Abstract:The demand for fiber optic time and frequency synchronization technology is gradually developing in the direction of long-distance, multi-node and high-precision. White rabbit (WR) technology, as an important implementation of fiber optic time and frequency transfer, is an innovative extension of the precision time protocol (PTP). Sub-nanosecond time deviation and nanosecond synchronization uncertainty between master and slave clocks by introducing synchronous Ethernet and digital double-mixed frequency time difference (DDMTD) measurement techniques. It also demonstrates significant advantages in terms of standardization, generalization and convenient networking. This paper systematically explains the synchronization principle of WR technology and its performance advantages. And deeply analyzes the key factors affecting the system synchronization performance. Meanwhile, a comparison is made between the advantages and disadvantages of existing research methods. Finally, the current technical challenges and future directions of the technology are further discussed. The study shows that WR technology has a broad application prospect in distributed systems. But it still needs to make breakthroughs in several key technical areas to meet the demands of diverse application scenarios, including high-precision absolute calibration technology, temperature compensation technology for long-distance fiber transmission, noise accumulation suppression in cascade systems, and time-frequency transfer implementation based on wireless links.

Abstract:The performance of an analog-to-digital converter (ADC) determines the quality of the entire acquisition system. With the increase in sampling rate and bandwidth, nonlinear errors pose a greater threat than linear errors. This paper proposes an adaptive parameter estimation method based on sine wave fitting combined with the normalized least mean squares (NLMS) algorithm to calibrate frequency-dependent nonlinear errors in ADCs through digital post-calibration. The method first performs sine wave fitting on the single-tone signal collected by the ADC and then reconstructs the nonlinear error signal based on the fitting output and the Volterra series model. The NLMS algorithm is employed to adaptively estimate the parameters related to the Volterra series model. When the parameters converge, the deviation between the corrected ADC sampling sequence and the fitting output approaches zero. This method requires only a small number of sampling points to converge and involves no complex computations, resulting in low resource overhead. Simulations and experiments have validated the effectiveness of the proposed algorithm. In the simulation, the second harmonic and third harmonic components are attenuated by more than 20 and 15 dB, respectively. Furthermore, the overall nonlinear spurious components of the wideband multi-tone signal are reduced by more than 15 dB after calibration. The proposed method was validated on a hardware platform with a sampling rate of 20 GSPS and demonstrated effective calibration performance, resulting in an overall improvement in the spurious-free dynamic range (SFDR) of approximately 10 dB.

Abstract:Model-assisted probability of detection (MAPoD) and sensitivity analysis are important to quantify the detection capabilities of eddy current nondestructive testing (ECNDT) systems. Due to the propagation of uncertainties in the MAPoD and SA problems of eddy current NDT, the traditional methods which are based on experiment and physical simulation models require a lot of time and labor costs. To reduce these costs, in this paper, the particle swarm optimization (PSO) algorithm optimized support vector regression (SVR) model is proposed to replace the traditional experiments and physical simulation models to predict the response of eddy current NDT models, thereby accelerating the analysis of MAPoD and SA problems. In addition, to the novelty, this paper combines the hyperparameter optimization algorithms such as grid search, random search, simulated annealing algorithm and PSO with SVR to test the accuracy and efficiency of them for the optimization of key parameters, and verify the advantages of PSO-SVR over other optimization algorithms based SVR. Finally, the PSO-SVR model is applied to the ECNDT problem, and the uncertainties in length of the surface slot is studied in MAPoD and SA analysis. The results show that the proposed method not only ensures the accuracy, but also accelerates the study for the MAPoD and SA analysis of eddy current NDT systems. It also reduces the computational costs, which accounts for 3.5% and 0.06% of those of the pure physical model in average.

Abstract:To address the challenges of existing PCB defect detection models, which suffer from excessive parameters, high computational complexity, and limited deploy ability on industrial edge devices with constrained computing resources, we propose a lightweight defect detection algorithm based on STR-DETR. First, we construct a novel backbone network, G-StarNet, by integrating group convolution into the lightweight StarNet architecture. This modification significantly reduces model complexity while preserving multi-scale feature extraction capabilities. Second, within the adaptive feature interaction module, a statistical feature-based self-attention mechanism replaces the conventional multi-head self-attention, effectively lowering computational overhead. Third, the RetBlockC3 module is designed by combining the Manhattan self-attention mechanism and its decomposed form. By incorporating a distance-dependent attenuation strategy, this module prioritizes local feature representation and reduces computational complexity from quadratic to linear scaling. Finally, we introduce a new loss function, FSN Loss, which mitigates the adverse effects of shape/scale variations and imbalanced sample distributions on bounding box regression, thereby enhancing both localization and classification accuracy. Experimental results demonstrate that the improved algorithm achieves an mAP@0.5 of 96.7%. Compared with the baseline model, it reduces parameters by 50.8%, computational load by 55.4%, and increases detection speed by 23.7%. These findings validate the algorithm’s effectiveness in meeting the requirements of lightweight small-target detection tasks.

