Table of contents

The global climate observing system Reference Upper Air Network (GRUAN) provides reference measurements of the essential climate variables and their measurement uncertainty. An important aspect of the measurement uncertainty is the interpolation uncertainty. This problem arises, for example, when GRUAN processes atmospheric profiles collected by Vaisala RS41 radiosondes. For various reasons, the radiosonde sensor may fail to collect some values along the vertical profile in the atmosphere. As a consequence, estimation techniques to fill the data gaps and to provide an evaluation of the related interpolation uncertainty are welcome. This paper aims to understand and quantify the interpolation uncertainty of the relative humidity (RH). In particular, we consider linear and Gaussian process interpolation either unidimensional or multidimensional. We provide results in terms of interpolation distance in seconds, which measures the distance of an interpolated data from the interpolating data. We also consider measurement altitude and launch site. Although the Gaussian interpolation results to be generally better than linear interpolation, the difference in uncertainty is relatively small. The average column uncertainty difference is in the order of 0.2% and for large data gaps around 0.4%. The interpolation uncertainty is larger in the lower atmosphere. For example at 2–4 km altitude, the uncertainty is around 2% at 10 s distance, around 5% at 40 s and 10% at 90 s. In the upper atmosphere, RH and its uncertainty decrease. For example, at 14–16 km altitude, the latter is smaller than 2% RH at all interpolation distances.

X-ray microscopy instruments have the unique ability to achieve nondestructive imaging with higher spatial resolutions than traditional x-ray computed tomography (CT) systems. This unique ability is of interest to industrial quality control entities, as they deal with small features in precision manufactured parts (with tolerances in the order of ±25 µm or better). Since many of today's technology and manufacturing companies demand increasingly higher levels of precision, accuracy, and reliability for dimensional measurements on feature sizes that are much smaller than 5 mm, it would be ideal to further expand the imaging capabilities of x-ray microscopy to the field of precision metrology. To address such demand, this paper describes the development of a measurement workflow, through a package consisting of hardware and software, to improve the accuracy of dimensional data obtained with 3D x-ray microscopes (XRMs)—also known as sub-micrometer CT systems. The measurement workflow, called Metrology Extension (MTX), was designed to adjust and configure the XRM instrument work-zone to perform dimensional measurement tasks. The main adjustments of an XRM instrument through the MTX workflow, which must be implemented before scanning parts of interest for dimensional evaluation, include applying a distortion map correction on the image projections produced by the x-ray detector and a voxel scale correction. The main purpose of this article is to present, evaluate, and analyze the experimental results of various measurement tests to verify the metrological performance of several XRM systems operating with the MTX workflow. The main results show that these systems can produce repeatable and reproducible measurements, with repeatability standard deviations of the order of 0.1 μm and reproducibility standard deviations of less than 0.5 μm. In addition, coordinate-based 3D XRM measurements produce dimensional accuracies comparable to those offered by high-precision tactile coordinate measurement machines (with deviations within the range of ±0.95 µm). Therefore, once the MTX workflow is executed, XRM instruments can be used to measure small volumes, in the order of (5 mm)³ or less, with improved dimensional accuracy.

This paper presents imaging of copper slag solidification processes using electromagnetic tomography (EMT). It describes key components of the measurement system including a sensor that can operate in the harsh temperature environments. An integrated process tomography system is detailed that includes the main processing unit, i.e. a field programmable gate array and other front end circuitries. It is capable of delivering a wide excitation signal frequency ranging from 0.1 Hz to 500 kHz and a fast speed of 131 frames s⁻¹ with a typical signal-to-noise ratio of 66 dB–95 dB. Measurement trials were conducted on a molten converter slag solidification process where the copper slag changes from the molten state to solidification state during the cool-down for more than an hour. The phenomenon in which the disorderly distributed metal gradually forms solid and permeable object was observed, which can indicate the status of the process. This is the first report of observing such a process using an EMT system. Moreover location-based convergence analysis has been carried out in the imaging space and useful new insights have been gained for the copper production process, which would be difficult to obtain otherwise. Verification and calibration using x-ray diffraction and scanning electron microscope indicate the viability of the EMT-based measurement method.

Ultrasonic transmission tomography is an effective non-intrusive method for detecting gas–liquid two-phase flow patterns. A specific interest is the many processes whose reaction utilizes a bubble column, where the fast estimation of cross-sectional gas-holdup ratio is important for monitoring and control. In this study reference indirect image-based estimates were obtained from reconstructed tomographic data. Direct (non-image) estimation of the gas holdup ratio was also obtained using trained neural processing networks. Two forms were trialled: a generalized regression neural network (GRNN); and a long short-term memory (LSTM) network. Comparison trials were carried out for single-bubble, dual-bubble, circulation and laminar flows. Relative cross-sectional gas holdup error was selected for evaluation. For the image-based indirect trials the Tikhonov regularization algorithm had the lowest error range: 2.15%–15.64%. For direct methods the LSTM network had the lowest error range: 0.41%–9.63%, giving better performance than the image-based methods. The experimental data were used to verify the effectiveness of the network. The root-mean-square error of the test metrics for GRNN and LSTM network were 6.4260 and 5.4282, respectively, indicating that LSTM network has higher performance in processing the data in this paper.

The size and shape of the focal spot, i.e. the area where x-rays are generated in an x-ray tube, is a key figure of merit of a radiographic imaging system, because it governs the achievable resolution and, if the shape is irregular, can introduce anisotropic blur in the radiographs. Size and shape are mainly determined by the diameter of the accelerated electron-beam and electron interactions with the target material. Since focal spot parameters change over time, due to target wear and drift in the electron optics, traceable methods to monitor the focal spot size and shape are required. Here, we present a method to map the 2D intensity distributions of focal spots, ranging from sub-microfocus (0.1 µm) to mesofocus (>50 μm). The method evaluates radial profiles of a circular aperture and applies filtered back projection to reconstruct the focal spot intensity distribution. Special emphasis is placed on traceability and influence factors, such as image noise and aperture geometry. The method was validated over a focal spot range from 0.3 µm to 200 µm. Experimental results were compared to complementary methods, using line pair gauges (JIMA) for microfocus (1 µm–10 µm), e-beam characterisation for sub-microfocus (0.3 µm–1.0 µm), and pinhole camera imaging for mesofocus spots (50 µm–200 µm). Good agreement between the focal spot reconstruction and the complementary methods was achieved. Limitations due to edge penetration, caused by the finite thickness of the circular aperture gauge, were experimentally verified. From these findings, we derived recommendations for an improved gauge geometry, regarding its thickness and the aperture diameters. The method perfectly complements the well-established pinhole imaging, which is limited to focal spots larger than ∼50 µm.

