Abstract:To solve the problems of poor global path quality and the tendency of local paths to fall into local optimality when mobile robots plan in complex environments and dynamic obstacles, a fusion algorithm based on the ant colony algorithm and the artificial potential field method is proposed. Firstly, in view of the poor global search ability and slow convergence speed of the traditional ant colony algorithm, its search method is optimized, a new pheromone update rule is constructed, the revised heuristic information is introduced, and a path node optimization strategy is designed to improve its path quality and search efficiency. Secondly, the problem of unreachable target and local minimum in the traditional artificial potential field method is solved by adding the relative distance from the mobile robot to the target point into the repulsive potential field function and setting sub-target points. Finally, the improved ant colony algorithm and the improved artificial potential field method are integrated to improve the path planning performance of the fusion algorithm in complex dynamic and static environments. The parameter combination of the improved artificial potential field method is selected through simulation analysis. Compared with the traditional ant colony algorithm, the simulation results show that, the optimal path of the improved ant colony algorithm is shortened by 26. 23% , the turning points of the path are reduced by 60. 00% , and the search efficiency is improved by 73. 75% . The improved artificial potential field method effectively solves the limitations of the traditional artificial potential field method and improves its local obstacle avoidance capability. The fusion algorithm can plan a collision-free and smooth path while maintaining compliance with the global optimal path. In actual scenarios, the experimental results show that: the path planned by the improved ant colony algorithm is shorter than that of the existing traditional algorithm. In the Gazebo physical simulation platform, the fusion algorithm can effectively avoid static obstacles, verifying its theoretical feasibility.

Abstract:Dynamic path planning of surface boats during navigation is of great significance to ensuring navigation safety. In view of the multi-target dynamic path planning of surface ships, puts forward an improved IRRT ∗ algorithm, fully considering the unique kinematic constraints of surface ships and the size of its hull, obstacles, the two-way search mechanism, search at the starting point and set a grading strategy, significantly improve the efficiency of path planning, and can better solve the optimal efficiency of multi-target path planning. Second, the path search process is optimized by introducing dynamic KD tree for nearest neighbor search, and reconstructing the KD tree regularly reduces the retrieval depth of the query nodes to further improve the search efficiency. Finally, the design is actually considering the cost function of ship turning angle and energy consumption, fusion the idea of artificial potential field method, introducing the gravitational field and repulsion gain coefficient as a local obstacle avoidance strategy, and finally adopt the adaptive third-order B spline curve optimization path, improve the smoothness of the path and real-time obstacle avoidance ability of water boats. Through simulation experiments and actual offshore tests in the Python environment, the results demonstrate the advantages of the algorithm in computing time, path length, performance of collision avoidance, and number of path turns. The research results provide a new idea for efficient path planning in complex waters, and contribute to the development of autonomous navigation technology of surface ships.

Abstract:The coated pipe is widely used in the petroleum, the chemical industry, and other fields because of its advantages of insulation, high-pressure resistance and vibration resistance. To ensure the safe operation of the coated pipe during service, it is necessary to detect its flaws. However, the high polymer material with high sound absorption and attenuation characteristics is usually used as the coating layer of the coated pipe, which makes the traditional high-frequency ultrasonic guided wave detection methods have low signal-to-noise ratio problems. On the other hand, the low-frequency ultrasonic guided wave can travel a long enough distance, but it has the disadvantages of large detection blind areas and low resolution. For this reason, a high-sensitivity method based on nonlinear ultrasonic zero-frequency wave low attenuation is proposed to detect the flaws of coated pipe. Firstly, the generation mechanism of nonlinear ultrasonic zero-frequency wave propagating along the pipe structure was theoretically analyzed. Then, the propagation characteristics of nonlinear ultrasonic zero-frequency wave in free and coated pipe are investigated through finite element simulation and experimental verification. Finally, the notched defects in the coated pipe are successfully detected by using a nonlinear ultrasonic zerofrequency wave in the experiment. The results show that the nonlinear ultrasonic zero-frequency wave has a low attenuation characteristic, and the nonlinear ultrasonic zero-frequency wave can still accurately identify multiple defects in the coated pipe when the linear fundamental frequency ultrasonic wave has been completely attenuated. The nonlinear ultrasonic zero-frequency wave in the method has the characteristic of low attenuation because the carrier frequency is zero, and can be used in the nondestructive testing of the circular pipe structure with high attenuation of the coated layer.