Abstract:An extensive investigation was conducted on the electronic system of the high-temperature guarded hot plate method thermal conductivity apparatus. It was observed that the electronic system had several issues, including large fluctuation range of temperature measurement data, prolonged measurement cycles, and inaccurate calculation of heating power, all of which negatively impacted the measurement repeatability of the apparatus. By designing a new generation of electronic systems for the high-temperature guarded hot plate method thermal conductivity apparatus, the thermocouple cold junction compensation temperature control structure and the heating circuit sampling resistor heat dissipation structure were constructed in the design of the new generation electronic system, which reduces the fluctuation range of temperature measurement data and improves the accuracy of heating power calculation. Moreover, the temperature measurement circuit was reconstruct by using dual temperature conversion chips to increase the conversion frequency of temperature measurement. The experimental results demonstrate that the measurement repeatability of the high-temperature guarded hot plate method thermal conductivity apparatus, after the electronic system enhancement, has significantly improved from ±3.67%~8.93% to ±2.69%~4.34% within the temperature range of 200 ℃~600 ℃. The new generation of electronic systems has markedly enhanced the measurement reliability of the apparatus, providing more reliable data support for the measurement of insulation materials at high temperatures. It is of great value in practical applications and is expected to promote the in-depth development of research on high-temperature insulation materials.

Abstract:In response to the problem that the traditional capacitive road covering detection method has weak identification ability and low discrimination at low temperatures，the dielectric loss and relaxation phenomena of capacitors and covers are studied based on the series equivalent model of the capacitor and starting from the principle of dielectric polarization. The effects of covers and detection frequency on the equivalent series capacitance and equivalent series resistance of the capacitor was analyzed and a capacitor high-pass filter circuit was designed and built. By comparing the amplitude attenuation ratio and phase difference between the output and input signals of the high pass filtering circuit, the changes in equivalent series capacitance and equivalent series resistance of capacitors can be indirectly measured to achieve the identification of the type of covering material. The sample was tested in the incubator. The experimental results show that when the temperature is between 10 ℃ and 60 ℃, the phase difference during drying is greater than 30°, and the decay ratio is less than 0.8. The phase difference is less than 10°, and the decay ratio is close to 1. When the temperature is between -30 ℃ and 0 ℃, the decay ratio of dry, freezing and snow cover is crossed. The phase difference varies slowly with temperature, with an average phase difference of more than 40° when dry, less than 30° when frozen, and somewhere in between when covered with snow. The neural network classification model is constructed by using temperature, phase difference and decay ratio, which is deployed to the single chip computer and measured. Measured data show that the method achieves an accuracy of 95% in distinguishing between dry and stagnant water between 0 ℃ and 60 ℃, and the accuracy of distinguishing between dry, frozen, and snow cover is about 83% in the range of -30℃ and 0℃, which can meet the needs of road covering detection.

Abstract:Ultrasonic welding technology is widely utilized in industrial manufacturing, however, factors such as welding parameters, equipment conditions, and operational techniques often lead to diverse welding defects. To enhance welding efficiency, this study proposes a lightweight deep-learning-based network, ultrasonic welding defect detection network, for ultrasonic welding surface defect detection. First, to address the limitations of conventional convolutional networks, which are often insensitive to fine details and prone to losing critical small-scale defect features in welding defect detection, this study introduces a novel stepwise attention convolution module. The SA-Conv architecture enhances the model’s ability to perceive defect features while reducing computational overhead. Second, to tackle the challenge of extracting complex welding defect features, this study designs a defect feature extraction network incorporating a deformable convolutional network module and a welding defect feature extraction module based on deformable convolution and SA-Conv. This network significantly improves defect representation in complex backgrounds, enabling the effective extraction of welding defect features with varying shapes and intricate characteristics. Finally, quantitative and qualitative experimental analyses demonstrate that UWD-Net achieves superior detection performance on both a self-constructed welding defect dataset and the publicly available NEU-DET dataset. On the self-constructed dataset, UWD-Net achieves an F1-score of 0.952 and a mAP@0.5 of 93.6%, while on the NEU-DET dataset, it attains an F1-score of 0.710 and a mAP@0.5 of 78.6%, outperforming other benchmark algorithms. Furthermore, UWD-Net has a lightweight model size of only 1.818×106 parameters and achieves an FPS of 145.80, effectively balancing detection accuracy and inference speed. These characteristics make UWD-Net well-suited for real-time defect detection and deployment in industrial applications.

