is the maximal temperature difference feature, derived from temperature difference is used to quantitatively analyze the thermal effect caused by different cavities. The subsurface heating simulation results showed that the highest temperature difference between the region with cavity and defect-free region is up to 88% higher than that of traditional surface heating. The experiments revealed that the deep cavities of different lengths, located at 7 mm, 9 mm, and 11 mm below the glass surface, can be successfully detected using subsurface heating and reference-free calibration. A quadratic linear model is proposed to reflect the relationship between the depths and lengths of cavities and the evaluated feature. In conclusion, the proposed method can protect the HFGCW from deep and invisible cavities which can reduce its adhesion and strength.

The accuracy of tesla/volt calibration factors for superconducting quantum interference devices (SQUID) gradiometers is essential to take advantage of the magnetic gradient tensor (MGT), but the accuracy of the published calibration procedures ranges from tenths of percent to a few percent. In this paper, we propose an efficient calibration method for planar SQUID gradiometers based on MGT components. The factors affecting the calibration accuracy and the optimal inclination angle for calibration were investigated by analyzing the response characteristics of planar SQUID gradiometers to the MGT components. By superposition of square coils and gridding of pickup loops, numerical models of the response of planar gradiometers to different types of Maxwell coils were built to simulate theoretical calibration accuracy and calibration errors due to position, attitude and imbalance. We built a calibration setup to achieve high-precision simultaneous calibration of multi-channel planar SQUID gradiometers and investigated the effect of eddy current on the calibration accuracy. The results show that simulation and experimental results are in general agreement by taking into account various error factors. A calibration error of about 0.4% is achieved.

The microsphere profile of probe tips has to be measured and compensated to improve the measurement accuracy of micro-coordinate measuring machines (micro-CMMs) to less than several hundred nanometers. A high-precision radius and sphericity measurement method for the microsphere of the CMM probe tip is proposed in this article. Different circumferences of the tested microsphere are measured, and the true radius of any measuring point on the surface can be obtained after separating the runout errors. Then the sphericity of the tested microsphere can be evaluated by the minimum zone sphere method. A corresponding measuring system is developed based on the analysis of the primary error model and mechanical model, and verification experiments are conducted using a ruby microsphere (A-5000-7801, Renishaw Corporation) as the reference, whose claimed diameter and sphericity are 700 µm and 130 nm, respectively. Six groups of repeated experiments are performed, and 18 000 measurement points on 15 circumferences are recorded in each group of experiments. Results show that the average radius of the tested microsphere is 350.003 µm, and the average sphericity error is 208 nm with a standard deviation of 5.3 nm. Finally, the expanded uncertainty of the measurement results of the developed system is calculated as 209 nm (k = 1.96, normal distribution). The proposed method and system can be used in the precision measurements and compensations for probe tips of micro-CMMs and other parts, such as microspheres used in micro-bearings and micro-manufacturing machines.

This study formulated and clarified the presentation of the noise-related error terms in the power spectral density (PSD) and height–height correlation function (HHCF) through a roughness analysis of self-affine fractals. Assuming the noise as white noise, this study simulated the PSD and HHCF of the noisy roughness profile in a typical situation: line-edge roughness measurements involved in semiconductor lithography technology. It was found that conventional noise-correction techniques can bias the PSD/HHCF from those of the true roughness profiles because they do not account for the residual noise-induced error. The findings obtained from the roughness analysis and simulation techniques are expected to benefit various applications, such as clarifying the quantitative residual error in specific situations, selecting the optimum analysis method (PSD or HHCF), estimating the required number of averaged profiles (measurements) that yields sufficiently small noise errors, and considering strategies for noise reduction.

The effectiveness of model-based diagnosis strongly depends on the model's authenticity and is impacted by various uncertainties. Measurement uncertainty is governed by the probability method, and parameter uncertainty can be handled by the linear fractional transformation, whereas structure uncertainty and errors are rarely considered. In this paper, an improved bond graph (BG) model is proposed, which adopts subsystems to substitute modelling errors, namely parameter uncertainty and structure uncertainty. A multi-dimensional Fibonacci optimization algorithm is developed to identify the parameters of subsystems to obtain the subsystem-based diagnostic hybrid BG (SDHBG) model. Fault diagnosis is realized by comparing the residuals and thresholds derived from the SDHBG. Experiments are conducted to validate the key concepts of the proposed methods. Subsequently, the results suggest its effectiveness.

During long-term operation of analogue circuits, fault diagnosis is important for preventing the occurrence of hazards. However, noise often accompanies sampled signals and makes the task of fault diagnosis more difficult. Therefore, developing a robust feature extraction technique is an indispensable part of fault diagnosis. The locally linear embedding (LLE) algorithm has recently emerged as a promising technique for dimensional reduction and feature extraction because it preserves linear neighborhoods, and it is quite effective when there is a locally linear dependent structure embedded in fault data. However, LLE is sensitive to noise. Therefore, the maximum correntropy criterion is adopted to resist non-Gaussian noise by seeking the optimal weight coefficient, and a half-quadratic optimization procedure is introduced to address the objective function. Moreover, softmax regression is applied to locate faults. Finally, two typical analogue circuit systems are used to demonstrate the robustness of the modified algorithm to non-Gaussian noise. The experimental results show that the robust LLE algorithm can outperform LLE in the extraction of fault features when there is non-Gaussian noise in the fault signals, and the proposed fault diagnosis method has a better effect in locating faults compared with other feature extraction methods.

Binocular vision can get the three-dimensional information of the objects according to two-dimensional images. However, when the background texture information of the workpiece to be measured is weak, or the depth information cannot be recognized due to the change of viewing angle, it will lead to poor three-dimensional measurement accuracy. To address this problem, the paper proposes a multi-view workpiece 3D measurement method based on binocular vision. First, an experimental bench with a Chessboard is designed. The corner point reconstruction is realized by extracting the corner point of the calibration plate. The checkerboard plane is fitted by the least squares method to obtain the checkerboard plane mathematical model. Then, the vertices of the workpiece are extracted at the subpixel level, and a minimum distance sparse vertex stereo matching algorithm (EDMS) based on Euclidean distance metric is proposed to achieve accurate and fast corner matching. Finally, the three-dimensional dimensions of the workpiece are calculated. Through experiments on multiple angles of the two workpieces, the results show that the average absolute error measured by the method at different angles is 0.33 mm, the total relative error is 0.90%, and the variance is less than that 0.01 mm², realizing the more accurate measurement of multi-view three-dimensional dimensions of small workpieces. This paper provides a new binocular vision handheld mobile 3D measurement equipment method.

Aiming at the disadvantages of the existing conveyor belt flow monitoring, this research builds a conveyor belt optical dynamic flow-meter based on field programmable gate array (FPGA) by using FPGA image processing technology and the method of measuring material volume to measure the volume of conveyor belt material, which uses the principle of laser triangulation. The flow-meter takes FPGA as the core and drives two complementary metal oxide semiconductor (CMOS) cameras to collect light strips on material surface at the same time. Then, filter, binarization, center extraction and image stitching are performed on the collected light strips. Finally, the material volume is calculated by linear interpolation and Riemann sum. The image acquisition and processing of the flow-meter and the final volume calculation are implemented in the form of hardware in FPGA, the calibration and result display of system parameters is realized in the form of software in advanced RISC machine (ARM). It can be shown that the relative error of the volume measured by the flow-meter is less than 0.05 through experiment, and the material volume flow rate of the conveyor belt can be measured and displayed accurately in real-time, which makes up for the deficiency of the existing flowmeter. This system is of great significance for the informatization transformation of the conveyor belt and the realization of digital production and intelligent manufacturing.

To ensure the nanometer-level running accuracy of the ultra-precision spindle, the radial error motion should be precisely measured. However, the asynchronous errors induced by the measurement instrument will mix in the measured data and significantly deteriorate the measurement precision. In this study, a novel adaptive digital filter method (ADFM) is proposed to suppress the asynchronous errors and lower requirements for the precision of measurement instruments based on the Donaldson reversal method. Specifically, the Gaussian mixture model is first used to investigate the characteristics of asynchronous errors and obtain appropriate initial filter parameters. The least mean square algorithm is utilized to adaptively optimize the filter parameters. Monte Carlo simulation results indicate that ADFM can suppress asynchronous errors by more than 90% compared with the traditional methods. Experimental results show that the standard deviation uncertainty is reduced by more than 60%. Besides, the measurement results of two sensors with different precision verify the feasibility of ADFM.

In a laser tracking control system, a motor drives the load to achieve continuous forward and reverse rotation speed control, and the magnitude of the load changes periodically during the motor rotation to realize tracking. Based on these characteristics, we propose an integral-separated improved variable universe fuzzy proportional-integral (PI) control method with torque feedforward compensation method in this study. The proposed method achieves the control performances of fast response, high steady-state accuracy, and strong anti-interference. The conventional fuzzy PI control method achieves good performance only in the unidirectional acceleration of a motor; therefore, we have improved its fuzzy rules, and designed a novel universe contraction–expansion factor. Additionally, a load torque observer is introduced to observe and compensate for the external load and to further improve the anti-interference performance of the tracking control system. The experimental results demonstrate that the integral-separated improved variable-universe fuzzy PI control method achieves better performance in the forward and reverse speed regulation processes of the motor. The overshoot is 92% less than that of the conventional PI control when the target speed is set to ±500 rpm. The Luenberger observer is introduced to observe and compensate for the load torque. After the stabilizing the speed and applying an external load of 0.2 N m to the motor, the maximum speed reduction is observed to be 59% less than that obtained using the conventional PI control. The improved PI control method satisfies the speed loop control requirements of the laser tracking control system and can improve the steady-state tracking accuracy and robustness of the system.