Quality monitoring for laser powder bed fusion (L-PBF), particularly in-process and real-time monitoring, is of importance for part quality assurance and manufacturing cost reduction. Measurement of layer surface topography is critical for quality monitoring, as any anomaly on layer surfaces can result in defects in the final part. In this paper, we propose a surface measurement method, based on the use of scattered light patterns and a convolutional autoencoder-based unsupervised machine learning method, designed and trained using a large set of scattering patterns simulated from reference surfaces using a scattering model. The advantage of using an autoencoder is that the monitoring model can be trained using solely data from acceptable surfaces, without the need to ensure the presence of representative observations for all the types of possible surface defects. The advantage of using simulated data for training is that we can obtain an effective monitoring solution without the need for a large collection of experimental observations. Here we report the results of a preliminary investigation on the performance of the proposed solution, where the trained autoencoder is tested on experimental data obtained off-process, using a dedicated experimental apparatus for generating and collecting light scattering patterns from manufactured L-PBF surfaces. Our results indicate that the proposed monitoring solution is capable of detecting both acceptable and anomalous surfaces. Although further validation is required to fully assess performance within an on-machine and in-process setup, our preliminary results are encouraging and provide a glimpse of the potential benefits of using our surface measurement solution for L-PBF in-process monitoring.

Optical time-of-flight sensors have potential in the revolution of distance measurement. These sensors can continuously monitor the distance and track the movement of objects. However, the existing sensing methods for such distance optical sensors mainly calculate the flight time, e.g. pulse transmission and receiving time, without considering the environmental effects. Therefore, the measurement accuracy is severely reduced. There are other technologies with higher accuracy in distance measurement. Nonetheless, they are too expensive due to the high accurate power supply. In this paper, we innovatively improve the accuracy of continuous distance measurement using the artificial neural network (ANN) technique. The proposed method can be applied for very cheap optical distance sensors with analog output in a real-time system. Moreover, the proposed method can self-calibrate and be miniaturized for cheap analog sensor applications. The prototype is built with the infrared sensor GP2Y0A02YK0F and an Arduino control board (ESP32_DevC), and the ANN is implemented using the deep learning algorithm. The test results show that the distance measurement accuracy is significantly improved and the measuring range is increased from 15 to 150 cm. In addition, we calculate mean squared error, mean absolute error, mean bias error, and R² for further performance evaluation. The experimental results have proven the superiority of the proposed ANN method in optical distance measurement. The proposed method can be applied to many types of sensors.

To locate the impact source of a composite material structure with a series of fiber Bragg grating (FBG) sensors, a hybrid correlation interpolation method is proposed, in this paper, based on a weighted correlation analysis and a unidirectional cubic spline interpolation. First, the signals collected by FBG sensors are preprocessed on the frequency domain to reduce noise interference. Second, the impact similarity map is built by using the weighted cross-correlation value to estimate the location of the impact roughly. Third, the similarity map is further subdivided by the unidirectional cubic spline interpolation to increase the impact localization. Fourth, the FBG sensor network is experimental analyzed to optimize the layout and the structure of the FBG sensor network. The validity of the multi FBG sensor-based impact localization method is verified on a carbon fiber composite plate with a size of 650 × 500 (mm). The experimental result shows that the mean absolute error is 9.74 mm for our suggested method and 15.51 mm for the bicubic polynomial interpolation (BPI) algorithm, which is reduced by 5.77 mm. The maximum errors are 36.40 mm for the BPI algorithm while are reduced to 21.55 mm for our suggested method. Furthermore, the effect of FBG sensor combination layout and number on impact location accuracy is analyzed and compared. It shows that the sensor network layout of rectangular structure using four FBG sensors is a better choice after the comprehensive consideration of economic benefits, localization accuracy, and error rate.

Mineral insulated metal sheathed (MIMS) base metal thermocouples experience thermoelectric drift over their lifetime caused by use at high temperatures and metallurgical changes, causing spurious measurement errors. CCPI Europe Limited and University of Cambridge have designed a MIMS thermocouple with an additional inner sheath, in order to protect the thermoelements from the effects that cause thermoelectric drift. The performance of these dual-wall thermocouples and conventional type N and type K thermocouples are assessed at six different National Metrology Institutes (NMIs) using two different testing regimes: isothermal testing at 1200 °C, and thermal cycling tests between 300 °C and 1150 °C. The investigation demonstrates that in both testing regimes, the type N dual-wall thermocouples showed a significantly reduced thermoelectric drift by about a factor of three compared to the conventional thermocouples. There was no significant difference between the type K dual-wall and conventional type K thermocouples in the isothermal tests, and the type K dual-wall thermocouples showed greater drift than the conventional thermocouples in the thermal cycling tests, but the conventional type K thermocouples were less robust than the dual-wall type K thermocouples. The results presented in this paper represent an impartial assessment of the thermoelectric stability of both dual-wall thermocouples and conventional thermocouples, which may provide assurance to potential users.

Deep-learning-based fault diagnosis (FD) methods have shown remarkable superiority in the field of fault prognostic and health management (PHM). However, the performance of a deep neural network relies heavily on a substantial labeled training dataset, which is rare in actual industrial scenarios. Moreover, the extracted fault features under different working conditions follow different joint distribution. As a result, a deep model trained under one condition cannot be extended to others. To address the existing problem, this paper proposes a step-by-step gradual domain adaptive neural network to conduct cross-domain FD, which can realize precise alignment between the source domain and the target domain. Firstly, the maximum mean discrepancy is used to perform primary domain adaptation. Furthermore, two classifiers are set up to limit the discrepancy of target domain data in the classification decision. Finally, the exact alignment of class-level features is achieved by category prototype alignment. The extensive experimental results show the superiority and stability of the proposed method when compared with other conventional approaches.

Compared with the traditional assembly simulation based on theoretical models, this paper proposes a new pre-assembly analysis method of aircraft components based on measured data. Specifically, before the actual assembly of the product, digital measurement methods are used to obtain the measured data of the target features of the manufactured parts. Subsequently, the measured data is processed and reconstructed to obtain the actual geometric shape of the part, based on which the product is pre-assembled and analyzed to evaluate the assembly quality in advance. Finally, according to the analysis results, the assembly process is adjusted in time to reduce assembly trial and error and improve assembly quality and efficiency. This article systematically introduces the implementation process of the method, which is illustrated through two cases study on aircraft wing box assembly process. Experimental results demonstrate the feasibility and effectiveness of this proposed method for assembly of large aircraft components.

In order to realize the rapid start-up of a vehicle-carried inertial navigation system, this paper studies an in-motion alignment method based on an onboard rotating inertial navigation system (RINS) and GPS velocity information. Working with the rotation characteristics of the RINS, a Kalman filter model was established, that included the installation angles between gyros, the lever arm and the asynchronous time, by measuring the velocity difference between RINS and GPS. Multiple simulations and experimental verifications of the proposed in-motion alignment method have been carried out. The results show that the proposed in-motion alignment method can realize a rapid estimation of platform deflection angle, gyro installation errors, lever arm and the asynchronous time of RINS in the process of motion, which verifies the effectiveness of the proposed in-motion alignment method.