Abstract:Temperature has a certain impact on the magnetic properties of soft magnetic materials, and temperature rise parameters are crucial considerations in high-frequency transformer design optimization. However, in magnetic characteristics test of soft magnetic materials at different temperatures, the temperature variation time relies on subjective judgment of experimenters. If the heating time is insufficient, the testing accuracy is relatively low, while excessive heating results in low efficiency. Therefore, a self-heating magnetic property measurement method for magnetic rings was proposed, aiming to ensure the measurement accuracy while improving efficiency. Nanocrystalline materials, preferred materials for high-frequency transformers, are used as the study object in this paper, aiming to investigate factors influencing temperature rises of nanocrystalline magnetic rings, and an excitation circuit is subsequently designed to rapidly heat the magnetic rings. Once the magnetic rings reach the target temperatures, they are demagnetized through low-frequency AC attenuating excitations to meet the testing conditions at designated temperatures. The magnetic property parameters are then obtained by using the excitation and measurement circuit. Finally, a traditional constant temperature box of nanocrystalline magnetic rings and a self-heating magnetocaloric characteristic testing system are designed and built. The comparison between measurement results from the two methods validates the improved efficiency of the proposed method on the premise of ensuring the accuracy. By simulating actual temperature rises of magnetic materials, the proposed measurement method reduces the heating time by over 90%, and changes the temperature of the magnetic rings by a constant value, thus achieving advantages of precise, controlled, and rapid temperature changes. This method provides an experimental and data foundation for multi-physics coupling and design optimization in high-frequency transformers.

Abstract:To address the issue of poor performance of traditional image quality enhancement algorithms across different scenes, a novel image quality enhancement method based on dynamic adaptive optimization model is proposed to meet the diverse requirements of various scenes and improve the effectiveness of image quality enhancement. Firstly, a dynamic adaptive optimization model is constructed based on the atmospheric scattering characteristics of the enhanced image. And the objective function of the model is designed using image quality assessment metrics, PSNR and SSIM, to provide evaluation standards for image quality enhancement in different scenes. Based on this, a cooperative-competitive learning operator is designed and cooperative-competitive human learning optimization algorithm is proposed to calculate the optimal transmission threshold t0, filtering window size n, and weighting parameter ω. Then the optimal dynamic adaptive optimization model is constructed to achieve image quality enhancement in different scenes. Finally, image quality enhancement experiments are conducted using images from the SOTS benchmark test set and six real scene images. The proposed method is compared with three other methods, i.e. CLAHEMF, IDCPLT and DCP-PSO. Experimental results demonstrate that the proposed method outperforms the three comparison methods in terms of both subjective visual effects and objective evaluation metrics, thereby fully validating the effectiveness and feasibility of the proposed approach.

Abstract:Aiming at the current situation that the existing target detection method has low accuracy and the model is too large in the surface defect detection of photovoltaic array, it is difficult to apply to the lightweight UAV detection equipment. An improved lightweight YOLOv8 model is proposed. The use of CSPHet module ensures the ability to extract features while reducing model parameters and improving operating efficiency. The PSA attention mechanism is introduced to integrate the global information into the feature map, which improves the network’s ability to distinguish the target and background, and reduces the influence of noise on target location and classification. The CCFM neck structure is adopted, and the complexity of the model is reduced by adjusting the number of output channels of the model, so as to achieve a more efficient and lightweight network architecture. The global sensing module ACmix is added to enhance the global sensing ability of the model, reduce the interference of irrelevant information, and improve the robustness of the model. The experimental results show that the parameters of the improved YOLOv8 model are reduced by 37%, the calculation amount is reduced by 27%, and the detection accuracy mAP@0.5 is increased to 81.2%. The parameter quantity and calculation amount are significantly reduced, and the detection accuracy is improved while achieving lightweight. Compared with other models, it is more suitable for deployment on lightweight UAV equipment for target detection of surface defects of photovoltaic arrays.