To address the issues of low classification pickup efficiency and slow response time of parallel robots positioning (RP) methods based on machine vision, this paper proposes a deep learning hybrid method RP-YOLOX-Deeplabv3+ (YOLOX-DL method for RP) for parallel robots to achieve accurate picking. Firstly, the RP-YOLOX lightweight network is used to complete target recognition classification and rough positioning. A new feature enhanced network called Depth Slim Path Aggregation Network is proposed to optimize the original up-and-down sampling structure, and the computational efficiency is improved through the attention mechanism and deep convolution. The loss function in network evaluation is enhanced, and an emphasizing the target Binary CrossEntropy loss function is proposed for the objective loss strategy. Secondly, the DL network is used and the pooling structure is improved to obtain rich multi-scale information by using different types of Atrous convolution. Then, the extracted semantic segmentation center coordinates are finely positioned, and by using a hybrid positioning strategy, the RP-YOLOX and DL network modules are combined to obtain the best positioning coordinates of the target. Finally, a hand-eye calibration operation was performed to convert the robot, the camera, and the conveyor belt to eye-to-hand coordinate computation. The experimental results indicate that the hybrid method achieves a pick-up rate of 92.56% and a response time of 2.357 s, showing better performance than the traditional algorithms Faster-RCNN, YOLOv3, and YOLOv5. Meanwhile, the identification efficiency is increased by 2.41% compared to the YOLOX algorithm. These results verify the efficiency and robust adaptability of the mixed method. This study has a certain reference value for applying the deep learning method to robots positioning pick-up.

In engineering practice, device failure samples are limited in the case of unexpected catastrophic faults, thereby limiting the application of deep learning in fault diagnosis. In this study, we propose a prior knowledge-based residual shrinkage prototype network to resolve the fault diagnosis challenges under limited labeled samples. First, our method combines general supervised learning and metric meta-learning to extract prior knowledge from the labeled source data by utilizing a denoised residual shrinkage network. Further, the knowledge extracted from the supervised learning is used for prototype metric training to achieve a better feature representation for the fault diagnosis. Finally, our approach outperforms a series of baseline methods in the few-shot cross-domain diagnostic task on the gearbox and bearing datasets. A diagnosis accuracy of more than 95% has been achieved in a variety of working conditions for diagnostic tasks, which is far higher than the existing basic method.

Data-driven machine learning (ML) for rolling bearing remaining useful life (RUL) prediction is a promising method in condition-based maintenance. However, due to the uncertainty of optimal hyperparameter tuning of the ML model, it is very difficult for a data-driven method to accurately predict the RUL of rolling bearings. Aiming to address this problem, this paper proposes a hybrid model-based on continuous wavelet transform (CWT), convolutional neural network (CNN), Bayesian network and long short-term memory network for estimating the remaining usage of rolling bearings lifetime. Firstly, the one-dimensional vibration signal of a bearing is divided into six segments and then it is converted into the corresponding two-dimensional time-frequency feature images via CWT. Secondly, the two-dimensional images are input into the two-dimensional CNN for deep feature extraction in order to obtain a series of one-dimensional feature vectors. Finally, it is input into a Bayesian-optimized long short-term memory model to obtain a prediction of the RUL of the bearing. The effectiveness of the proposed method is verified using bearing data. The verification results show that the proposed method has better prediction accuracy than the other two compared prediction methods, which indicates that the proposed method can effectively extract the bearing fault features and accurately predict the RUL of rolling bearings.

This paper aims to employ high-rate global navigation satellite system real-time kinematic (GNSS-RTK) and triaxial accelerometer sensors to evaluate the modal frequencies of a damaged long-span bridge based on the proposed modified variational mode decomposition algorithm. Firstly, in the modified VMD, the relative root mean square error was presented to optimize the decomposition parameter K, the correlation criterion was introduced to select the suitable intrinsic mode function components. The above procedure is called the RCVMD algorithm. Subsequently, the performance comparison evaluation experiments proved that the RCVMD algorithm was more advantageous in the decomposition and reconstruction of signals, had a better noise reduction effect in the signal with different noise levels, and was suitable to process the actual measurement response. Finally, in the field experiment, the modal frequencies of the damaged bridge were correctly detected from the GNSS-RTK displacement and acceleration responses according to the RCVMD algorithm, and the accelerometer picked up the high-order structural frequencies, demonstrating the reliability of the proposed method. The frequency characteristics of the damaged bridge are different from the normal state. The first and third-order frequencies increase, which is especially obvious in the first-order frequency, yet the second-order frequency decreases. The measured results have practical applications and can provide data support for the maintenance of the damaged bridge.

In response to the problems of biased estimation of instantaneous frequency (If) and poor noise immunity in current time–frequency (Tf) analysis methods, the adaptive scale chirplet transform (ASCT) is proposed in this paper. The core idea of the proposed algorithm is to use a frequency-dependent quadratic polynomial kernel function to approximate the IF of the signal and to use the time-varying window length to overcome the frequency resolution problem due to the change in signal modulation. This method can dynamically select suitable parameters and overcome the disadvantage of unfocused energy of TF distribution. The experimental results show that the ASCT algorithm has high TF aggregation and can suppress noise interference well. In practical signal processing, the advantage of the ASCT algorithm is that it can accurately depict the characteristic frequency of the signal and detect the fault in the bearing signal. Both simulation and experimental results prove the strong realistic relevance of this algorithm.

To prepare for the lunar water exploration in the Chinese Lunar Exploration Project IV, a new apparatus for studying characteristics of the water vapor conductance and water detection at low temperatures was built based on the flowmeter method, and the performance of the miniature time-of-flight mass spectrometer (TOF-MS) for water detection in the range −20 to 20 °C was measured. The through-put within the range of 3 × 10⁻⁹–9 × 10⁻⁶ Pa m³ s⁻¹ was provided at temperatures in the range −60 to 20 °C. The conductance of the orifice with a diameter of 21 µms for three gases (N₂, Ar, and H₂O) was measured at low temperatures. The diameter change of the metal orifice caused by the cold contraction in the molecular flow state is the main factor affecting the conductance, and the viscosity characteristic of gas in the viscous flow state is the main factor affecting the conductance. Therefore, the conductance of the orifice increases with the decreasing temperature when the through-put is high. In addition, the water vapor can be stably supplied by the orifice under low temperatures, and the conductance of water vapor through the orifice is measured through the on-line method and the time-ratio method. The adsorption rate and amount of water vapor on the metal surface increase with the decreasing temperature, hence the concentration of water molecules in the test dome significantly decreases as the temperature decreases. Finally, the through-put of water vapor of 8.75 × 10⁻⁸ Pa m³ s⁻¹ was detected by the miniature TOF-MC at −20 °C on the apparatus.

During the assembly process of deep groove ball bearings, due to defective parts and unqualified assembly process, various indentations and scratches on the dust cover will often result in reducing the service life and reliability of the bearing. Therefore, the online monitoring of the assembly quality of the dust cover ensures the necessary detection process of the bearing surface quality. This paper proposed a bearing dust cover defect detection method based on machine vision and multi-feature fusion algorithm, which can effectively detect bearings with dust cover defects. The algorithm first performs Laplace transform and Sobel operator image enhancement on the collected bearing images. Extract and fuse multi-source fault feature with the scale-invariant feature transform (SIFT), bag-of-visual-words (BoVW) and GLCM-Hu methods. Machine learning and deep learning models were constructed, and the performance of each model was compared through feature visualization and misclassified analysis. The results show that the extracted multi-source features are more representative and robust. The SIFT-BoVW-GS-SVM model achieved the best detection results in detecting bearing dust cover defects with an accuracy of 91.11%. The processing and program detection time for each bearing image is about 0.019 s. The accuracy and speed of detection and judgment meet the needs of online defect detection of bearing dust cover.

In order to detect the radial is characterized and the isolated points that deviate from the target point cloud are removed using statistical section dimensions of ring parts in the forging process, we present a new method that processes laser scanning data of the outer radial section profile of ring forgings. Based on regression analysis and point cloud filtering, a multiple regression filtering algorithm is proposed to process the point cloud data for the measurement of the outer diameter of ring forgings. First, the laser scanned point cloud data filtering methods. The outlier clusters are then removed using regression analysis and processed by Gaussian regression filtering algorithms to obtain point cloud data with good circularity characteristics. Next, the Least Squares principle is used to fit a circle to the processed data to obtain information on the external diameter of the ring forgings to be measured. Finally, the feasibility of the method is verified by measuring the outer dimensions of the elliptical ring forgings in the experiment and performing error analysis.

High-G accelerometers are critical for measuring high shock signals and must be calibrated to improve measurement accuracy. A laser Doppler velocimeter (LDV) is required to calibrate a high-G accelerometer to provide a high-precision reference velocity. The LDV signal must be demodulated to obtain the velocity. However, the phase method is susceptible to noise interference, while the conventional periodic distribution method is challenging to demodulate and severely affected by signal oscillations. We propose a novel periodic distribution method based on enhanced complementary ensemble empirical mode decomposition (CEEMD) and threshold correction to demodulate the LDV signal. First, the LDV signal is processed with CEEMD to obtain multiple intrinsic mode functions (IMFs) and the residual. Next, each IMF is partially zeroed to obtain the noise-reduced LDV signal. Then, the over-threshold peak of the noise-reduced LDV signal is calculated. Finally, the demodulated velocity of the LDV signal is obtained by correcting the noise-reduced LDV signal according to the over-threshold peak point and calculating all the zero points. Simulation and experimental results show that the proposed method outperforms the phase method based on enhanced CEEMD and the periodic distribution method based on enhanced CEEMD and can significantly reduce noise interference. The results show that the proposed method can accurately demodulate the LDV signal to obtain a highly accurate reference velocity, improving the reliability of accelerometer shock testing.