Non-destructive stress measurement using ultrasonic techniques is based on calculation of the acoustoelastic coefficient obtained from the relationship between material stress and sound wave velocity. Longitudinal critically refracted (L_CR) waves are the most suitable for ultrasonic stress measurement testing. This is because these waves exhibit a relatively large change in flight time in relation to a change in stress. In general, to calculate the acoustoelastic coefficient, tensile tests are performed with specimens extracted from the test objects to be inspected, and the difference in the speed of ultrasonic waves is measured in relation to the stress applied. However, train rails are usually subjected to compressive loads and no research has been conducted to confirm whether the results obtained using this method match the coefficients obtained from test objects in full cross-section. Therefore, results from compression tests on full cross-section specimens are needed as reference data to determine the reliability of the results obtained using extracted specimens. In this study, experiments were conducted to compare the result from applying compressive loads to full cross-section specimens to the experimental result obtained from specimens extracted from each rail section. For this purpose, ultrasonic probes capable of generating L_CR waves were designed and manufactured to apply the technique to measure the applied stress. Based on the results obtained using this probe, the acoustoelastic coefficient values obtained from each specimen were compared. Then, the feasibility of calculating the internal stresses in railway rails using the acoustoelastic constants obtained from the extracted specimens was determined, and the reliability of the method was verified. It was concluded that there was no significant difference in the acoustoelastic coefficients calculated using the data from extracted specimens and those using full cross-section specimens.

A method for measuring the notch length of plastic parts without stopping the machine in real-time industrial environment is proposed, which can be used to measure the notch length on the assembly line. A camera is used for capturing the live images of plastic images being moving straightly. Firstly, invalid frames, clear and blurred keyframes are preprocessed respectively. Then, for the processed keyframes, the corner extraction algorithm based on contours are used to obtain all corner points and pixel length of notch. Finally, the actual length is calculated by setting constraints on positional parameters. When tested in industrial environment, our algorithm can extract the corner points and calculate the notch length accurately and quickly. The average error of corner extraction is about 0.4%, and the measurement error of notch length is less than 1%. Compared with other methods, the algorithm has the advantages of speed and accuracy in dealing with rectangular plastic notches.

In actual industrial production, labeled sample data of a ball mill is difficult to obtain under variable working conditions. Aiming to realize the soft measurement of ball mill load under variable working conditions, a joint discriminative high-order moment alignment network (JDMAN) is proposed, based on the deep transfer learning in this paper. With this method, discriminative features were learned through jointly training labeled samples belonging to the source domain and unlabeled samples belonging to the target domain. Simultaneously, the features learned by a deep convolution network were aligned and clustered through central moment discrepancy and center loss to accomplish transfer. The comparison with other transfer methods indicates that the proposed JDMAN can effectively promote the accuracy of soft measurement under variable working conditions.

Rolling bearings are a widely used component in rotating machinery, and deterioration of their performance may lead to machine breakdown and accidents. However, because rolling bearings often work under time-varying working conditions it is a great challenge to accurately separate fault features from working condition features. This paper presents a new neural network framework for variable speed fault diagnosis. The proposed neural network can eliminate the influence of condition information during training. Firstly, samples of the same category under different working conditions are distinguished. Secondly, a projection matrix is designed at the last layer of the neural network to project the features extracted through forward propagation to a hyperplane of the same latitude, so that the fault features are more obvious. Finally, a network model that can distinguish fault features from working conditions is obtained through back propagation training. In addition, we propose a multi-loss function, which can reduce the within-class distance and enlarge the between-class distance. The method is validated using bearing datasets under time-varying rotation speeds. The results show that our method has higher accuracy with a simple structure, and is a great improvement on the traditional method of bearing fault diagnosis.

The quantitative evaluation of bearing damage size is of great significance for accurately predicting the remaining useful life of bearing. Existing studies have shown that rolling elements will produce corresponding vibration features when they enter and leave localized defects, and these features are the premise of accurately estimating the defect size. This paper presents an accurate quantitative estimation method of bearing localized defect size based on vibration instantaneous energy analysis. The time interval of rolling element entry and exit the damaged area is determined by analysing the instantaneous energy of fault impulse. The effectiveness of the proposed method is verified by the experiments of different defect sizes of the outer race and inner race at various speeds. Besides, the proposed method has higher accuracy compared with the conventional methods, which provides a new idea for fast and convenient estimation of the ball bearing localized defect size.

The bearing is the core component of mechanical equipment, and attention has been paid to its health monitoring and fault diagnosis. Bearing fault diagnosis technology based on deep learning has been widely developed because of its powerful feature learning and fault classification ability. However, the traditional deep learning-based bearing fault diagnosis methods fail in mining the relationship between signals explicitly, which is beneficial to fault classification. Therefore, this paper proposes a new method based on a multi-head graph attention network (MHGAT) for bearing fault diagnosis. Firstly, it employs dynamic time warping to transform the original vibration signals into graph data with topological structure, so as to exploit the intrinsic structural information of the independent samples. Next, the graph data is input into the MHGAT, and the weights of neighbor nodes are learned automatically. Then, the MHGAT extracts the discriminative features from different scales and aggregates them into an enhanced, new feature representation of graph nodes through the multi-head attention mechanism. Finally, the enhanced, new features are fed into the SoftMax classifier for bearing fault diagnosis. The effectiveness of the proposed method is examined by two bearing datasets. The superiority of the proposed method is verified by comparison to traditional deep learning diagnosis models.

In this work, a novel bearing fault identification scheme making use of deep learning has been proposed. Initially, the raw vibration signal is passed through a time-varying filter based empirical mode decomposition (TVF-EMD) to obtain different modes. Filter parameters of TVF-EMD are optimized by a newly developed optimization algorithm i.e. ameliorated African vulture optimization algorithm The Kernel estimate for mutual information has been considered as the fitness index for the developed optimization algorithm. The mode having the least value of fitness index is known as a prominent mode from which sensitive features representing different bearing conditions are extracted. These extracted features help in preparing the data matrix which is further utilised to build fuzzy-based classification models. The results obtained revealed that the linguistic hedge neuro-fuzzy classifier obtained maximum performance with the least computational time. The comparison of the developed method has also been done with other classification models viz., KNN, SVM, ELM and random forest that revealed the superiority of the developed method.