Abstract:To address issues in existing defect detection for irregularly patterned transparent packaging bags—such as multi-scale anomalies, pattern interference, and missed detections, which are caused by low contrast and insufficient sensitivity—an improved YOLOv8s-CBW detection algorithm based on the YOLOv8s framework is proposed. In this algorithm, a coordinate attention (CA) mechanism is embedded into the C2f module of the YOLOv8s backbone network to enhance the model’s spatial feature localization and refined identification capabilities for low-contrast and minute defects. The original PANet structure is replaced with a bidirectional feature pyramid network (BiFPN) to optimize multi-scale feature fusion efficiency. Finally, a dynamic focusing WIoU-v3 loss function is introduced, replacing the traditional CIoU loss function, to improve bounding box regression accuracy for irregularly shaped defects and enhance the model’s overall generalization performance. Experimental results show that, compared to the baseline YOLOv8s model, YOLOv8s-CBW, with only a 0.11×106 increase in parameters and essentially unchanged GFLOPs, achieved an mAP@0.5 of 82.2% (an increase of 1.3%) and an mAP@0.5:0.95 of 49.3% (an increase of 7.1%) in defect detection tasks. Compared to mainstream models such as YOLOv5s and YOLOv6s, our algorithm improved mAP@0.5 by 2.3% and 10.6%, respectively, achieving superior detection accuracy while maintaining essentially the same GFLOPs. This demonstrates that the lightweight improved YOLOv8s-CBW can ensure efficiency and significantly enhance stability in detecting multi-scale defects, providing a reliable solution for automated quality inspection of packaging bags.

Abstract:With the widespread application of clustering algorithms in intelligent measurement systems, multi-source sensor data analysis, and embedded state recognition, ensuring fairness with respect to sensitive attributes while maintaining clustering quality has become a key challenge that limits their effectiveness in critical measurement tasks. To address this issue, we propose a population optimization combined with robust distance metric for fair K-means clustering method (PODM-Kmeans). The proposed method balances clustering quality and fairness by incorporating an enhanced Cuckoo Search algorithm to achieve a trade-off between global and local search capabilities during the initialization of cluster centers, thereby improving clustering stability. Furthermore, fairness constraints and cluster size constraints are effectively integrated into the iterative clustering objective function. A flexible weighted Euclidean norm is adopted as the distance metric to mitigate the negative impact of outliers, contributing to improved fairness. Extensive experiments conducted on five synthetic and five real-world datasets demonstrate the superior performance of PODM-Kmeans compared to existing methods. Notably, on the Adult, Bank, Census1990, and CreditCard datasets, PODM-Kmeans achieves a fairness ratio (FR) exceeding 0.95 while maintaining high clustering quality.

Abstract:Terahertz noise sources are critical tools for noise figure measurement and performance evaluation of high-frequency devices. Traditional solid-state noise sources based on electronics face challenges in achieving high excess noise ratio (ENR) and flat power spectral characteristics due to the bandwidth limitations of electronic components, restricting their application in higher frequency bands. To address this issue, this study developed a prototype terahertz noise source using photonic methods. By utilizing two beams of incoherent light for photo-mixing in a high-speed photodetector, the system generates terahertz noise with a frequency range of 140~220 GHz, a maximum ENR of 47 dB, and a flatness better than ±2.0 dB. The ENR can also be tuned by adjusting the optical power. Stability tests show that the prototype maintains an ENR stability of 0.35 dB over 12 hours of continuous operation and an output repeatability of 0.39 dB over 10 power cycles. Additionally, the noise figure of a mixer module was measured using the noise source, with the measurement uncertainty being less than 0.48 dB. The development of this photonic terahertz noise source has increased the maximum operating frequency of noise sources in China, providing essential testing instruments for the design and optimization of terahertz devices.

Abstract:In the study of pipeline magnetic leakage detection, intelligent recognition models often struggle due to the limited number and significant variability of defect samples. To address this, a data augmentation method based on an improved Generative Adversarial Network is proposed. A multi-class mixed estimation approach provides prior information to the generator, enhancing its random noise input. A multi-head attention mechanism is integrated into the generator to capture global features, improving the quality of generated samples. Additionally, a sample selection method based on variational autoencoder reconstruction error filters higher-quality generated samples, improving the training efficiency of the recognition model. Finally, selected generated and original samples are combined to form an augmented defect sample dataset. Classification methods are applied to classify the augmented leakage magnetic defect signals. Results show that under small sample conditions, the proposed method achieves an average recognition accuracy of 93%, outperforming similar methods.

Abstract:A smart medical measurement system based on LoRa Mesh was designed to address the need for synchronized monitoring of physiological parameters and environmental factors in chronic disease management. The system employs the RadioHead protocol stack to achieve self-organized multi-hop mesh communication and is deployed within an endocrinology ward, integrating environmental sensing with patient health monitoring. Data from wearable terminals and environmental nodes are transmitted via LoRa modules to a central gateway and uploaded to a cloud server for storage and visualization. By extending the LoRaMeshSim simulation platform, the system was tested in a hospital scenario with 74 nodes distributed across 18 rooms and corridors, under varying payload lengths (30, 90, and 150 bytes) and transmission rates (1 to 26 packets per hour). Simulation results showed that the delivery rate remained above 99.5% when the packet generation rate was 1 packet per hour, but dropped to approximately 83.2% at 26 packets per hour. Moreover, increasing the packet length significantly elevated collision occurrences, with 150-byte packets encountering approximately 2.1 times more collisions than 30-byte packets. The analysis demonstrates that the proposed system maintains good stability and scalability under moderate to low data loads, while signal collisions and transceiver contention become critical challenges at higher loads. Future work will focus on introducing adaptive spreading factor adjustments and intelligent routing algorithms to further enhance the system’s reliability and energy efficiency. In addition, a small-scale LoRa Mesh network was physically deployed in a laboratory setting to experimentally validate the system’s communication reliability and practical applicability.