Due to the random surface texture and weak features of the grinding processing, most current visual measurement of grinding surface roughness is predicted by designing feature indicators, but its imaging environment is more demanding and the indicator design is more artificial. While deep learning can achieve feature self-extraction, the existing deep learning-based feature extraction uses a single convolutional model to extract features, which tends to make the extracted features noise, with low resolution and poor perception of details. To address the above problems, this paper proposes a feature fusion-based method for measuring grinding surface roughness. The method adopts different feature extraction ways and fuses the extracted features to obtain more features while also improving the generalization ability of the model, and verifies the adaptability of the model in different lighting environments. The experimental results show that deep learning self-extracted features based on feature fusion can effectively solve the problem of weak feature information on grinding surface roughness that is difficult to identify, and the model has high detection accuracy across different lighting environments, thus laying the foundation for the automated visual online measurement of grinding surface roughness.

Domain adaptation has shown good performance in fault diagnosis of bearing transferable working conditions. However, most of the existing cross-domain diagnosis models are carried out in linear space, which are subject to strong linear constraints and ignore the manifold correlation information between samples, which greatly limits the learning ability of the diagnosis model. Therefore, we propose a manifold embedding adaptive graph label propagation fault diagnosis method for rolling bearings under variable working conditions. The purpose is to reduce the cross-condition domain offset and mine the neglected nonlinear manifold information in the linear space to improve the adaptive ability of the diagnosis model. First, a similarity graph is constructed on the source domain of the known fault type and the target domain of the unknown information, and the cross-domain propagation of the label information is performed on the graph. Then, in this process, the clustering domain adaptation is performed simultaneously, the accuracy of label propagation is improved by reducing the cross-domain offset, and the graph is updated according to the new sample distribution. Finally, according to the known graph and target domain labels, the regression residuals of linear labels are introduced to relax the strong linear constraints in the original space, so that the linear label space approximates the manifold label space and obtains more accurate fault identification results. Experiments show that the new method, as an unsupervised domain adaptive method in transfer learning, has better fault recognition ability than similar transfer learning methods on Paderborn university datasets and own test rig datasets, and the convergence performance of the algorithm is excellent. The highest fault recognition accuracy can reach 100%.

In metrology and industrial design, the evaluation of measurement uncertainty and error is crucial to the measurement process. The Guide to the Expression of Uncertainty in Measurement and its supplementary documents have established a unified framework and standard for evaluating measurement uncertainty. However, a reasonable method for evaluating dynamic measurement uncertainty has not yet been proposed. By analyzing the dynamic measurement system, and using the long short-term memory time neural network to model the nonlinear dynamics represented by a piezoelectric drive platform, this paper evaluates the system's dynamic measurement uncertainty through deep integration methods. Bayesian theory is used to propagate probability densities, and experimental results demonstrate the effectiveness of this method for assessing dynamic measurement uncertainty.

The measurement of NOx emissions in the selective catalytic reduction (SCR) system of boilers has problems with poor real-time performance and abnormal measurements during purging. It is necessary to accurately estimate NOx emissions. For this reason, the NOx emissions prediction method of boiler based on mutual information feature reconstruction and optimization of extreme learning machine (ELM) is proposed: firstly, delay estimation and data space reconstruction of input features are performed based on mutual information; Then the conditional mutual information based on greedy selection strategy is adopted to rank and choose the input features; Finally, the hybrid quantum sparrow search algorithm (QSSA) was proposed by combining Lévy flight strategy and quantum strategy in the sparrow search algorithm, and QSSA is used to optimize the weights and biases of the ELM. Taking the operation data of the SCR system of a 1000 MW thermal power unit as an example for verification. The results show that the proposed method can effectively improve the accuracy and generalization ability of the ELM, and provide a new method for NOx emissions estimation of boilers.

In response to the problem that the input data and combination method of existing polar motion (PM) prediction methods are relatively single, which leads to the limited satisfaction of PM prediction accuracy by major satellite navigation orbiting systems and deep space exploration projects. This study borrows the idea of differential method and proposes to push back the forecast after selecting several samples within the PM Y, X and Y–X series by sliding average. In the constructed new series, the high-frequency terms are effectively attenuated. Then, the pushing back forecasts are combined in pairs with those of the traditional method. After least-squares extrapolation and autoregressive (LS + AR) modeling, the optimal combination was found. Among them, the prediction of PMX is obtained by subtracting the forecast of PMY of traditional method and the prediction of PM(Y–X) of the sliding average method, the forecast of PMY is obtained by adding the forecast of PMX of the sliding average method and the forecast of PM(Y–X) of the traditional method. The results of the 418-week hindcast experiment from 2012 to 2021 show that the proposed method has a greater improvement than the traditional method, and the corresponding 1–365-day mean absolute error (MAE) are improved by 31.46% and 21.11%, respectively, on average. It has certain advantages over the IERS Bulletin-A in the medium-long term, and the 150-day lead time predictions, the MAE of PMX and PMY were 14.678 and 17.232 mas, respectively, which were less than the 17.833 and 20.769 mas predicted by IERS Bulletin A. This not only verifies that the stability and ability of the proposed method have some competitive ability, but also provides new ideas for other time-series forecasting studies.

In this paper, a new kinematic parameter error calibration method based on the Levenberg–Marquard and the beetle antennae search algorithm is proposed to enhance the positioning accuracy of the industrial robot. Firstly, the Modified Denavit–Hartenberg model is chosen to establish the kinematic model for the industrial robot. Secondly, the kinematic parameter errors are calibrated by Levenberg–Marquard algorithm and then obtain the kinematic parameter errors of the industrial robot. Thirdly, these kinematic parameter errors are taken as the center values for the initial individual of the beetle antennae search algorithm. The kinematic parameter errors are accurately calibrated using the beetle antennae search algorithm to obtain the best value for the kinematic parameter errors. Finally, experimental verification results demonstrate that the positioning error of the industrial robot is decreased from 0.7332 mm to 0.1392 mm by using the proposed Levenberg–Marquard and beetle antennae search algorithm. The results also demonstrate that the proposed Levenberg–Marquard and beetle antennae search algorithm calibrate the kinematic parameter errors of the industrial robot effectively and enhance the positioning accuracy of the industrial robot significantly.

An apodized fiber Bragg grating (FBG) is designed to investigate the impacts of side lobe elimination in quasi-distributed sensing for the estimation of measurands (like temperature and strain) to assess the condition of civil structures, such as bridges. The adjacent FBG spectrums may overlap with each other because of the impacts of temperature and strain due to the presence of a high range of side lobes in a quasi-distributed sensing network. Therefore, elimination of side lobes is necessary, by introducing a method of apodization. The sensitivity of the designed apodized FBG is estimated by analyzing the variations in the Bragg wavelength due to the impacts of temperature and strain. The changes in Bragg wavelength due to the measurands can affect the grating period and the grating index of the FBG. The period of the grating and the grating index of the FBG are simultaneously varied by temperature and strain. To measure the physical parameters effectively, it is essential to distinguish whether the changes in the Bragg wavelength are owing to the impacts of temperature or to the impacts of strain. The effect of cross-sensitivity between the temperature and the strain is a key problem in any FBG-based sensing application as both the measurands can affect the Bragg wavelength. In this work, machine learning methods (the support vector machine, K-nearest neighbors, logistic regression, naïve Bayes, decision tree, and ensemble models) are introduced to differentiate between the effects of temperature and strain on a single Bragg wavelength shift measurement. An artificial neural network is used for the predictive analysis of physical parameters, to identify any measurements of potential concern. It has been noted that the performance of the proposed ensemble model is higher compared to other models for the classification of temperature and strain.

This paper studied the problem of ice and snow pollutants identification and freezing temperature detection under winter runway operation, and designed an active detection system for runway ice and snow pollutants and freezing temperature. The temperature change trend during the liquid freezing process was researched, and the freezing temperature detection model based on sequence segmented linear fitting and inflection point identification was proposed in combination with active cooling technology. The differences in dielectric properties of runway snow and ice cover are studied, and a multi-frequency detection-based forked-finger planar capacitive sensor was identified for snow and ice pollutants detection considering practical application scenarios. A finite element simulation model of the forked-finger planar capacitive sensor was established, to optimize the structural parameters of the sensor and verify the feasibility of the capacitor sensor. Finally, the detection device and control system were designed and fabricated, and built an experimental platform for system test. The field experimental results showed that the system has good reliability and stability, and the error of freezing temperature detection model is less than 0.3 °C. At the same time, it can identify three types of pollutants, including water, ice and ice–water mixture, with an accuracy of 89%.

The fault diagnostic approach based on sparse optimization has been receiving considerable attention. Which shows superior robustness and noise adaptability. However, in the real working environment, the collected bearing signals are often accompanied by random impact interference owing to alterations in working circumstances and load mutations. In this paper, the nonlinear sparse metric is used to reduce the interference of random shock excitation (FNCSF). The nonlinear activation has different activation coefficients for different amplitudes, which can change the sparsity distribution of the raw data. Firstly, the influence of different nonlinear activation functions on the diagnostic performance is studied. Then, in order to solve the scale inconsistency caused by nonlinear activation, the nonlinear function is created to the generalized form to further improve the noise adaptability of feature extraction. The second-order Gaussian fitting is used to improve the performance of the learning filter. Simulation and experimental results verify the performance of the proposed method. The results demonstrate that the suggested technique can significantly reduce the interference of random impact components.