In recent years, data-driven intelligent diagnosis methods have been widely applied in the field of bearing fault diagnosis. However, these methods involve some expert experience and knowledge, and cannot accurately mine bearing fault characteristics under different loads. To solve this problem, this paper proposes a first-order differential filtering spectrum division (FDFSD) method and an information fusion multi-scale network (IFMSNet) to realize bearing fault diagnosis under different working conditions. First of all, the proposed spectrum division method based on the first-order differential filtering, the first-order differential processing of time domain signals, and the introduction of triangular filter, reclassify the spectrum features, highlight feature information, can accurately extract bearing fault features. Secondly, a new multi-scale network model of information fusion is constructed in this paper. Convolution kernels of different sizes are used to extract fault features of bearings of different scales, and information fusion is carried out to identify bearing working conditions and realize intelligent diagnosis of bearings under different loads. Finally, in order to verify the effectiveness and accuracy of the proposed method, it is verified on a variety of bearing experimental data sets. The results show that the average prediction accuracy of the proposed method is 99.11% and 97.74%, respectively. Compared with the proposed three single-scale network, K-nearest neighbor, Naive Bayes, support vector machine and random forest methods, the proposed method has more advantages in bearing intelligent diagnosis under different loads.

The precision measurement of blades has always been a complex problem in free-form surface inspection. For contact inspection of blades by coordinate measuring machine (CMM), the probe radius compensation has a critical influence on the measurement accuracy. In this study, an iterative approach based on the normal vector approximation is proposed to measure the blade profile curve, which aims to reduce the measurement error caused by the conventional method relying on a computer aided design model for compensation. The proposed method combines the local reconstruction of blade surface and CMM measurement software. It first takes the measured results of conventional method as a base, then the resulting corrected points of several sectional profiles are fitted to a B-spline surface to provide an alternative nominal model for CMM software, leading to the next same measurement task. Thus, the measurement-fitting process iterates, in which the current nominal model in CMM software is updated by the reconstructed model of foregoing iteration, until a preset condition is reached. Consequently, the proposed method is implemented on a four-axis CMM, and two blade samples are designed to verify the approach's effectiveness. Experiments show that our approach is superior to the conventional one in blade profile measurement accuracy.

In recent years, increasing numbers of deep learning methods for fault diagnosis of rolling element bearings (REBS) have been proposed. However, in industry, the scarcity of available data to monitor the health condition of REBS leads to low recognition accuracy of the trained intelligent diagnostic models. To solve this problem, we propose a simulation-data-driven subdomain adaptation adversarial transfer learning (TL) network (SAATLN). Firstly, a defect vibration model is introduced to simulate vibration signals of different types of REBS faults. And the real signal and simulated signal are used as the target domain and source domain of the TL fault diagnosis methods, respectively. Secondly, SAATLN uses the designed residual squeeze-and-excitation (Re-SE) blocks to extract transfer features between different domains. Meanwhile, it combines adversarial learning and subdomain adaptation to adapt the marginal distribution and conditional distribution discrepancies of high-level features. Also, the local maximum mean discrepancy is introduced as the subdomain adaptation metric criterion. Finally, different transfer tasks are performed on the artificially damaged and run-to-failure REBS data sets. The results demonstrate the effectiveness and superiority of the SAATLN in the simulation-data-driven REBS fault diagnosis.

Vibration measurement provides valuable information pertaining to the mechanical properties of a vibrating structure. However, anomalies caused by sensor faults, system malfunctions, and environmental effects impede the accurate measurement of vibration responses that are essential in determining these properties. Usually, responses with anomalies need to be discarded to attain meaningful interpretation of vibration. In this study, a neural network (NN) model able to classify vibration responses with missing data anomalies is proposed, using convolutional recurrent neural network (CRNN) as the fundamental component for spatio-temporal feature extraction. The proposed model was employed using contact measurement in carbon fiber reinforced plastic plate to identify partially missing responses of different load locations. The integrity destruction algorithm is introduced to simulate responses with missing data anomalies. The proposed model utilized a multi-input structure to reduce the computation cost of the training process. Two CRNNs were evaluated and the results showed that the hybrid architecture of convolutional neural network and long short-term memory (LSTM) is a better choice for the proposed model compared to that of convolutional long short-term memory (ConvLSTM). The proposed model was evaluated by samples with missing data anomalies. The experimental results show that the proposed model has good per-class precision and recall (above 80%) for the classification of measurements with missing data anomalies.

Support matrix machines (SMMs) take a matrix as the modeled element and can fully mine the structural information of matrix samples. However, relying solely on a pair of parallel hyperplanes limits the performance of SMMs in classifying complex data. Therefore, this paper proposes an adaptive interactive deviation matrix machine (AIDMM). In the AIDMM, a sensitive margin parameter is introduced to construct two deviation hyperplanes, so that the parameter margin between the two deviation hyperplanes becomes flexible. Compared to the original fixed maximum-margin method, the parameter-margin AIDMM can better adjust the boundary of the deviation hyperplane according to the data, which contributes to improving insensitivity to noise and enhancing robustness. In addition, a multi-rank projection matrix is introduced to obtain a low-rank solution, which gives AIDMM a better fitting ability and avoids the problem of large training errors. Two roller bearing fault datasets are applied for experimental verification, and the experimental results show that AIDMM has excellent classification performance in roller bearing fault diagnosis.

Laser distance sensors are a widespread, fast and contactless approach for distance and surface topography measurements. Main characteristics of those sensors are given by resolution, measurement speed and sensor geometry. With decreasing sensor size, the alignment of the optical components in sensor setup becomes more challenging. The depth response of optical profilers is analyzed to obtain characteristic parameters and, thus, to value the alignment and the transfer behavior of those sensors. We present a novel miniaturized sensor setup comprising of confocal and interferometric confocal signals within one sensor in order to compare both principles simply by obscuring the reference arm by an absorber. Further, we introduce a theoretical signal modeling in order to analyze influences such as spatial coherence, Gaussian beam characteristics and tilted reflectors on depth response signals. In addition to this, we show that the coherent superposition significantly reduces the axial resolution due to the confocal effect in interferometric signals compared to simple confocal signals in measurement and simulation results. Finally, an appropriate fit function is presented, in order to figure out characteristic sensor parameters from the obtained depth response signal. In this context, a good agreement to simulated and measured signals is achieved.

Although traditional deep learning improves the accuracy of intelligent fault diagnosis, it suffers from a problem, which is that a change in working conditions may reduce the diagnostic accuracy. The reason for this phenomenon is that a change of working conditions influences the probability distributions. To solve this problem, domain adaptation is adopted to perform intelligent fault diagnosis. However, the design of regularization methods, such as maximum mean discrepancy (MMD), neglects the phenomenon of fault extension. Considering the property of fault extension, the paper sums up a concept named short-time consistency which means that 'during stable operation, a failure does not expand over a short time period.' Moreover, short-time consistent regularization is proposed to ensure that the output of the model meets the requirement for short-time consistency, and closed-set regularization is proposed to further solve the problem of 'types of label drop' when short-time consistent regularization is used. When the problem occurs, the number of predicted label types in the target domain is smaller than that in the source domain in the closed-set domain adaptation. Two types of regularization, namely entropy-based regularization and regularization based on the $L2$ norm, are easily adopted in the final loss function. The proposed method is verified by experiments.