Abstract:To address the critical challenges in surface character defect detection of consumer batteries, including dynamic defect localization, multi-scale adaptability, and fine-scale defect recognition, this paper proposes an innovative DDP-YOLOv8 framework. Firstly, to resolve the limitation of YOLOv8 in effectively adjusting feature map weights during feature extraction, we design a DCNv3-LKA attention module to achieve adaptive spatial weight adjustment through dynamic convolution and large-kernel attention fusion. Secondly, aiming to overcome the fixed sampling positions and poor multi-scale adaptability of YOLOv8’s neck network in character defect detection, we restructure the neck architecture by adopting a CCFM framework and propose a dynamic sampler (DS-CCFM module) incorporating dual-driven dynamic sampling mechanism. Finally, to mitigate the insufficient feature representation and information loss caused by standard convolution layers in YOLOv8’s detection head when handling small-scale battery characters, we introduce a P2 small-target detection layer and integrate multiple self-attention mechanisms from DynamicHead into the detection head (P2-DynamicHead module) to improves small defect recognition. Experimental results demonstrate that the DCNv3-LKA, DS-CCFM, and P2-DynamicHead modules achieve mean average precision (mAP) mAP@0.5 of 91.8%, 91.2%, and 92.4% respectively on the character defect dataset, representing improvements of 1.7%, 1.1%, and 2.3% over baseline YOLOv8n. DDP-YOLOv8 achieves a final mAP@0.5 of 94.0%, representing a 3.9% improvement over the baseline model YOLOv8n. With an FPS of 85.1, the model meets the requirements of high accuracy and real-time performance for character defect detection in large-scale customized battery production.

Abstract:The coordinate rotation digital computer (CORDIC) algorithm has the feature of simple hardware implementation. It has been widely applied in various fields, such as electronic measurement, radar detection, and image processing. High-radix and parallel CORDIC effectively reduce CORDIC iteration latency to meet the real-time requirements. However, both approaches introduce a variable scaling factor, increasing the computational complexity and results in additional resource consumption. In comparison, scaling-free (SF) CORDIC algorithm eliminates the variable scaling factor. However, the most existing SF-CORDIC algorithms still require improvements in resource consumption and latency performance, while maintaining acceptable accuracy and supporting a wide convergence range. Therefore, this article proposes a hybrid CORDIC algorithm and its computing architecture design, which combines the look-up table (LUT) and parallel SF iterations. A method is proposed to determine the angle boundary between the LUT and parallel SF iterations using approximate angles with fewer non-zero terms, which extends the convergence range supported by the parallel SF iterations; furthermore, a method is proposed to divide the parallel SF iterations into two-parallel and four-parallel SF iterations to balance the computational complexity of each iteration stage, ensuring the overall design performance. Specifically, the LUT is used to rapidly fold a large-angle input located in the range (-π/2,π/2) into the convergence range supported by the two-parallel SF iterations. Then, the two-parallel SF iterations are performed to bring the residual angle into the convergence range supported by the four-parallel SF iterations. Finally, the four-parallel SF iterations are performed and the CORDIC iteration results are output. The proposed design is implemented in Verilog hardware description language and validated on field-programmable gate array (FPGA). Experimental results demonstrate that, compared with the existing designs，the proposed design reduces resource consumption by 23.1% and latency by 22.1%, while maintaining comparable accuracy and convergence range.

Abstract:To address the challenge of detecting weak target signals on the ocean surface under strong sea clutter backgrounds, this study investigates the theory of chaotic phase space reconstruction and the improved Newton-Raphson optimization algorithm. A novel method for weak signal detection in chaotic backgrounds is proposed, based on an optimized bidirectional long short-term memory network (BiLSTM). The reconstructed phase space signal is used as the input to the BiLSTM network, with the length of the training data determined by the embedding dimension and delay time. The parameters of the BiLSTM model are optimized using the improved Newton-Raphson optimization algorithm, and the model is trained with an adaptive weighted error (AWE) loss function. Both approaches work together to enhance prediction accuracy, improve runtime speed, and reduce the detection threshold. A single-step prediction is performed using the BiLSTM model, and weak target signals are detected from strong chaotic background noise by analyzing the prediction errors. Simulation experiments were conducted using the Lorenz chaotic system as the chaotic background to detect superimposed weak signals. The results demonstrate that the proposed method effectively detects weak signals. Further validation was carried out using the IPIX radar dataset and sea surface detection data from Yantai, confirming the method’s robustness and effectiveness.