Deep learning-based bearing fault diagnosis methods have been developed to learn fault knowledge from massive data. Owing to the deficiency of fault samples and the variability of working conditions, these deep learning-based methods are limited in industrial applications. To address this problem, this study proposes a prior knowledge-based self-supervised learning (PKSSL) method for bearing fault diagnosis. In the PKSSL method, prior diagnostic knowledge is extracted by meta-learning from a few samples. Prior diagnostic knowledge is then utilized to guide the self-supervised learning (SSL) process to reduce reliance on training data. Furthermore, a graph convolutional network is introduced to fuse the information obtained by meta-learning and SSL, which makes the model fully utilize the learned information and improves the accuracy of the fault diagnosis. The effectiveness of the proposed method was validated using two datasets. The results demonstrate that compared to other existing approaches, the proposed method exhibits a strong generalization ability to transfer diagnostic knowledge from artificial damage data to real damage data under varied operating conditions.

In this paper, a hybrid convolutional neural network (CNN)-bidirectional gated recurrent unit (BiGRU) model is integrated with the bootstrap method to endow the deep learning (DL) based prognostic method with the quantification capability of the prognostic intervals. The proposed hybrid method contains three parts: (I) The complete ensemble empirical mode decomposition with adaptive noise and principal component analysis and the CNN-BiGRU are utilized to automatically construct the health indicator (HI). (II) 3σ criterion is employed to detect the first predicting time based on the HIs of rolling bearings. (III) The bootstrap method is imposed to endow the proposed DL method with the quantification capability of the prognostic intervals. The experimental validation is carried out on the XJTU-SY bearing dataset and the proposed method outperforms the other four methods in the majority of cases. In addition, the proposed method not only comprehensively considers the fault prognosis error caused by model parameters and noise, but also considers the prediction error caused by different combinations of features on the model.

Accurately monitoring the blood levels of testosterone (TST) in a sensitive, simple, and rapid manner is of paramount importance for the diagnosis and treatment of various medical conditions, as well as for controlling pharmaceutical quality and facilitating doping detection. TST, the primary male sex hormone, plays a crucial role in facilitating human physical performance, protein synthesis, and the development of muscle mass. Consequently, TST and its analogues are frequently abused by athletes as performance-enhancing steroid in order to increase muscle mass and enhance their performance. The use of such steroids is strictly prohibited to guarantee fair play. In this study, we employed a pencil graphite electrode that was electrochemically modified with CuO nanoparticles (CuONPs) for determination of TST. The electrode response was significantly enhanced by approximately fourfold compared to the unmodified pencil graphite electrode (PGE) when electrodeposition CuONPs onto PGE surface was performed at a potential of −0.6 V for 200 s. The success of modification was confirmed through morphological analysis using scanning electron microscopy and energy-dispersive x-ray spectroscopy. Using square wave adsorptive stripping voltammetry analysis in Britton–Robinson buffer at pH 6.0, we demonstrated that the proposed sensor exhibited sensitivity to detect TST within a linear range of 5–200 nM. The detection limit of sensor was calculated 4.6 nM (1.32 ng ml⁻¹). The sensor platform developed for the accurate, sensitive, and specific determination of TST holds tremendous potential for the development of point-of-care devices and their integration into lab-on-a-chip research.

A novel hybrid interferometer sensor composed of a tapered seven-core fiber (TSCF) and a polydimethylsiloxane (PDMS) cap at the end face of a TSCF is proposed for simultaneous measurement of temperature and gas pressure. TSCF forms a Michelson interferometer (MI), and the PDMS cap on the end surface of TSCF forms a Fabry–Pérot interferometer (FPI). The sensing head consisted of a cascade of MI and FPI. When the external temperature or gas pressure changes, owing to the thermal effect or elastic deformation of PDMS, the interference spectrum of the FPI shifts significantly, so the FPI is very sensitive to temperature and gas pressure. MI, which is made of quartz optical fiber, is sensitive only to temperature and is not to gas pressure. The experimental results show that FPI has a temperature sensitivity of −0.22 nm °C⁻¹ in the temperature range of 40 °C–80 °C, and a gas pressure sensitivity of −2.27 nm MPa⁻¹ in the gas pressure range of 0–0.3 MPa. MI has a temperature sensitivity of 0.05 nm °C⁻¹ in the temperature range of 40 °C–80 °C, and a gas pressure sensitivity of zero in the gas pressure range of 0–0.3 MPa. Using the temperature and gas pressure sensitivities of FPI and MI to construct a measurement matrix, it is possible to simultaneously measure temperature and gas pressure, eliminating their cross-sensitivity. This sensor has the comprehensive advantages of compact structure, small size, easy manufacturing, low cost, high reliability, and high sensitivity, and is expected to be applied in industrial practice.

Because the working conditions of the rotate vector (RV) reducer are often reciprocating at variable speeds and the working environment is relatively complex, so the vibration signals collected have non-stationary vibration and complex environmental noise. A fault diagnosis method for RV reducer (WDT-IF-SS-VMD) is proposed, which combines instantaneous frequency (IF) trend graph based on current signal and the parameter adaptive variational mode decomposition (VMD) algorithm. Firstly, the current signal and vibration signal are collected synchronously, and the steady-state phase of vibration signal is intercepted according to the IF trend diagram obtained by wavelet decomposition transform (WDT) of the current signal. Secondly, the intercepted vibration signals are transformed into scale space, and the parameters of VMD are selected adaptively based on fuzzy C-means clustering. Then, the maximum kurtosis criterion is used to complete the extraction of sensitive components. Finally, the envelope analysis is carried out to complete the fault diagnosis. The measured signal analysis results show that this method can effectively separate and extract fault features of the RV reducer.

This work presents a study of a deflection laser sensor using a pump light source with different polarization states and shows that controlling the polarization state of the pump source can achieve better control in the tuning of an erbium-doped fiber laser. Laser tuning uses a selective wavelength filter manufactured using a thin core fiber section between two single-mode fibers, while the deflection is applied using an angular mechanism. In addition, the sensor was analyzed according to the wavelength shift of the laser emission as a function of the angular micrometric deflection, and a sensitivity of −33.01 pm µrad⁻¹ was obtained in a dynamic range from 0 to 89.3 µrad with an adjustment parameter R² = 0.993 61. We achieved dual-wavelength tuning with gradual shifting and single-wavelength tuning from 1531.5 nm to 1547.7 nm. This sensor exhibits potential applications in the bionic and robotic detection fields owing to its high sensitivity, good linearity, simple fabrication, and low cost.

This work explores the use of microelectromechanical system (MEMS) sensors in electronic nose systems for the measurement of prominent industrial pollutants and proposes a methodology for the correction of drift induced in sensors during prolonged use for measurement of corrosive compounds. MEMS sensors have the advantages of low recovery and response time, lower power consumption and improved sensitivity. An electronic nose system is developed to quantify the four major pollutants emitted from various process industries, i.e. hydrogen sulfide (H₂S), carbon monoxide (CO), ammonia (NH₃) and ethanol (C₂H₅OH). All four pollutants represent major environmental and human health hazards and rapid quantification is vital. A drift correction algorithm was developed using experimental design and robust regression to correct the drift induced in the sensors. The experiments consisted of eight experimental sets with mixtures of H₂S, CO, NH₃ and C₂H₅OH measured on the same sensor array at the start and after a gap of 1 year. The methodology consisted of mapping the response of sensors after 1 year compared with that observed at the very start (day 1) of the experiment. The results showed successful implementation of the methodology, with the root mean square error values being significantly reduced at 0.235, 0.354, 0.145 and 0.651, respectively, for the four studied odorants (H₂S, NH₃, CO, and C₂H₅OH).

Rolling bearings play a crucial role as components in mechanical equipment. Malfunctioning rolling bearings can disrupt the normal operation of the equipment and pose safety hazards. Traditional deep learning-based methods for diagnosing faults in rolling bearings present several issues, such as insufficient feature information of fault samples, high model complexity and low accuracy. To overcome these challenges, this paper introduces an intelligent approach for rolling bearing fault diagnosis using intrinsic feature extraction and convolutional block attention module (CBAM)-enhanced InceptionNet. In our researches, variational mode decomposition (VMD) is adopted to decompose the original signal into multiple band-limited intrinsic mode functions (BLIMFs). In the decomposition process of VMD, the number of decomposition layers k is determined by center frequency method and the optimal BLIMF is chosen based on minimum envelope entropy. Subsequently, the continuous wavelet transform is employed to transform the optimal BLIMFs into time-frequency images. Finally, the obtained time-frequency images are fed into the proposed CBAM-enhanced InceptionNet for fault state diagnosis. Experiments on two different datasets prove that the method has stable and reliable accuracy. Comparative experiments have demonstrated that this method can reduce network model parameters and improve diagnosis efficiency while achieving high accuracy.

When a large amount of full life-cycle data are available, similarity-based methods are the preferred method for remaining useful life (RUL) prediction due to their reliability and accuracy. Traditional similarity-based RUL prediction methods use a single model and single-scale degradation features, which are incapable of fully capturing the degradation behavior of the system. Additionally, the similarity of spatial orientation is neglected in the similarity-matching process. To fill these research gaps, a novel method is developed based on multimodal degradation features and adjusted cosine similarity (ACS) to tackle complex-system RUL prediction in this paper. Complete ensemble empirical mode decomposition with adaptive noise is employed to decouple global degradation and random fluctuations in run-to-failure sensor data. Slow feature analysis is utilized to obtain local degradation features, and residual terms are used as global degradation features. Then, multimodal degradation features are transformed into one-dimensional health degradation indicators by bidirectional gated recurrent unit autoencoder. An ACS is developed to estimate the matching similarity between the test degradation curve and the training degradation curve. The proposed scheme captures the time-varying multimodal degradation behavior and provides libraries of health curves with multiple degradation patterns. The designed scheme is evaluated on the C-MAPSS dataset and the results illustrate the competitiveness and effectiveness of the proposed method.