Compared with signals collected by the single sensor, the collected multivariate signals contain more information to reflect the state of mechanical equipment, which has a positive effect on fault diagnosis. However, different acquisition channels and various operating conditions interfere with the extraction of fault features of rotating machinery. To solve this problem, taking rolling bearings as an example in this paper, a novel method is adopted to alleviate these interferences and combined with an improved extreme learning machine (ELM) to achieve intelligent fault diagnosis of rolling bearings under various operating conditions. First, adaptive projection intrinsically transformed multivariate empirical mode decomposition is used to decompose multivariate signals and obtain intrinsic mode functions of each channel to construct feature matrices. Then, nuisance attribute projection (NAP) is employed to alleviate the interference components in the feature matrix, which are originated from different channels and operating conditions. Finally, vectors belonging to the processed feature matrix as samples are input into the proposed weighted extreme learning machine (WELM) for intelligent fault classification. The weighted matrix of the WELM can compress the dimension of the sample and extract sensitive features, and the effectiveness of the proposed fault diagnosis model via the above methods is verified by experiments. Furthermore, comparative experiments show that the proposed fault diagnosis model has higher accuracy than the model combined with NAP and traditional single-hidden layer feedforward neural network or ELM. Therefore, the proposed fault diagnosis model may potentially aid experts on fault diagnosis of rotating machinery under various operating conditions.

A novel method for the lateral and impact force measurement of shock absorbers using a three-component column-type sensor is proposed in this study. The strain gauge configuration and the bridge connection were designed, and the outputs of each designed measuring bridge under both alignment and eccentric loading were theoretically analyzed. Simulations and experimental tests were conducted to validate the measuring bridge output of the theoretical analysis; the results from the theoretical analysis, simulation, and experiments agree well with one another. In accordance with the specifications put forth in the SAE J2570 standard, the designed sensor was calibrated to obtain its performance parameters and calibration coefficients, which were then applied to acquire the lateral and impact force of a shock absorber. The calibration results demonstrate that the maximum nonlinearity and hysteresis were 0.51% and 0.87% (full-scale), respectively, and the sensor had good linearity and weak coupling between dimensional characteristics.

A pulsed eddy current testing (PECT) method is proposed to measure the gap of non-ferromagnetic metal casing. Firstly, the detection sensitivity to the casing gap of the voltage signal induced respectively by time-harmonic excitation current and by pulsed excitation current are calculated theoretically and compared. Then the advantages of PECT method for gap measurement are discussed. Secondly, the difference curve of the time-domain induced voltage signal between two gaps near the nominal value is presented. And the value of the induced voltage signal corresponding to the peak time of the difference curve is extracted to serve as the signal characteristic. Furthermore, using a calibrated casing, the function relationship between the signal characteristic and the casing gap is derived by curve fitting. Finally, the casing gap can be determined rapidly through the induced voltage at the peak time. This paper provides a reliable non-destructive testing method for the casing gap measurement, and compared with the existing sinusoidal ECT method, the PECT method can significantly improve the signal-to-noise ratio and the detection sensitivity to the casing gap.

Periodic impulse features caused by damage to rotating mechanical components are often overwhelmed by redundant components, which seriously affect the fault detection and diagnosis of equipment. Therefore, the time-frequency sparse (TFS) strategy based on optimal flux atom (OFA) and scale lp approximation operator (lp-AO) is proposed to extract periodic fault features. The OFA is determined by short-time Fourier transform (STFT) and correlation analysis of the signals. The convolutional coefficients are obtained by one-dimensional convolutional denoising based on the OFA. The convolution coefficients retain the main timing features of the signal. The scale lp-AO sparse model extracts the main frequency features of the convolutional coefficients in the frequency domain. The solution of the lp-AO sparse model relies on the iterative reweighed least squares algorithm. The effectiveness of the TFS is demonstrated by the analysis of simulated and several experimental signals. The two methods of fast spectral kurtosis and l_q sparse model are used as comparisons. The TFS is demonstrated to be more effective for extracting periodic fault features.

The inertial navigation system (INS) has been commonly adopted for unmanned ground vehicles, but it needs other systems to correct for error divergence, such as the Global Navigation Satellite System (GNSS). However, GNSS failures are inevitable, and INS must navigate independently in such situations, meaning that navigation errors will diverge fast. To solve this problem, the vehicle model aided INS is proposed. In this paper, three commonly used vehicle models are analyzed, first to discuss their disadvantages, that they contain the vehicle kinematics model (VKM), the non-holonomic constraint of VKM, and the vehicle dynamics model (VDM). Against their disadvantages, the multi-layer vehicle model aided INS is proposed. The proposed method divides the INS error parameters into the sensor layer, system layer I, and system layer II. Then, the navigation information from the INS is fused with our developed VKM and VDM at the system layer and the sensor layer respectively. Additionaly, the design of the adaptive Kalman filter is based on the VKM error model, such that the estimations can be protected from observation errors when the VKM accuracy declines. Compared to the traditional vehicle model aided INS, the proposed method can improve the accuracy and robustness of the navigation system with acceptable computational complexity.

As a pivotal part of a machine driven system, the health states of rolling bearings usually determine the normal operation of a whole item of equipment. Consequently, it is important to make accurate and timely judgments as to the operating conditions of rolling bearings. In this paper, a synthesized diagnosis technology, including fault pre-judgment and identification for rolling bearings is proposed. In the first section, a threshold value is defined on the basis of the sensitivity of amplitude-aware permutation entropy (AAPE) to bearing faults. Whether the bearing has defects is judged is based on this value. If a defect exists, a feature extraction scheme combining the modified complete ensemble empirical mode decomposition with adaptive noise (MCEEMDAN) and the modified hierarchical AAPE (MHAAPE) is adopted, to fully mine the hidden state features. Firstly, the scheme uses MCEEMDAN, which benefits from a good time-frequency decomposition capability, to divide the signal of trouble into a group of intrinsic mode functions (IMFs). Second, the MHAAPE of each IMF component is computed to form the candidate state features. Then, multi cluster feature selection is employed to compress the high-dimensional fault features to form the low-dimensional sensitive feature vectors required for subsequent classification. Finally, the sensitive feature vectors are input into a random forest classifier for training and classification, so as to ascertain the different fault types and severity. In addition, different contrastive methods are tested based on experimental data. The experiment results indicate that, compared to contrastive methods, the proposed scheme enjoys better performance, which can effectively judge whether the bearing is healthy and accurately identify different fault states in bearings.