Abstract:To ensure the safe operation of Tokamak experiments, the reliability assessment of key vacuum acquisition equipment, specifically molecular pumps, is crucial. However, limited degradation data has resulted in low accuracy of existing predictive methods. To address this challenge, a degradation data generation method for molecular pumps based on time series generative adversarial networks (TGAN) has been proposed, aimed at augmenting the dataset through generated data to enhance the accuracy and reliability of predictive models. This method innovatively combines Transformer networks with TGAN and improves the quality of the generated data by incorporating Weibull distribution. Furthermore, long short-term memory networks are utilized for degradation prediction of the generated data. Experimental results demonstrate that TGAN-Transformer can effectively generate data that meets the needs of molecular pump degradation prediction, significantly enhancing prediction accuracy and reliability, thereby providing solid support for the reliability analysis and safe operation of molecular pumps. Through comparative experiments, TGAN Transformer has improved RMSE indicators by 51%, 48%, 36%, 40%, and 30% compared to GAN, DCGAN, RCGAN, VAE, and CVAE, respectively. On the MAE index, they increased by 52%, 49%, 38%, 42%, and 33% respectively, demonstrating their effectiveness in predicting molecular pump degradation. Future research may further optimize the structure of the generation network and explore more variants of generative adversarial networks to improve the diversity and authenticity of generated data, thereby further enhancing the accuracy and reliability of degradation predictions.

Abstract:The variational modal decomposition (VMD) method has a better modal decomposition effect in the denoising of ultra-low altitude magnetic anomaly signals, however, it needs to rely on the manual setting of the penalty factor and the modal decomposition parameters in practical detection, and the magnetic anomaly signals are weak and the environmental noise is complex. Aiming at the above problems, this paper proposes an improved particle swarm optimized variational modal decomposition (IPSO-VMD) combined wavelet threshold denoising method. Firstly, by introducing the adaptive inertia weights and learning factor strategy, and utilizing the arrangement entropy as the fitness function, the self-adaptation to the above parameters is realized. After that, the optimal parameter combination is used to decompose the signal, and wavelet threshold denoising is applied to the abnormal components. Finally, the signal is reconstructed and the denoised signal is obtained. The simulation experiment results show that the method improves the SNR by about 9.44 dB compared with other methods, and the correlation coefficient reaches about 0.74, obtaining a good denoising effect. The field experiments show that the magnetic anomaly location of the measured signal after denoising is obvious, which effectively reduces the interference of environmental noise on the signal and shows the potential of application in the exploration of ultralow altitude magnetic targets in the field.

Abstract:The surface shape of the train shaft is irregular after hot forge, and the shaft body may deform during the cooling process, which poses challenges and difficulties for the positioning of the axis point in the primary turnery process. The existing methods, such as two-point method, optical projection method, rotation axis method, etc., or only consider the center and local surface, ignoring the deformation of the axis and the irregular shape of the surface, have problems such as low efficiency, inability to manufacture, and large product loss and so on. This paper proposes a marking method for the optimal manufacturing of train axle in primary turnery to address the problems, which in existing methods. Firstly, obtaining a 3D point cloud of the axle by a scanner. Then, the point cloud is successively transformed into another coordinate system, cut into discrete slices, and calculated to getting the initial machining axis. Next, the spatial margin distribution of the product machining computer aided design (CAD) model in the axle point cloud is analyzed. Simultaneously, the gradient descent optimization strategy is used to adjust the machining axis position. Finally, the optimal marking point for rail axle to processing optimal manufacturing is calculated, and then marked on the axle by the laser marking machine. This method is implemented using the mixed encoding of C++and point cloud library (PCL), and has been validated on site by China railway rolling stock corporation (CRRC) for up to a month, and data statistics show an accuracy of over 98%, with an efficiency improvement of 3~6 times compared to the operator. This method improves the production efficiency of the primary turnery process for rail axles, reduces the scrap rate in the production process, and ensures the margin adequacy and rotational balance during the turning process.