The BeiDou-3 global navigation satellite system was operated successfully in July 2020. It broadcasts precise point positioning (PPP)-B2b signals and provides real-time PPP services free of charge, which greatly meets the navigation needs of real-time users, especially warship navigation. This paper first introduces the PPP-B2b real-time positioning model, and then expounds the two-velocity measurement algorithm of Doppler and carrier phase time difference. Next, it combs the extended Kalman filter (EKF) and factor graph optimization (FGO) integrated navigation algorithm in detail, and finally comprehensively evaluates the real-time integrated navigation performance of PPP-B2b/strapdown inertial navigation system (SINS) through the measured data at sea. The experimental results show that the real-time dynamic positioning accuracy of PPP-B2b can reach the decimeter level, and the positioning accuracy is not significantly improved when integrated with SINS. Furthermore, dynamic alignment experiments results show after adding time-differenced carrier phase and Doppler velocimetry information, the convergence accuracy of yaw misalignment angle has been greatly improved. The accuracy of the EKF algorithm is 16.7% higher than that of the FGO algorithm after the convergence of the yaw angle error.

Missing data, caused by many factors such as equipment short circuits or data cleaning, affect the accuracy of condition monitoring for rotating machinery. To improve the precision of missing data recovery, a compressed sensing-based vibration data repair method is developed. First, based on the Gaussian random matrix, an improved optimized measurement matrix (OMM) is proposed to accurately sample data. Then, a sparse representation of the vibration signal, through a discrete cosine transform, is utilized to make the noisy vibration signal sparse. Finally, the orthogonal matching pursuit algorithm is employed to reconstruct the missing signal. The effectiveness of the proposed method is verified by analyzing constant and variable speed time series of rolling bearings. Compared with other data repair methods, it is shown that the OMM has a higher repair precision at different loss rates.

It is well known that complex mechanical equipment has many differnt the failure modes, and monitoring data for fault conditions are scarce. Therefore, research on fault detection for reciprocating plunger pumps with fault-free data is significant for theory and application. Due to the lack of prior knowledge of faults, it is always a great challenge for researchers to extract fault features from signals. In this paper, an innovative fault detection method for a reciprocating plunger pump is proposed based on an unsupervised feature encoder (UFE) and minimum covariance determination (MCD). Firstly, a criterion based on mutual information maximization for local and global features is proposed for unsupervised feature extraction. In addition, an unsupervised training strategy based on negative sampling is proposed to train the encoder, so that the model can converge quickly and improve the stability of model training. A fault detection algorithm for a reciprocating plunger pump is proposed based on a UFE and MCD. Finally, the effectiveness and superiority of the proposed method are verified with the measured data for a reciprocating plunger pump. The results show that the proposed method can accurately detect the faults in the reciprocating plunger pump with a detection accuracy of more than 98%. Compared with other methods, the proposed fault detection algorithm has better applicability and accuracy for fault detection with fault-free data.

The recent research shows that data-driven inertia navigation technology can significantly alleviate the drift error of micro-electro-mechanical system inertial measurement unit (MEMS-IMU) in pedestrian localization. However, most existing methods must rely on attitude information provided by external procedure (such as smartphone API), which violates the original intention of full autonomy of inertial navigation, and attitude information is also inaccurate. To address the problem, we propose a pedestrian indoor neural inertial navigation system that does not rely on external information and is only based on low-cost MEMS-IMU. First, a deep learning based neural inertial network was designed to estimate attitude. Then, in order to obtain position estimation with both global and local accuracy, an invariant extended Kalman filter (IEKF) framework was proposed, where 3D displacement and its uncertainty regressed by a deep residual network are utilized to update IEKF. Extensive experimental results on a public dataset and a self-collected dataset show that the proposed method provides accurate attitude estimation and outperforms state-of-the-art methods in position estimation, demonstrating the superiority of our method in reliability and accuracy.

To obtain geometric information and color texture simultaneously, a surface structured light sensor consisting of a monochrome camera, a color camera, and a projector is proposed. The sensor uses a color camera to acquire surface color information while using it as a geometric measurement unit and matching with the monochrome camera to obtain geometric information. Due to the Bayer array and demosaicing algorithm of the color camera, pixel RGB components are always coupled with interference from other channels. However, existing color de-crosstalk in reconstruction is merely applied to the decoupling of color composite patterns, ignoring the intensity errors present in color fringe patterns under monochrome illumination. In our sensor, de-crosstalk of monochromatic patterns is considered to guarantee the reconstruction accuracy. The high-accuracy measurement of the sensor is validated by reconstructing standard steps, yielding a mean absolute error of 0.008 mm for distance measurements. In addition, the reconstruction experiment of a terracotta warrior verifies that the proposed sensor has potential application in the digital preservation of cultural relics.

High-precision positioning with global navigation satellite systems (GNSS) remains a significant challenge in urban environments, due to the outliers caused by the insufficient number of accessible satellites and environmental interference. A GNSS outlier mitigation algorithm with effective fault detection and exclusion (FDE) is required for high-precision positioning. The traditional methods are designed to deal with zero-mean noise in GNSS, which leads to instabilities under biased measurements. Considering that GNSS data are typical time series data, a dynamic FDE scheme is constructed by combining a prediction-model-based method and a dissimilarity-based method. First, a hybrid prediction model which combines autoregressive integrated moving average (ARIMA) model and multilayer perceptron (MLP) model is proposed to provide pseudo-GNSS series by predicting the vehicle's location for several future steps. Then, a dissimilarity-based method of dynamic time warping measure is utilized to analyze the pairwise dis-similarity between the pseudo-GNSS series and the received GNSS series. The performance of the different models in forecasting is evaluated, and the results show that the positioning accuracy is significantly improved by applying the ARIMA-MLP. The effectiveness of the proposed FDE method is verified through simulation experiments and real experiments based on a typical urban canyon public dataset collected in Tokyo.

In the application of small field angle lidar for robot SLAM (simultaneous localization and mapping), livox mapping can provide accurate odometer information and point cloud information of the environment with good precision for the robot in a short time. However, over long periods of motion, the laser odometer calculated by livox mapping will produce a large offset, which will reduce the localization accuracy and mapping accuracy of the robot. To overcome above problem, a lidar-inertial navigation odometer compact fusion method based on the idea of complementary filtering is proposed in this paper. By taking advantage of the good static performance of the accelerometer for a long time, the angle value obtained by the gyroscope integration is corrected. In the back-end optimization, the Jacobian matrix obtained by the residual calculation of the acceleration in the navigation coordinate system obtained by IMU and the gravitational acceleration is tightly coupled with the Jacobian matrix of the lidar residual. Different weights are given to the residual of each part, and the odometer is solved iteratively to further improve the pose accuracy of the whole SLAM system. In this paper, the method is applied to Livox-Mid40. The experimental results show that it can reduce the drift of long time and long distance and improve the accuracy of the system localization and mapping.

Vibration sensors are key components in low-frequency micro-seismic monitoring, and their performance directly determines the accuracy of monitoring results. In response to the current problem that fiber Bragg grating (FBG) vibration sensors are difficult to effectively monitor micro-seismic low-frequency vibration signals, a rigid L-shaped beam FBG vibration sensor based on bearings is proposed. Firstly, a sensor model is established and theoretically analyzed; secondly, key parameters are optimized using differential evolution algorithm and imported into COMSOL simulation software for static stress analysis and dynamic characteristic analysis; finally, the sensor prototype is developed and a low-frequency vibration test system is set up to verify the sensor performance. The results reveal that the inherent frequency of the sensor is 57 Hz, with a flat response band of 0.3–35 Hz, a frequency lower limit of 0.05 Hz, a transverse interference degree of 4.5%, an average sensitivity of over 800 pm g⁻¹, a dynamic range of 67.75 dB, favorable linearity, and the ability to achieve temperature self-compensation. Research findings provide new insights into low-frequency micro-seismic monitoring.

Rolling bearings are critical components in many industrial fields, and their stability directly affects the performance and safety of the industrial equipment. Accurate prediction of remaining useful life (RUL) of rolling bearings is a heated topic in modern research. Traditional strategies are unable to efficiently exploit the significant features of the data, resulting in the inability to determine the starting time of prediction along with a reduced prediction accuracy. Accordingly, this paper proposes a novel data-driven prediction model named ConTriFormer, which incorporates multi-feature triggers focusing on various scales of input signals, and the ConvNeXt V2 sparse convolution strategy within the contextual Informer architecture for estimating RUL. Firstly, significant feature indicators of the original data are calculated to construct feature triggers, resulting in a multi-feature fusion. Secondly, the starting time for prediction is obtained through quantified results from fault-sensitive triggers. Thirdly, the original signal with triggers embedded is encoded and organized into sparse matrices to facilitate the simplification of subsequent computations. Sparse features and dynamic context information reflecting bearing state changes are obtained through ConvNeXt V2 sparse convolution, which is input into the Informer structure with contextual attentive structures inside for better adaptability to long time-span dynamic data and lower spatiotemporal complexity for feature mining and prediction. Finally, the prediction results are obtained by mapping output values to the remaining life through a fully connected layer. The proposed algorithm is compared with mainstream deep learning algorithms such as Bi-LSTM and Convolutional Transformer using the XJTU-SY dataset and PHM 2012 dataset, and the effectiveness of model is verified with ablation study. Results show that, the proposed method can more accurately predict RUL, providing a high-precision and intelligent method for prognostics health management of rolling bearings.

Multimodal natural frequency is a crucial factor in determining the structural stability of bridges. Global navigation satellite system (GNSS) has become an increasingly important tool for monitoring the structural health of long-span bridges. This paper proposes a method for accurately determining multimodal natural frequencies in these structures using GNSS monitoring data. The proposed method involves decomposing GNSS displacement data into several signals that correspond to each mode using auto-regressive power spectrum decomposition, extraction of Intrinsic Mode Functions (IMFs) using empirical mode decomposition (EMD), identification of multimodal natural frequencies from the extracted IMFs using random decrement technique and Hilbert transform. The proposed method was validated through a simulation test and was applied to the Yingwuzhou Yangtze River Bridge. Results showed that this method was able to accurately identify the first six modal frequencies with a relative error of less than 8.09% compared to the theoretical values obtained through a finite-element model. This method outperforms other methods such as peak-picking, Complete Ensemble EMD with Adaptive Noise, and empirical wavelet transform, which can only identify the first three modes or fewer. Finally, four fieldwork experiments with different GNSS data show that the maximum range of relative errors of each identification is 3.65%, which fully demonstrates the effectiveness and universality of this method.