Random impact interference has always been an important subject in fault diagnosis. In the fast kurtogram (FK), kurtosis is a sparsity index used to locate optimal resonance frequency bands. However, the shortcomings of the kurtosis measure are a tendency to select the random impact with a large amplitude and a relatively weak sparse measure ability, which reduces the accuracy and anti-interference of the FK method. To overcome the effects of random impact interference, this paper proposes a novel bearing fault diagnosis method named the fast nonlinear Hoyergram (FNH), in which the nonlinear Hoyer index is employed to replace kurtosis to improve the fault representation capability under random impulse interference. Firstly, Z-score normalization and a generalized nonlinear sigmoid activation function are used for signal preprocessing, and the scale distribution of the signal is changed to weaken the impact of random impact interference. Secondly, the Hoyer index, which can be viewed as the normalized form of the ${l_2}/{l_1}$ norm, is used to replace kurtosis to improve the sparsity measure capability. Thirdly, in the Hoyergram, the frequency band with the largest Hoyer value can be chosen as the best resonance frequency band for the squared envelope analysis. Finally, the proposed FNH is compared with the FK method through simulation and experimental signals, and the effectiveness of the proposed method is verified.

Inspired by the lateral line system of fish, an artificial lateral line system is proposed for underwater target detection. The dipole is treated as a standard and simplified target. In previous studies, most researchers focused on the dipole at a fixed position and the trajectory tracking of a moving dipole was barely considered. In this paper, a new trajectory tracking method for a moving dipole is proposed. First, based on the instant pressure amplitude and loss function, the dipole trajectory is tracked by particle swarm optimization (PSO). Then, the PSO-tracked trajectory is optimized by using recursive filters such as a Kalman filter (KF) and a particle filter (PF) to reduce the tracking error. The experiment result showed that when the trajectory of the dipole was rectangular, the target tracking accuracy of PSO was competitive compared with the Gauss–Newton method. The mean error distance (MED) of PSO was 12.51 mm. The PF showed better optimization performance than the KF in this study, and the corresponding MED of the PF was 7.064 mm. The main factor that caused tracking errors was pressure mismatch. In the simulation, when pressure mismatch was not considered, the performance of the proposed dipole tracking method was highly improved.

Studies show that to monitor the status of a rotating device accurately, vibration and rotation should be monitored simultaneously. In this regard and based on the effects of triboelectricity and electrostatic induction, a multifunctional sensor (MS) with a rotating-sleeve structure and two independent sensitive units is proposed in the present study. Nylon fabric and polytetrafluoroethylene film are used in the sensitive units, and the bounce of the copper ball on the sandwich elastic sensitive layer is used to determine the vibration level. Then theoretical analysis and finite element simulations are carried out to study the charge transfer characteristics of the sensor. It is found that the MS can truly reflect the vibration and rotation characteristics of the rotating device. Moreover, experiments are carried out and the obtained results show that the detection range of the sensor and its sensitivity are 0–36 m s⁻², and 40.38 mV s² m⁻¹, respectively. It is concluded that the proposed MS can accurately identify the vibration frequency. The obtained results demonstrate that at rotation speeds of up to 2000 rpm, this MS can be used to measure the direction of rotation, speed, and angular acceleration with a relative error of less than 0.95%. Considering the superior characteristics of the proposed MS, it can be used to monitor the motion state of objects in real-time and has potential application prospects in the field of internet of things.

Here, we report a vibration sensor based on a single-electrode mode triboelectric nanogenerator (TENG). The main objective of this study is to develop a vibration sensor (architecture) that can be employed in any application with minor design changes to meet individual objectives. Hence, a cantilever-based vibration system is selected, which offers optimum design control in fine-tuning the sensor to operate in the desired frequency spectrum. The cantilever's proof mass is suspended by isotropic linear elastic material constituting a scalable and tunable cantilever–mass system. The oscillations create contact separation between the triboelectric-active layers (i.e. fluorinated ethylene–propylene copolymer and screen-printed zinc oxide), which develop triboelectric waveforms. This voltage waveform is used for both sensing and powering mechanisms. At resonance, the device produces peak-to-peak voltage, short-circuit current, and power density of 25 V, 10 µA, and 1.38 W m⁻², respectively. To measure the influence of change in cantilever properties, we varied the number of cantilevers and evaluated the sensor performance. The sensor is reliable with >99% accuracy in a broad frequency range of 0–400 Hz. The sensor exhibits a maximum sensitivity of 14 V g⁻¹ and can charge a 1 µF capacitor to 2.75 V in <150 s. The sensor is further tested on a lab-scale vacuum pump with known (induced) faults to estimate the sensor's competence in detecting the machinery faults. Considering the market acceptability, the sensor is developed with established manufacturing techniques such as screen-printing, and laser cutting. This study hopes to bridge the lab-to-market gap for TENG-based (vibration) sensors.

Alignment of superconducting cavities is one of the important issues for the China Spallation Neutron Source Phase II (CSNS II) linac. In order to obtain the cavity displacement in the process of cooling down to the liquid helium temperature, a laser-based Poisson Spot Monitor (PSM) system was newly proposed and a verification system in the laboratory was built. The PSM system uses the diffraction spot formed on the CMOS camera after a beam of parallel laser passes through a spherical target to monitor the position of the measured object, where the spherical target is fixed on the monitored object. The displacement during the cooling process of the cryomodule is replaced by the movement of the high-precision two-axis motorized translation stage. A spherical target is fixed on the translation stage as the monitoring object. A beam of parallel laser passes through the spherical monitoring target to form a Poisson spot image on a CMOS camera. The coordinates of the Poisson spot center are obtained through image processing. Through experiments, the PSM system obtained a high accuracy within 5 μm, which meets the displacement monitoring requirement of the CSNS II cryomodule components. The system is fairly simple and able to be constructed without highly specialized parts and can also be used in other high-precision alignment and monitoring fields.

The existing methods of contact respiration rate (RR) measurement can bring discomfort to the person being measured. However, RR is a human index that has to be monitored in clinical medicine. To overcome the limitations of contact measurement methods, a non-contact RR measurement method based on an infrared thermal camera is proposed. This is based on the phenomenon that human breathing causes periodic temperature changes around the nostrils. First, an infrared thermal camera is used to collect image sequences of the human face. And then, to track a region-of-interest (ROI) in moving image sequences, the You Only Look Once V3 deep learning method is used to track the nostril ROI from the recorded facial image sequences. The performance of the deep learning model is analyzed through experiments. The average temperature of the nostril ROI of the tracked image is calculated, and the temperature change value of the continuous image series can be obtained. The temperature change curve reflects the respiratory process. A Butterworth low-pass filter is used to filter the continuous temperature value to obtain robust respiratory signals, to improve the signal-noise ratio. Last, the respiratory rate is calculated by time-domain and frequency-domain methods. RR is calculated in different conditions, such as normal respiration, fast respiration, and slow respiration. Through the training, the training accuracy of the model reaches 97.9%. Experimental results show that the method can achieve high precision, and the maximum error is not more than 2%. Those indicate the proposed method can effectively measure the RR. Therefore, the proposed non-contact RR measurement method can be a useful reference for clinical RR measurement and other applications.