Abstract:To address the issue of significant time delay estimation errors in pipeline leakage localization, which stem from the low signal-to-noise ratio (SNR) of detection signals and the existence of diverse noise interferences, a cascaded arctangent least mean square (LMS) adaptive time delay estimation method is proposed. First, the arctangent function is incorporated into the LMS adaptive filter to improve the filter’s robustness against non-Gaussian noise. Next, two channels of leakage signals are fed into the first stage adaptive filter to suppress correlated Gaussian noise. Subsequently, the two output signals from the first stage filter serve as the input and desired signals for the second stage filter to further eliminate noise. Finally, the time delay estimation is obtained by analyzing the weight coefficient curve of the second stage filter. In the simulation, under the influence of correlated Gaussian noise and non-Gaussian noise with three distinct distributions, when compared with the cross-correlation method, the arctangent LMS method, and the cascaded LMS method, the proposed method exhibits the optimal noise suppression performance, and the signal correlation peak is the most pronounced. As the SNR gradually declines, this method can attain superior time delay estimation accuracy at a lower SNR. Finally, the effectiveness and practicality of the proposed method are further validated through an actual pipeline leakage location experiment. Under the influence of noise, the method can precisely locate the leakage point, with an average relative location error of 2.31% and a standard deviation of 2.08%.

Abstract:The rapid development of the shipping industry has led to a significant increase in exhaust emissions from ships. Ship plume emissions are characterized by wide distribution and high mobility. These emissions are often hidden, uneven, and highly variable, making their regulation extremely challenging. In response to this, the present study designed and developed a high-precision, high spatiotemporal resolution UV imaging remote sensing system for the real-time, remote monitoring of SO2 emissions from ship exhausts. The system employs a three-channel design, utilizing dual-wavelength channels at 310 nm and 330 nm to eliminate interference and accurately capture SO2 signals, with a spectral channel used for cross-verification of accuracy. The monitoring system integrates the 2-IM sky background reconstruction method, a self-calibration technique, and an optical dilution effect correction algorithm, enabling the precise acquisition of optical thickness images and real-time inversion of SO2 concentrations. Additionally, through an emission rate inversion algorithm, the 2D SO2 concentration data are converted into intuitive emission rate information, further enhancing the practicality and interpretability of the monitoring data. Experimental results show that the self-calibration technique can fit calibration curves in real time with an error of only 2.35%. After optical dilution correction, the camera’s detection limit reaches 3.84 ppm·m at 623 m, and it still maintains a high sensitivity of 6.24 ppm·m at 1 932 m. These results fully demonstrate that the system meets the performance requirements for monitoring distant, low-concentration, mobile pollution sources. The development of this system not only provides robust technical support for the monitoring and control of marine pollutants but also aids in understanding the characteristics of ship emissions and the diffusion mechanisms of gaseous pollutants.

Abstract:The current infrared and visible light image fusion algorithms often fail to fully extract image features, resulting in the loss of detail information. In real-world scenarios, infrared and visible light images are typically unregistered, and existing registration algorithms still suffer from artifacts and biases. To address these issues, this paper proposes an adaptive feature enhancement multi-scale infrared and visible light image registration and fusion algorithm. First, multi-scale convolutional kernels and dense connections are used in the registration network to extract features at different scales and prevent information loss. Additionally, an ORB feature point detection algorithm and a designed feature enhancement module are introduced to fully extract features and adapt to complex environments. Secondly, a lighting enhancement module is designed by incorporating channel attention and self-learning parameters to improve the information expression of visible light images. Then, in the fusion network, adaptive multi-scale pooling convolutions are designed using different pooling strategies and variable convolutions to extract detail information at multiple scales. An EMA feature fusion module is designed to integrate local and global features. Finally, a flow consistency loss function is introduced to minimize registration errors. To better validate the practical applicability of the proposed method, an infrared and visible light image dataset is established. Comparative and ablation experiments are conducted on the public datasets TNO, Roadscene, and a self-constructed dataset. The experimental results show that, in terms of subjective evaluation, the registered images have minimal bias and no artifacts, while the fused images are clear and visible. On objective evaluation, it improves about 20%, 7%, 4%, 15%, and 8% on the metrics MSE, MI, NCC, SD, and EN compared to other algorithms. Additionally, target detection performance experiments on YOLOv8 show that the fusion results exhibit good detection performance.

Abstract:During the casting process of the high-voltage coil in dry-type transformers, significant deviations in the concentricity and verticality of the inner and outer molds can lead to an asymmetric coil structure, affecting the consistency of electrical parameters and potentially causing local overheating and electric field distortion. To address this issue, a detection and adjustment algorithm for mold concentricity and verticality in dry-type transformer coils is proposed. The algorithm utilizes high-precision laser displacement sensors to obtain the relative position of the mold within the three-dimensional framework and calculates the concentricity and verticality based on the measured position data. Subsequently, the gradient descent algorithm is employed to optimize the calculation and accurately determine the concentricity and verticality deviations. Upon completing the calculations, a microcontroller generates adjustment commands and drives push rods via stepper motors to precisely adjust the mold position. The system displays real-time detection data and integrates closed-loop feedback control to enhance adjustment accuracy and stability. Additionally, the system can automatically record detection data, supporting trend analysis over multiple measurements to optimize adjustment strategies. Experimental results demonstrate that the proposed algorithm operates stably and achieves high adjustment precision, maintaining a concentricity error within 2 mm and a verticality error of less than 1.5°. Compared with particle swarm optimization algorithm and genetic algorithm, this method has higher computational efficiency. Under the condition of a concentricity deviation of 20 mm, the optimization time only takes 11.2 s, which is suitable for real-time detection and online adjustment. It can effectively improve the manufacturing accuracy and consistency of dry variable coils, reduce manual intervention, and improve the automation level of production.