Despite the diverse number of machine learning algorithms reported in the literature for machine fault detection, their implementation is mainly confined to laboratory-scale demonstrations. The complexity and black-box nature of machine learning models, the processing cost involved in appropriate feature extraction, limited access to labeled data, and varying operating conditions are some of the key reasons that curtail their implementation in practical applications. Furthermore, most such models serve as decision support tools, aiding domain experts in root cause analysis, and are not truly autonomous by themselves. To address these challenges, we present a lightweight autoencoder-based unsupervised learning framework to accurately identify machine faults against the changing operating conditions in a real-world scenario. The fault detection strategy is further strengthened by a model agnostic Shapley Additive exPlanations (SHAP)-based method (kernel SHAP) for identifying the most prominent features contributing to fault detection inference, the findings of which are then explored for identifying trends and correlations among prominent features and various types of faults. The framework is validated using two widely used and publicly available datasets for machine condition monitoring, as well as a large industrial dataset comprising 18 machines installed at three factories in India, monitored for several months.

Incipient faults features are often extremely weak and susceptible to heavy noise, making it challenging to obtain the concentrated faulty energy ridges in the time-frequency domain. Thus, a novel impulse-enhanced sparse time-frequency representation (IESTFR) method is proposed in this paper. First, the time-rearranged multisynchrosqueezing transform is utilized to produce a time-frequency representation (TFR) with a high energy concentration for faulty impulses. Next, a new non-convex penalty function is constructed by the hyperbolic tangent function, which can enhance the periodic impulsivity of sparse TFR for more obvious fault characteristic frequency. Moreover, the time-frequency transform is evaluated and compared by simulated signals and a selection strategy for the regularization parameter is designed. Simulated signals and two experimental signals are applied to verify the effectiveness of IESTFR, and the results show that IESTFR is effective and superior in bearing incipient fault feature extraction.

Vibrational analysis is one of the usual methods to monitor rotational machines that uses sensors fixed on the frame. Although effective, this method can be expansive and uneasy to use particularly at slow speeds (<60 rpm). This paper offers to evaluate the capabilities of a 6-axes low-cost inertial measurement unit which is positioned on the shaft to detect and diagnose bearing defects. This study is based on experimental data allowing to (i) choose the most relevant axis for better defect detection, (ii) to compare statistical indexes using usual vibratory analysis for a speed range from 5 to 400 rpm and for five defect severities, (iii) to compute the defect severity from a shocks local study.

Remaining useful life (RUL) prediction is crucial in the field of engineering, which can reduce the frequency of accidents and the maintenance cost of machinery. With the increasing complexity of rotating machinery, the data analysis methods based on deep learning have become the mainstream methods of prediction work. However, most of the current RUL prediction methods only use single-sensor data as input, which cannot effectively use multi-sensor data. In addition, as an advanced deep learning prediction method, temporal convolutional network (TCN) only uses the past time information of vibration data to determine the current health status of bearings, while ignoring the importance of future time information of vibration data. To solve the above problems, a bearing RUL prediction method based on multi-sensor data fusion and bidirectional-temporal attention convolutional network (Bi-TACN) is proposed in this paper. In multi-sensor data fusion, multi-sensor data are combined into multi-channel data, and a channel-weighted attention is designed to emphasize the importance of each sensor data. Compared with traditional multi-sensor data fusion, the proposed fusion method allows deep prediction networks to learn more useful feature information from multi-sensor data. Then, Bi-TACN is developed to predict the RUL of bearings. Bi-TACN is mainly composed of the forward TCN block and the backward TCN block, both of which can learn the past and future time information of multi-sensor data simultaneously. Moreover, a temporal attention mechanism is embedded in Bi-TACN to adaptively calibrate the weights of the two TCN blocks, so as to achieve dynamic feature fusion of past and future time information. RUL prediction experiments are carried out through Xi'an Jiao tong University bearing dataset and PHM 2012 bearing dataset respectively. Compared with the advanced prediction methods, the proposed method can accurately predict the RUL of more types of bearings and has low prediction errors.

Digital fringe projection (DFP) is widely applied in three-dimensional (3D) shape measurements. However, its performance is severely disturbed while measuring complex surfaces with an extensive range of reflectivity. In this paper, to enhance the underexposed regions and reduce the saturated regions, an automatic and effective method is proposed for the multi-exposure determination of the DFP system. During the whole process, only one unsaturated image must be captured preliminarily, which could establish an adaptive strategy based on the intensity value distribution function. Raw absolute phase maps captured with different exposure times are synthesized at the pixel level for phase retrieval and 3D reconstruction. To evaluate the effectiveness of our method, a high dynamic range method and a global optimal exposure method are introduced for comparison. Verification experiments demonstrate that our method performs better for the 3D reconstruction results of complex surfaces with a large reflectivity range.

Optical coherence tomography (OCT) has the advantage of high precision in measuring micrometre-scale precision structures, but its measurement range is limited by the correlation with Rayleigh length of Gaussian beam. Therefore, this study used the Airy beam to extend the focal depth of the OCT system. The system's coupled power was increased by 7.34 times when the sample position was at an optical defocus of 3 mm. Measurements of the resolution plate and silicon wafer sample showed that Airy beam OCT had a lateral resolution of 7.4 μm and the focal depth was >6 mm. While the focal depth of the Gaussian beam OCT with the equivalent lateral resolution was less than 1 mm. This improved OCT system demonstrates significant potential for sample measurements at large aspect ratios in the industrial surface detection field.

Microfluidic flows feature typically fully three-dimensional velocity fields. However, often the optical access for measurements is limited. Astigmatism or defocus particle tracking velocimetry is a technique that enables the 3D position determination of individual particles by the analysis of astigmatic/defocused particle images. The classification and position determination of particles is a task well suited to deep neural networks (DNNs). In this work, two DNNs are used to extract the class and in-plane position (object detection) as well as the depth position (regression). The performance of both DNNs is assessed by the position uncertainties as well as the precision of the size classes and the amount of recalled particles. The DNNs are evaluated on a synthetic dataset and establish a new benchmark of DNNs in defocus tracking applications. The recall is higher than compared to classic methods and the in-plane errors are always subpixel accurate. The relative uncertainty in the depth position is below 1% for all examined particle seeding concentrations. Additionally, the performance on experimental images, using four different particle sizes, ranging from 1.14 $\unicode{x03BC}\mathrm{m}$ to 5.03 $\unicode{x03BC}\mathrm{m}$ is analyzed. The particle images are systematically rearranged to produce comprehensive datasets of varying particle seeding concentrations. The distinction between particles of similar size is more challenging but the DNNs still show very good results. A precision above 96% is reached with a high recall above 95%. The error in the depth position remains below 1% and the in-plane errors are subpixel accurate with respect to the labels. The work shows that first, DNNs can be trained with artificially rearranged data sets based on individual experimental images and are therefore easily adaptable to various experimental setups and applicable by non-experts. Second, the DNNs can be successfully adapted to determine additional variables as in this case the size of the suspended particles.

To improve the efficiency of port bulk handling, a fast volume measurement algorithm for irregular bulk cargo is proposed in this paper. The elevation laser scanner and solid-state Lidar are used to determine the geometric information of bulk piles. The 3D point cloud data of the irregular bulk cargo was extracted, and the volume of the pile was calculated using the point cloud. To realize fast measurements, the algorithm first obtains a series of sliced point clouds and generates the slice matrix via dimensionality reduction and rasterization. Next, the area of the slice matrix is filled by the X-scan line algorithm. Finally, the volume of the whole point clouds is obtained by integrating the area of each slice matrix. Extensive experiments on datasets of realistic scenarios demonstrate that the proposed measurement method can complete point cloud reconstruction and volume calculation for different types of stockpiles with a good balance of accuracy, robustness, and execution efficiency.

A calibration method for line-structured light (LSL) by using a virtual binocular vision system (VBVS) composed of one camera and a front coating plane mirror is promoted in this work. The front coating plane in the VBVS can generate much less coplanarity error in lithographic feature points and remarkably decline the imaging distortion during back coating. An encoded target is proposed to distinguish between real corners and virtual corners (mirrored corners) and achieve high-precision matching between real and virtual corners when the target is occluded during the VBVS calibration. A parameter optimization method based on 3D constraints is presented in the work to obtain accurate structural parameters and thus guarantee precise reconstruction of the LSL. Moreover, the laser stripe and its mirrored image meet the auto-epipolar constraint. Therefore, the matching between the real and virtual stripes can be realized based on the vanish point. The performance of our method is verified in the experiments.

This paper proposes a new optical flow (OF) method for particle image velocimetry applications. The proposed method is based on the use of an a priori sparse knowledge of the flow. A particular insight is given to the optimization derivation based on an image-independent method. Two alternatives are introduced. The first one uses particle-tracking velocimetry estimates as subpixel information to describe the finest velocity scales. The expected true displacements related to the motion of the individual particles are used as anchors for the optimization procedure when the density of the particles is large enough. Alternatively, the second method solves the well-known median problem based on new image-independent functions in areas of low particle density. Studies have been carried out on synthetic images to characterize the error and analyze the impact of image parameters (particle density, particle size, or noise) on the methods. The new methods are compared with a reference method against synthetic data: two Lamb-Oseen vortex rings and a 3D Turbulent Homogeneous and Isotropic flow. The results show that the performances of the new method exceed those of the reference method in almost all tested cases, except for images with particles of relatively small size. It is notably shown that the new method is less dependent on the particle density and the noise embedded in the images than other OF estimators.