The eddy current flow meter (ECFM) has been used to measure velocities and temperatures of conductive flows, such as liquid metal flows in a nuclear fission reactor. The goal of this paper is to develop a finite element electromagnetic model that can characterize the ECFM sensor performance and validate this finite element model with detailed velocity measurements of a controlled, well-characterized moving conductive solid rod. Both measurements and modeling were performed for various parameters that are important for ECFM performance such as rod velocity, rod material, ECFM sensor coil length, number of sensor coils, applied alternating current (AC) current amplitude, and applied ECFM AC frequency. For all parametric scans, the measurement and modeling agree well in both magnitude and trend. The normalized root-mean-square error between measurement and modeling is less than 10% for all cases. These results suggest that electromagnetic modeling could eventually be used to cost-effectively design future ECFM sensors in arbitrary geometry for more challenging applications such as liquid metal nuclear fission reactors.

Phase fraction is one of important indexes to characterize multiphase flow. In order to measure each phase fraction of oil–gas–water three-phase flow, liquid level height is detected by time-of-flight—TOF of reflected ultrasound at gas–liquid interface, while oil phase fraction in reflection path is calculated according to the ultrasound attenuation. By studying interactions between multiphase flow and the ultrasound propagation in certain flow patterns, a prediction model for phase fraction measurement of three-phase flow is proposed based on ultrasound transmission attenuation and reflection TOF in the process of horizontal flow with actual phase distributions. Simulation and experimental results under conditions of oil–water two-phase structure with stratified gas in a horizontal pipe show that the proposed method and the established model can accurately detect gas–liquid interface, so that measure oil, gas, water phase fraction. The mechanism prediction model and the measurement device effectively solve the nonlinear response of the ultrasonic measurement parameter, so that can estimate phase fractions of liquids and gas in two-phase as well as three-phase flows simultaneously, which extends the measurement range and the applicable scope of ultrasonic technique to multiphase flow.

Magnetic resonance imaging (MRI) is a medical imaging technique which is widely used in clinical routine. Standard imaging methods lead to so-called contrast-weighted images. The contrast arises from several tissue-related parameters such as the relaxation times T₁ and T₂. The aim in quantitative MRI (qMRI) is an estimation of these quantitative parameters. Magnetic resonance fingerprinting (MRF) is a promising technique in qMRI that allows for the simultaneous determination of multiple tissue-related parameters within a short acquisition time. The conventional MRF method utilizes an approximate inverse Fourier transformation for the highly undersampled data in the Fourier domain, leading to aliasing errors in the reconstructed magnetization courses. Computationally expensive statistical MRF modeling approaches overcome this issue by modeling the data directly in the Fourier domain. However, this leads to a non-convex and large-scale optimization task that is challenging to solve and requires expensive calculations. We compare two recent approaches, namely the conventional MRF method and a statistical MRF modeling approach based on Bayesian statistics in terms of their accuracy, reliability and computational costs. The comparison is carried out for simulated data with known ground truth for different signal-to-noise ratios, in the presence of errors in the physical model, and for several Fourier domain sampling schemes. It is demonstrated that a residual analysis can help to decide if the conventional MRF method is sufficient or if the complex Bayesian Fourier domain modeling approach can lead to a significant improvement.

Coal petrography extraction is crucial for the accurate analysis of coal reaction characteristics in coal gasification, coal coking, and metal smelting. Nevertheless, automatic extraction remains a challenging task because of the grayscale overlap between exinite and background regions in coal photomicrographs. Inspired by the excellent performance of neural networks in the image segmentation field, this study proposes a reliable coal petrography extraction method that achieves precise segmentation of coal petrography from the background regions. This method uses a novel semantic segmentation model based on Unet, referred to as M2AR-Unet. To improve the efficiency of network learning, the proposed M2AR-Unet framework takes Unet as a baseline and further optimizes the network structure in four ways, namely, an improved residual block composed of four units, a mixed attention module containing multiple attention mechanisms, an edge feature enhancement strategy, and a multiscale feature extraction module composed of a feature pyramid and atrous spatial pyramid pooling module. Compared to current state-of-the-art segmentation network models, the proposed M2AR-Unet offers improved coal petrography extraction integrity and edge extraction.

This study proposes a near-infrared (NIR) imaging method for two-dimensional distributions of water vapour (WV) in the open air. A wavelength of 1872 nm, at which the dependence of the absorption coefficient on the WV concentration is the largest in the v₂ + v₃ band of water molecules, was employed for imaging. The measurement system was a star imaging system, comprising a laser light source and an NIR camera. A measurement space with 0.08 m diameter and 0.30 m length was ensured to obtain the 2D absorbance images of moist air emitted from a gas nozzle and the breath air exhaled from a human subject with and without a mask. The images clearly showed the distributions of WV that flowed around a pipe and impinged on a plate. The exhaled air was clearly visualised, and the difference in air permeability between the masks was demonstrated. These results indicate the potential of the NIR imaging method for various applications that require WV flow visualisation, WV source identification, and evaporation process observation.

Sparse features and repetitive textures are frequently presented in microelectronic microscopic images. Therefore, it is challenging for image stitching to meet the requirements of high-speed precision manufacturing. A novel image stitching method for tiled images is proposed to generate panoramic images of microelectronics quickly and accurately. According to the preset scan trajectory, grids were established between adjacent images for feature matching. The clustering algorithm was used to screen reasonable and multiple sets of registrations. Then, all registrations were used as connecting edges, and images were used as nodes, to create a multigraph. The unique registration in multigraph was solved by a non-linear minimization problem with linear constraints. Finally, image transformations were computed in global optimization for rendering panoramic images via image warping. The experimental results show that the proposed method improves the stability and efficiency of image stitching, furthermore, it maintains an equivalent level of precision as the Fiji and microscopy image stitching tool methods.

A novel polarimetric imaging system is proposed, which is based on one complementary metal-oxide-semiconductor (CMOS) camera and double-layer prisms pair. Four polarizers, whose axes were set at different angles and affixed to the four prisms. The CMOS camera is used to simultaneously record the intensity of the light passing through the prisms with four specific polarization information. An algorithm is adopted to align and correct parallax to generate accurate Stokes parameters. Indoor and outdoor experimental performances verify the effectiveness and high measurement accuracy of the proposed system.