Abstract:As one of the most widely used resonance demodulation methods in planetary bearing fault diagnosis technology, modulation signal bispectrum demodulates the fault feature information from the resonance band via inhibiting the background noise and interference frequencies. In addition, modulation signal bispectrum has the characteristics of preserving phase information, detecting secondary phase coupling signals, and removing Gaussian noise. Therefore, the ability to detect amplitude and phase modulation is crucial for fault feature extraction of modulation signal bispectrum. To address the challenge of analyzing non-Gaussian noise in modulation signal bispectrum, an autoregressive model filter based on non-Gaussian noise suppression is proposed to improve its performance in planetary bearing fault diagnosis and monitoring. Autoregressive model filters effectively capture key features of data series and are applied with superior performance in removing non-Gaussian noise. Therefore, the autoregressive model is considered as a pre-filter process unit to reduce non-Gaussian noise in the original signal to improve the accuracy of modulation signal bispectrum. The order of the autoregressive model is determined adaptively using an indicator called kurtosis, further improving the effectiveness of the autoregressive model. Finally, the non-Gaussian noise analysis model signal is processed using modulation signal bispectrum to remove Gaussian noise and decompose coupled modulation components, thereby accurately identifying the frequency components of planetary bearing faults. The simulation and experimental analysis results demonstrate that non-Gaussian model modulation signal bispectrum achieves higher accuracy in diagnosing planetary bearing fault characteristics than fast kurtogram and modulation signal bispectrum.

Abstract:Mining bearing fault characteristics in high-power variable-frequency wind turbines is challenging, and existing deep learning models suffer from poor interpretability. To address these issues, a new intelligent diagnosis framework of lightweight space-time information fusion model, named BSTA-Net, is developed to enhance bearing fault identification in practical engineering applications. Firstly, a bearing fault feature space-time information fusion module is designed, and a new bidirectional timing information feature fusion strategy is creatively developed. The strategy is cleverly applied to the proposed BSTA-Net framework to fully extract the finegrained features from the fault data, marking the first attempt to apply such an approach to wind turbine bearing condition monitoring. Secondly, the feature focusing module is introduced into the proposed framework for optimization, enabling it to effectively prioritize critical fault-related information while discarding irrelevant or noisy features. This ensures that the model maintains robust learning capabilities even under complex conditions such as alternating voltage shocks and variable loads. Finally, based on the same data set, the diagnostic performance of 8 methods such as BSTA-Net framework is compared from multiple dimensions, and the diagnostic results are compared with 7 methods such as BST-Net. The results show that the proposed framework exhibits superior superiority and strong generalization ability, providing a new idea for bearing fault identification. Furthermore, T-SNE and significance region detection technology are introduced into the BSTA-Net framework to explain the physical attribution of fault feature mining process, thereby improving the reliability of the framework in the decision-making process.

Abstract::This article proposes a variable step size LMS algorithm based on deep reinforcement learning to address the problem of the difficult balance between convergence speed and steady-state error in the fixed step size LMS algorithm, as well as the high dependence on initial parameter selection, heavy workload, and subjective defects of traditional variable step size algorithms. This algorithm has a low dependence on initial parameters and avoids the cumbersome parameter tuning process. Firstly, an algorithm model integrating deep reinforcement learning and adaptive filtering is constructed, which utilizes deep reinforcement learning agents to control the change of step size factors, replacing the nonlinear function used for step size adjustment in traditional variable step size algorithms, thereby avoiding the cumbersome experimental parameter tuning process and reducing the complexity of algorithm use. Secondly, the error-based state reward and step size-based action reward functions are proposed. Dynamic rewards and negative reward mechanisms are introduced, which effectively improves the convergence speed of the algorithm. In addition, a network architecture based on incomplete encoders is designed to improve the inference ability of reinforcement learning strategies. Through experimental verification, compared with other newer variable step size algorithms, the algorithm proposed in this article can quickly adjust to a reasonable step size value under different initial parameters and reduce the workload of experimental parameter tuning, obtaining faster convergence speed and smaller steady-state error. The trained network has been applied to practical fields, such as system identification, signal denoising, and filtering of water level signals at the closure gap, and has achieved good performance, further confirming the generalization ability and effectiveness of the algorithm.

Abstract:In urban canyon environments, the multi-path effect and non-line-of-sight phenomenon significantly affect the positioning accuracy of GNSS, which in turn impacts the positioning performance of the INS / GNSS integrated navigation system. Traditional INS / GNSS integrated systems, however, struggle to accurately determine the rapidly changing GNSS measurement noise in such environments. To improve the robustness and estimation accuracy of the integrated navigation system, this paper proposes a robust factor graph optimization algorithm based on adaptive MCMC. The main issue addressed is the inaccuracy of measurement noise covariance in traditional factor graph optimization, which reduces state estimation accuracy. First, adaptive MCMC is introduced into the factor graph optimization framework, incorporating both prior and posterior stages. In the prior stage, the MCMC algorithm transforms posterior probability sampling into the product of prior and likelihood probability sampling, with an adaptive strategy enhancing sampling efficiency to obtain the posterior sample set. In the posterior stage, KL divergence minimizes the difference between the approximate and true posterior, allowing for accurate estimation of GNSS time-varying measurement noise covariance. Additionally, an innovation chi-square detection algorithm is used to detect and eliminate outliers by constructing hypothesis test statistics and identifying abnormal boundary values. The proposed method effectively estimates GNSS time-varying measurement noise while reducing outlier interference. Simulation and field tests of the INS / GNSS integrated navigation system show that the proposed method reduces horizontal positioning root mean square error by 20. 4% , 11. 9% , and 71. 6% , 25. 2% respectively compared to the standard factor graph optimization and the robust adaptive factor graph optimization algorithms based on variational Bayesian, respectively. The method also demonstrates improved robustness.