This study presents a comparative analysis of droplet and conical liquid film velocities in sprays by a pressure swirl injector with Abramovich geometric constant K = 1.986. The velocities were measured simultaneously by schlieren image velocimetry and sequential images of the liquid film, using a high-speed camera with 8192 fps and shutter of 2 μs. Other parameters, such as discharge coefficients, spray cone angles and breakup lengths were also determined, using water as test fluid, for injection pressures from 0.05 to 0.5 MPa. Experimental velocity data were compared to results from different semi-empirical equations. The breakup lengths decreased continuously from around 20 mm to 15 mm for injection pressures from 0.1 to 0.5 MPa, while spray cone angles increased continuously from about 42°–53°, for pressures from 0.2 to 0.5 MPa. Mean axial droplet velocities varied from 4.7 m s⁻¹ to 14.5 m s⁻¹, while the mean total droplet velocities varied from 5 m s⁻¹ to 16.2 m s⁻¹ and the total liquid film velocity increased from 5.7 m s⁻¹ to 20.2 m s⁻¹, approximately, for increasing injection pressure. Liquid film velocities were about 15%–28% higher than the droplet velocities in the pressure range considered, due to the energy required for liquid film breakup and the air drag on the droplets. The current findings underscore that significant discrepancies may arise when relying on inadequate velocity data, particularly when employed in the computation of key parameters such as the Reynolds and Weber numbers.

The finite difference time domain (FDTD) method was used to solve the Maxwell's equation to obtain the reverse time migration (RTM) of ground penetrating radar (GPR) signals, namely, the FDTD-RTM. In order to ensure that the correct numerical solution of iterative calculation was achieved, it was necessary to obtain high-resolution signals, which greatly limits the applicability of FDTD-RTM in engineering. Based on the characteristics of wavelet multi-resolution analysis, this study proposed a super-resolution signal reconstruction method to improve signal resolution, with the view to completely solving the problem of FDTD-RTM limitation caused by insufficient signal sampling using GPR. The results of electromagnetic simulation showed that the signals reconstructed by the above method were highly similar to the signals sampled with the same resolution. On this basis, reverse time migration electromagnetic simulation and physical model tests were designed. The results of both experiments showed that the under-sampled GPR signals could achieve FDTD-RTM following super-resolution reconstruction, and that the migration imaging results of the target were basically consistent with the design scheme. The signal super-resolution reconstruction method based on the wavelet theory was thus shown to successfully achieve the overall application of FDTD-RTM in GPR signal analysis.

Electrical resistance tomography (ERT) can be applied to two-phase flow pattern identification which is a key research direction for improving the operational safety of different industrial equipment systems with complex flow fields. Aiming at the existing problem that the traditional algorithm for defining flow patterns cannot accurately establish the mapping relationship between the measured voltage from ERT system and the two-phase flow conductivity distribution, a novel one-dimensional structure reparameterized convolutional neural network (1D-SRPCNN) algorithm for two-phase flow pattern image reconstruction based on ERT is proposed. First, finite element method and deep learning software framework are used to build dataset and train the neural network model respectively. Second, a deep residual network (ResNet) is used as the main network structure in the algorithm, and the one-dimensional multiscale feature extraction block (1DMSFE-Block) is improved by structural reparameterization. Then, multiscale convolution is introduced to 1DMSFE-Block for extracting features of different receptive field sizes and performing linear fusion, and the predicted two-phase flow conductivity pixel vector is obtained by the feature map passing with three fully connected layers. The results show that 1D-SRPCNN has high reconstruction performance, the average relative image error is 5.15%, the average correlation coefficient is 97.2%, and it has high anti-noise performance and generalization performance. Different experimental data also show that 1D-SRPCNN has high image reconstruction accuracy and efficiency. The research will provide important theoretical support for accurately identifying two-phase flow patterns in different fields.

To reduce the influence of material particle size on coal gangue identification, a particle size identification method, and an adaptive image enhancement method are proposed, which can accurately identify the particle size of poorly segmented and mutually blocked materials, effectively reduce the reflection and blur of the image surface and enhance the texture details. Through the research of coal gangue images with different particle sizes, it is found that the image quality and feature curve distribution of small particle size are different from those of large particle size, and the gradient features are worse. In this paper, the accurate identification of particle size is realized using the difference in image quality and texture, and the identification rate is 99.25%. Through the image enhancement method in this paper, 33.41% of the reflection on the image surface is removed, and the average gradient is improved by 74.01%, which effectively improves the image quality and the ability to express texture information. This algorithm has high environmental adaptability, and the identification rate can reach 99.16% in moderate illumination, 98.33% in dim illumination, and 96.33% in strong illumination. This research provides a valuable idea for image processing and identification technology based on machine vision.

The objective of an effective human-robot collaborative (HRC) task is the maximization of human-robot competencies while ensuring the user's convenience. In photometrically challenging and unstructured HRC environments, data obtained from vision sensors often tend to get degraded due to illumination irregularities and spatio-temporal complexities. To extract useful and discriminative features from the data under such situations, locality-sensitive methods like locality preserving projections (LPPs) become quite useful as it captures the local geometric structure of the high-dimensional data. In LPP, the local structural information is encoded in the form of weight values between two samples in the higher-dimensional Euclidean space. The weight values are learned in a regular and continuous manner which only depends on the spatial distribution of the data. Moreover, because of its weight dependency solely on the Euclidean distance, improper weight values can occur frequently, as the Euclidean distance is susceptible to noise, outliers, and different types of geometrical transformations. This paper proposes an adaptive weight learning method to be utilized in the weight computation of LPP, which allows it to adaptively select and extract more discriminative features from the higher-dimensional input data while preserving the intrinsic structural information of the data. Additionally, to alleviate the issues with spatial dependency, the concept of bilateral filtering that incorporates the range weights from the feature space along with the similarity weight in the Euclidean space has been utilized here. This paper proposes an augmented version of adaptive spatial-feature kernel-guided bilateral filtering inspired LPP which addresses two of these basic and fundamental issues of the conventional LPP.

Visual simultaneous localization and mapping (SLAM) algorithms face challenges in complex underwater scenarios, such as turbidity, dynamism, and low texture, where point features are unreliable and can lead to weakened or even failed systems. To overcome these issues, high-level object features are considered due to their accuracy and robustness. In this paper, we introduce an effective object-level SLAM method that employs a stereo camera to enhance the navigation robustness of autonomous underwater vehicles and generates a detailed semantic map. Point features and object features are integrated to serve the proposed approach. We begin by detecting 2D objects in images using a state-of-the-art neural network, followed by obtaining 3D objects described by the general model through the principle of multi-view geometry and eventually constructing semantic landmarks. To account for object data association, we present an object match method that takes into consideration the stereo camera characteristics in a single stereo frame and a filter-based approach to track the landmarks in odometry. Experiments are also conducted using the KITTI dataset and our sequences collected from the pool and coast. The evaluation results indicate that the proposed method can improve the performance of ORBSLAM2 in terms of both navigation robustness and mapping information in underwater scenarios.

To reliably measure multiple spatio-temporal distribution parameters of fragments in the blast field, a measurement method for spatio-temporal distribution of fragments based on a high-speed camera network is proposed. The method is based on the principle of binocular stereo vision measurement and consists of four steps: fragment image acquisition, target detection of multi-view images, fitting of motion trajectory, and calculation of the spatio-temporal distribution parameters. The test results demonstrate that the method can simultaneously measure multiple parameters in the intersecting field of view. Compared with the measurement results of the array target, the absolute value of the measurement relative error of velocity was less than 1.5%, and the measurement results of scatter angle and distribution density were consistent. It indicates that the measurement results of the proposed method are reliable, and the method is a feasible and effective measurement method for spatio-temporal distribution parameters of fragments in the explosion filed.

Electrical impedance tomography can reconstruct the complex conductivity distribution by injecting a current or voltage at a specific frequency into the target domain. The complex conductivity spectroscopy of numerous biological tissues is frequency-dependent. A suitable excitation frequency is vital to high-quality imaging over a wide frequency range. This paper investigates the relationship between the parameters of the biological tissue impedance model and the impedance spectroscopy. A frequency selection method based on the impedance spectroscopy is proposed, in which the impedance spectroscopy on a specific electrode at the domain boundary to be measured is first scanned. In TD-EIT, the difference between the target's impedance spectroscopy and the null field is calculated, and the frequency corresponding to its extreme value is used as the excitation signal for TD-EIT. The excitation frequency in FD-EIT is the frequency corresponding to the extreme point in the target impedance spectroscopy, and we have also used this method to image the interior of the maize ear. This study provides a quick and efficient method for determining the excitation frequency for EIT, allowing researchers to find the best excitation frequency for high-quality imaging during actual measurements.

A predictive model of tractor driver's eye positions based on the anthropometric parameters and tractor cab layout parameters was established using the statistical approach. To our knowledge, this model is the first published predictive model of tractor drivers' eye positions that takes account of their body dimensions and tractor types. In order to predict the distribution of the drivers' eye positions precisely, the fore-and-aft positions of the driving seat determined by 180 tractor drivers in three different tractors and the relative positions of the eyes to the designated reference points were recorded to quantitatively demonstrate the distribution rule of their eye positions. The multiple linear regression and principal component analysis were adopted to processing the feature parameters of human and tractor, and the prediction effect of the model was assessed according to the proportion of the driver's eye positions from side view those were within the range of the eyellipse model. The experimental result revealed that the centroid positions and principal axis dip angle of the model differed from each other respectively, while the axis lengths were the same. The principal axis dip angle was horizontal or oblique forward-and-upward. And the tractor layout parameters had a significant impact on the centroid coordinate and principal axis dip angle of the model, while the dimension of the model was associated with the anthropometric parameters mainly. The prediction accuracy of the model exceeded 80%. The regression predictive model provides a reference tool for the visual field design of tractor displays and cab layouts.