A computationally-supported experimental technique is presented, to measure height of luminous flames, using convolution and density-based spatial clustering for image processing. The experimental setup employs a high-definition camera array to capture flame imagery from 0°, 45° and 90° visualization planes. The volumetric fuel flow was ranged from 350 to 1800 cc min⁻¹ and images of the resulting flame structure were captured and measured. Results show that output measurements are affected by both the volumetric fuel flow and the visualization plane. Whilst the former is evidently the most significant factor, the latter was found to be relevant, since several flame features, particularly, flame tilt and flickering, are only perceivable through specific visualization planes; attributed to uneven flame structure due non-homogeneous thermal stress. The experimental technique proposed yields an overall statistical tolerance of 1.29 cm and an expanded uncertainty of 0.609 cm (∼11.5%). From these results, the test is considered successful and the proposed experimental technique is deemed to be on par with the already existing ones. The utilization of spatial density clustering of image gridded data has only been tested for this implementation; being severely constrained by the sample size and density variability of the data. Consequently, care should be exercised. Nonetheless, this approach was found to inherently recognize flame front edge, and mitigate variations of pixel value due change in flame intensity, normalizing image processing, hence, it is proposed as a viable alternative for flame feature/structure visualization and estimation, and further research is encouraged.

Near-infrared (NIR) spectroscopy is a promising candidate for low-cost, nondestructive, and high-throughput mass quantification of microplastics in environmental samples. Widespread application of the technique is currently hampered mainly by the low sensitivity of NIR spectroscopy compared to thermoanalytical approaches commonly used for this type of analysis. This study shows how the application of NIR spectroscopy for mass quantification of microplastics can be extended to smaller analyte levels by combining it with a simple and rapid microplastic enrichment protocol. For this purpose, the widely used flotation of microplastics in a NaCl solution, accelerated by centrifugation, was chosen which allowed to remove up to 99% of the matrix at recovery rates of 83%–104%. The spectroscopic measurements took place directly on the stainless-steel filters used to collect the extracted particles to reduce sample handling to a minimum. Partial least squares regression models were used to identify and quantify the extracted microplastics in the mass range of 1–10 mg. The simple and fast extraction procedure was systematically optimized to meet the requirements for the quantification of microplastics from common polyethylene-, polypropylene-, and polystyrene-based packaging materials with a particle size <1 mm found in compost or soils with high natural organic matter content (>10% determined by loss on ignition). Microplastics could be detected in model samples at a mass fraction of 1 mg g⁻¹. The detectable microplastic mass fraction is about an order of magnitude lower compared to previous studies using NIR spectroscopy without additional enrichment. To emphasize the cost-effectiveness of the method, it was implemented using some of the cheapest and most compact NIR spectrometers available.

Bearing fault diagnosis is a critical component of the mechanical equipment monitoring system. In the complex and harsh environment in which bearings operate, the fault diagnosis approach of multi-source information fusion can extract fault features more stably and extensively than the traditional single-source fault diagnosis method. However, most existing multi-source fusion methods are in infancy, and there are a number of pressing issues to address, such as subjective elements having a significant impact, excessive data redundancy, fuzzy multi-source signal fusion strategy, and insufficient accuracy. As a result, a new multi-source fusion fault diagnosis method is proposed in this paper. First, the residual pyramid algorithm is utilized to fuse the acoustic and vibration signals of multiple spatial positions respectively and then form two fused acoustic and vibration signals. Second, two improved 2D-CNN are used to extract the fault features contained in the above two signals separately to form a multi-source fault feature set. Third, an AdaBoost algorithm with a dynamic deletion mechanism is designed to fuse multi-source fault feature sets and produce the fault diagnosis findings. Finally, six different experimental data sets are used to test the performance of the model. The results reveal that the model has better generalization, higher and more stable fault diagnostic accuracy, and stronger anti-interference capacity.

Accurate measurement of soil moisture parameters plays a vital role in promoting agricultural production. Nuclear magnetic resonance (NMR) measuring instruments have become increasingly compact, especially in low field applications. However, NMR soil moisture measurement instruments still have problems with portability. We report on a compact NMR measuring instrument that focuses on the circuit's design and construction. The instrument includes an inside-out sensor, NMR spectrometer, radio frequency power amplifier, low-noise preamplifier, and duplexer. They are integrated into a slim 3D-printed cylinder. The integrated NMR instrument is portable and user-friendly and can reduce the influence of nonideal factors, such as transmission line effects and electromagnetic interference. We conducted different soil moisture experiments with the prototype using Carr–Purcell–Meiboom–Gill pulse sequence to obtain the corresponding transverse relaxation time (T₂) decay curves. Then, the T₂ 1D spectrum distributions were obtained by inverse Laplace transformation. Finally, the results reveal that the area of long T₂ component decreases as the soil moisture content decreases.

This study proposes a system design scheme of fault detection, isolation, and accommodation algorithms for the actuator fault of a fixed-wing unmanned aerial vehicle (FW-UAV) by considering system nonlinearity, external disturbance, and multi-actuator faults. The fault diagnosis scheme consists of a comprehensive observer and a bank of fault isolation estimators designed based on the actuator of the FW-UAV. In the diagnosis module, the system model is transformed into two systems by introducing a transformation matrix, so that both the actuator faults and the system disturbance are separated. Then, the concept of equivalent output injection is applied to construct a comprehensive observer to detect the actuator fault and estimate the unknown system disturbance. In the isolation estimation module, a set of sliding mode observers (SMOs) is constructed to isolate the multi-actuator faults, which reveals the fault source precisely. The stability analysis of the proposed SMOs was derived from the solution of linear matrix inequalities. In the event of a fault, the fault estimation provided by the fault diagnosis solution is used to accommodate the fault effects, while maintaining good attitude control and position tracking performance of the FW-UAV. The effectiveness of the proposed scheme is verified by the simulation model of de Havilland DHC-2 'Beaver' aircraft in different fault cases.

The time-zero correction is an essential step in the data pre-processing of ground penetrating radar (GPR) measurements to obtain an accurate signal propagation time between transmitting and receiving antennas. For a novel custom GPR monitoring system with about 3000 antennas and corresponding transceiver structures placed around a soil sample (lysimeter), an in situ approach for the time-zero correction is required. In particular, unknown material properties between any pair of transmitting and receiving antennas prevent a conventional time-zero correction. We present and compare two calibration approaches, namely a pairwise and a mesh calibration, both utilizing the ability of the monitoring system to conduct reciprocal measurements between any pair of antennas. The pairwise calibration enables an individual calibration for any antenna pair, whereas the mesh calibration reduces the influence of the soil between antenna pairs compared to the pairwise calibration. The developed approach is verified by utilizing a mathematical model. Experimental results from a simplified setup show that the lysimeter filling has a negligible impact onto the calibration approach based on adjacent measurements for the mesh calibration. In addition, it is shown that a state of the art time-zero calibration can be used to measure the signal delays within the analog circuit of the measurement system with an accuracy of ±4 ps. The simulation results indicate that by using the developed concept, no prior air calibration between every possible antenna combination is necessary. Thus, this work provides a crucial contribution towards an automated in situ time-zero correction for 3D GPR monitoring systems with many antennas.