Abstract:With the advancement of electronic device fabrication processes and the reduction of chip operating voltages, the performance requirements for rail to rail operational amplifiers have become increasingly stringent, particularly in critical parameters such as offset voltage and slew rate. This paper presents a low-offset, high slew rate rail to rail op-amp design. By cascading a high-gain low-bandwidth amplifier with a low-gain high-bandwidth architecture, constant transconductance is maintained across the rail to rail common-mode voltage range through current distribution principles. The output stage utilizes a feedforward Class AB push-pull amplifier to achieve rail to rail output with enhanced driving capability. A dedicated slew rate enhancement circuit is implemented to address the insufficient output slew rate under large input signals, thereby improving transient response and extending operational bandwidth. Additionally, to mitigate offset caused by process variations, a digital fuse trimming technique is incorporated at the input stage for load calibration. Operational stability is ensured through nested Miller compensation. Post-layout simulation results demonstrate that under a 2.2~5.5 V supply voltage with 1 kΩ and 100 pF load conditions, the op-amp achieves a gain-bandwidth product of 10 MHz, an open-loop gain of 145 dB, a phase margin of 62°, a slew rate of 11 V/μs, and a maximum offset voltage of 70 μV. Compared to conventional rail-to-rail op-amp designs, this architecture effectively reduces offset voltage through trimming technology and significantly enhances slew rate via dedicated enhancement circuitry, enabling the proposed design to drive heavy loads with high precision under constrained power consumption while maintaining superior performance metrics.

Abstract:Intermittent motion equipment is a critical component of intelligent logistics systems, and its operational condition directly impacts the safety and reliability of the entire system. In view of the complexity and uncertainty inherent in the operation of bearings within intermittent motion equipment, and the challenges posed by difficulties in acquiring effective data and the scarcity of fault samples under complex operating conditions, which result in low accuracy in rolling bearing fault diagnosis, this study proposes a fault diagnosis method for intermittent motion equipment under complex operating conditions based on the SResNet network. Firstly, an intermittent operating condition recognition method is proposed to enhance the effectiveness of state data. Secondly, the one-dimensional state data, after preprocessing, is transformed into a two-dimensional time-frequency spectrum using continuous wavelet transform (CWT), thereby enriching the fault feature information. Finally, an improved ResNet50 network is developed by incorporating a self-attention module (SAM). The SAM enhances the network’s focus on important features, thereby improving the accuracy and stability of bearing fault diagnosis. To validate the fault diagnosis performance of the proposed method, experiments were conducted using a bearing fault state simulation dataset. The results demonstrate that, under complex operating conditions, the proposed method can accurately classify and identify bearing faults, achieving a classification accuracy of over 99%. Compared to traditional fault diagnosis methods, the proposed approach exhibits significant improvements in diagnostic performance and generalization capability.

Abstract:Electrical impedance tomography (EIT) is a non-destructive visual detection technology, with no radiation, real-time, portable, low cost and other advantages, currently widely used in industrial testing and medical monitoring. But EIT technology also has low resolution and other shortcomings, which also greatly limits the rapid development of EIT technology. In this paper, aiming at the problems of unclear number of internal targets and excessive artifacts in the reconstructed image due to the “soft field” effect and under characterization in the process of electrical impedance imaging, this paper proposes an eight-modal data fusion electrical impedance imaging optimization algorithm, according to the characteristics of the eight excitation models of each imaging, with the help of the correlation coefficient between the reconstructed image and the actual distribution, the weight matrix is fused with the measurement value matrix obtained in eight single modes. The matrix was then used by the Tikhonov regularization (TR) algorithm for imaging. The simulation results show that the algorithm can effectively improve the resolution of the reconstructed image of the Tikhonov regularization algorithm, and the correlation coefficient of the reconstructed image after fusion is increased by 19.86% on average, and the relative error is reduced by 28.89% on average. This shows that compared with the traditional imaging under eight single models, the algorithm proposed in this paper has improved the number, size and position accuracy of reconstructed image targets, which provides a new theoretical basis and technical reference for EIT technology in the application practice of medical and industry and other fields.