This study presents the design and test of a novel ultra-simplified flexure-based spring-mass device (NUFSMD) with high sensitivity for relative gravimeters. The NUFSMD is mainly composed of a highly sensitive spring based on flexure hinges (SFH) and a cantilever beam. When the change of gravity causes SFH to rotate, an amplified displacement can be obtained on the tail of the cantilever beam. The displacement is measured by a capacitive displacement sensor. The change of gravitational acceleration can be calculated based on the output of the sensor. A theoretical analysis of the working principle of the NUFSMD is established. A prototype NUFSMD is machined from 304 stainless steel and a series of experiments are carried out. The effectiveness of gravity measurement is demonstrated by finite element simulations and experiments. The experimental results show that the sensitivity of the NUFSMD is about 15.27 μGal nm⁻¹, and the resolution is about 1.83 μGal. The designed NUFSMD shows outstanding performance. And the sensitivity and resolution of the NUFSMD can be further improved by increasing the number of flexible hinges connected in series. Compared with the existing spring-mass devices in high-precision relative gravimeters, the significant advantage of the NUFSMD is that it is simple in structure and easy to machine, which helps to greatly improve performance. Firstly, it is simple in structure, so it can be machined in one piece, which considerately contributes to stability and precision. Secondly, it is easy to machine, so it can be machined from high-performance but difficult-to-machine materials to improve its performance. Thirdly, there are a few parameters that need to be traded off for miniaturization because of the simple structure, so it is easy to be miniaturized to adapt to narrow measurement space. Therefore, it is expected to greatly improve the performance of the existing spring-mass device in relative gravimeters.

Currently, most of the electrical tomography measurement systems are developed on microcontroller units, digital signal processors, or field programmable gate arrays, whereas image reconstruction and display functions are implemented in a separate host computer. Therefore, the complete hardware usually comprises a measurement system and an image reconstruction computer. The current trend is to develop industrial standard electrical tomography systems, which can implement the image reconstruction and measurement functions into an integrated processor/chip. However, most of the above mentioned processors do not provide the necessary computational resources as required by the computationally intensive image reconstruction function. This paper describes a hardware scheme for implementing image reconstruction functions on a heterogeneous hardware platform, wherein the processor system and programmable logics (PLs) are tightly coupled and can achieve better resource utilization and overall system performance. By adopting a high-level synthesis method, the image reconstruction algorithms can be realized jointly by the PLs and processor system. In this hardware scheme, the algorithms are properly optimized to achieve better data throughput and execution efficiency, that is, by utilizing parallel computation in PLs. Details of the hardware scheme and method of accelerating the image reconstruction process are presented in this paper. Implementation results show that the proposed heterogeneous hardware scheme can achieve the image reconstruction rates of 24 and 1700 frames per second while employing the iterative and non-iterative algorithms, respectively.

Conventional automatic optical inspection (AOI) systems using rule-based image processing suffer from precision and velocity issues, particularly when simultaneously measuring multiple defects of a product in a high-speed production line. Such AOI stations usually become a bottleneck in the line. This paper presents a three-phase model for defect detection based on convolutional neural network to release the cycle time of the line. The phase I model using a deep residual network (ResNet50) performs defect classification of products with high accuracy. The phase II model is another independent ResNet50 that classifies defect products into defect categories, rescuing good products that have been mistakenly killed in the previous stage and replacing the conventional re-inspection labors. The phase III model is a you only look once—based network that detects multiple defects and their positions simultaneously in a single product, providing informative quality data for continuous improvement. The proposed model successfully resolves the issue of multiple-defect and multiple-len quality inspection in a high-speed production line. The proposed model resolved defect inspection by integrating object detection and defect classification simultaneously. By deploying the three-phase model in a tiny electronic connector component production line, the present model has demonstrated that it reaches high precision and facilitates prompt quality correction for high-speed production lines.

We represent small Z- and Y-gradient coils with higher gradient strength and slew rate for mouse MRI. The coil geometries are optimized based on predetermined conditions for the gradient strength and slew rate in terms of minimum power consumption and acceptable linearity for the variation of the magnetic fields using the particle swarm optimization algorithm in computer simulation technology. The simulation results verify that the magnitudes of the field strength and slew rate have reached up to $144\,\mathrm{mT\cdot m}^{-1}$ and $1297\,\mathrm{mT}\cdot(\mathrm{m\cdot ms})^{-1}$ for the Z-gradient coil and increased to $109\,\mathrm{mT\cdot m}^{-1}$ and $1264\,\mathrm{mT}\cdot(\mathrm{m\cdot ms})^{-1}$ for the Y-gradient coil, respectively. Furthermore, the optimal geometry of the Z-gradient coil derived from a simulation study was then fabricated, and its parameters were measured using a Hall sensor. The gradient strength and slew rate of the fabricated Z-gradient coil were measured at $131\,\mathrm{mT}\cdot{\mathrm{m}^{-1}}$ and $1180\,\mathrm{mT}\cdot(\mathrm{m\cdot ms})^{-1}$, respectively, which have only a 9% difference from the corresponding simulated parameters.

At present, pipeline inspection gauge (PIG) is the main means to ensure the reliability of oil and gas transportation pipelines. Due to the unstable friction between the PIG and the pipeline, the failure and blockage of the PIG often occurs. In this paper, the theoretical modeling and experimental measurement of fluctuating friction on PIG are studied. Firstly, considering the coupling relationship between cylindrical shell and conical shell, the contact radial force model between PIG and pipeline is derived. Based on which, the nonlinear frictional model of PIG is proposed, and the parameters are measured by several experiments. Then an experimental platform is developed to measure the fluctuating friction force of PIG in actual operation, and verify the accuracy of the proposed model. The motion characteristic analysis and dynamics research are carried out, and the PIG creeping phenomenon caused by fluctuating friction is explained. Finally, based on the proposed nonlinear friction model, the influence of parameters is analyzed. The conclusions based on the proposed model and the experiments results can provide technical supports for practical engineering application.

High-power diesel engines are core power equipment in some key fields, and fault diagnosis is of great significance for improving their long-term operational reliability and safety. The lack of sufficient fault samples, known as few-shot conditions, has resulted in the performance degradation of most current intelligent diagnostic methods under complex and variable working conditions. Moreover, the end-to-end deep learning diagnostic models based on implicit layer features do not have clear physical meanings, making it difficult for intelligent diagnostic models to be recognized by equipment managers in key fields. Mechanism-based fusion methods can effectively alleviate the above problems by incorporating fault-sensitive features with clear physical meanings. However, the lack of sufficient knowledge-related features that can characterize fault information is a challenge facing current mechanism fusion methods. In this work, we propose a progressive adaptive sparse attention mechanism-guided knowledge feature incremental mining method. Fault-sensitive segments are obtained based on attention-supervised learning with an improved activation function. Both knowledge-related features and hidden layer features from these segments are incorporated into the feature library. A differential constraint is designed between the distribution of knowledge features and hidden layer features, encouraging the model to focus on learning new knowledge-related features in the subsequent iteration process. Finally, a feature set containing a large amount of knowledge features is constructed to enhance fault diagnosis. Testing on a V12 diesel engine test bench under the conditions of rich and few-shot data sets, shows that the proposed method significantly enhances the fault recognition performance of diesel engines under complex variable working conditions and few-shot data set conditions.

As meter data becomes more and more important in the power industry, detection robots are led into substations for automatic collection of the pointer meters. However, the meter images captured in low illumination environments are unclear, resulting in poor recognition of the meter reading. A low-illumination image enhancement method based on virtual exposure is proposed in this paper, improving the dark and bright areas of low-illumination images, respectively. Then the image fusion was performed based on the Laplace pyramid to obtain clear meter images. In addition, the dial area was extracted using the Hough circle transform, and the pointer's rotation center was fitted using the least squares approach. Finally, the straight line of the pointer was extracted, and the data reading was based on the line segment detector algorithm. Case studies show the above method has good robustness in low illumination environment, with high rate, and accuracy during the image enhancement and automatic reading of the pointer meter.

Nowadays, the methods of remaining useful life (RUL) prediction based on deep learning only use single model, or a simple superposition of two models, which makes it difficult for to maintain good generalization performance in various prediction scenarios, and ignores the dynamic sensitivity of features in the prediction, limiting the accuracy. This paper proposes a method of RUL prediction of bearing using fusion network through two-feature cross weighting (FNT-F). First, a fusion network with two subnets is proposed in this paper to adapt to the prediction problem in different scenarios. Meanwhile, a method of cross weighted joint analysis of the two features is proposed to make up for the shortcomings of feature analysis and achieve complementarity between time-domain and time–frequency features.

The mapping of magnetic fields is an important task for characterizing permanent magnets, their systems, and the various devices that use magnets. However, commercial devices for this task are quite expensive and difficult to acquire. In this study, a design process for an automatized magnetic mapping setup utilizing a 3D printer was proposed via a do-it-yourself approach using common parts found in a laboratory. We used a 3D printer for two purposes: to create instrument parts to fix the Hall probe instead of the printer extruder, and to use the same 3D printer as the position controller of the magnetic probe. We describe the device assembly process, including hardware and software aspects. Several tests of the assembled setup were performed on samples with different magnetic texture dimensions from several centimeters (graduated permanent magnet) to tens of micrometers (flexible magnet with a planar Halbach structure and a 5¼ inch floppy disk). The spatial resolution was in the sub-millimeter range. The parameters of the mapping, such as the number of points and data acquisition time, were optimized for such samples.