Abstract:Utilizing mobile sensors to locate chemical gas sources in the air can be applied to security searches, disaster relief, and building environments. This study investigates the problem of gas source localization using mobile robots in indoor environments and proposes a time-weighted maximum likelihood estimation algorithm ( TWMLE). Based on a sampling time-weighting mechanism, the algorithm utilizes the observation samples that contain gas concentration, wind speed, direction, and its relative localization to iteratively estimate and approach the position of local plume source, accommodating the time-varying gas distributions and airflows in dynamic turbulent environments. Meanwhile, this study employs a local sensing window to constrain the feasible solution space of the estimated source location to ensure the feasibility of the estimation results, achieving short-term estimation of local plumes in unknown environments and effectively enhancing estimation stability. Additionally, this study weights average the multiple estimation results based on the gas detection condition, effectively enhancing the ability to search upwind when gas is detected and the ability to quickly rediscover the plume when gas is missed. The experiments are implemented to evaluate the proposed method in four indoor environments with different airflow conditions and obstacles, as well as in a real environment. The proposed TWMLE algorithm outperforms both the infotaxis algorithm and the surge-cast algorithm in terms of success rate and search performance. In the real environment, the success rate of the TWMLE algorithm reaches 90. 0% , which is higher than the 80. 0% of the infotaxis algorithm and the 60. 0% of the surge-cast algorithm. The results show that the proposed TWMLE algorithm can effectively locate the gas source in complex indoor environments

Abstract:Aviation composite materials, as the key technology of aircraft lightweighting, are widely used in the field of aviation. However, under the new situation of large-scale, complex and intelligent development, there is an urgent need for breakthroughs in noncontact multi-damage rapid detection technology for composite materials. Based on this, this paper proposes a fast multi-damage imaging method for composite materials based on Hilbert marginal spectrum and air-coupled ultrasonic guided wave composite path cross-imaging. First, the propagation characteristics of the guided waves in the composite plate are analyzed through numerical simulation to determine the optimal incidence angle of the ultrasonic probe and the propagation modes of the guided waves in the composite plate; second, in order to solve the problem of artifacts that may easily appear in the multi-damage imaging, the air-coupled ultrasonic guided waves are used for scanning in double orthogonal directions, and the signals of the guided waves are captured when they propagate along the direction of composite fiber layups at 0°,45°,90°and 135°; Then, the fully integrated empirical mode decomposition and adaptive noise algorithm combined with energy entropy are used to optimize the eigenmode function of the original signal in order to extract the main signal components and reduce the influence of noise, and then the Hilbert marginal spectrum is used to compute the energy of the optimized signal and determine the damage factor; finally, the array of the damage factor is expanded into a matrix and divided into two groups according to the orthogonal angle, and the matrix of the damage factor within a group the damage factor array is expanded into a matrix and divided into two groups according to the orthogonal angle, and the damage factor matrices within the groups are added together and then multiplied between the groups to obtain the final damage image. The experimental results show that the maximum localization error of single damage is 2. 3 mm, and the localization error of multi-damage is 12. 5 mm, and the efficiency of C-scanning is improved by 168. 9% compared with point-by-point C-scanning. The method proposed in this paper can effectively improve the inspection efficiency and multi-damage localization accuracy of large-size specimens.

Abstract:The deep stacked autoencoder, as a prominent deep learning architecture, has been widely applied in fields such as data science and pattern recognition. However, existing deep stacked autoencoders focus on transforming the features of individual samples, often overlooking the inter-sample correlation, which can lead to suboptimal feature quality. To address this limitation, this paper introduces a novel deep stacked autoencoder architecture called the dual-stage joint projection envelope embedded stacked autoencoder. Unlike traditional deep stacked autoencoders, the proposed model transforms deep features based on the correlation information between samples, rather than focusing solely on the samples themselves. The model is composed of two primary components: the dual-stage joint projection envelope and the embedded stacked autoencoder. The dual-stage joint projection envelope utilizes a manifold sample-pair envelope module to extract local correlation information from the original samples and reconstruct the first layer of enveloped samples. A descending clustering module is then employed to capture global correlations and reconstruct the second layer of enveloped samples. Additionally, the dual-stage inter-consistency maintenance module enhances the representational power of the second-layer enveloped samples. Subsequently, two sets of deep features are extracted by training two embedded stacked autoencoders on these two layers of enveloped samples. The paper concludes with four sets of experiments: ablation studies, algorithm comparisons, parameter sensitivity analysis, and complexity analysis. Experimental results demonstrate that the deep features extracted by the proposed dual-stage joint projection envelope embedded stacked autoencoder exhibit both high quality and stability.

Abstract:The failure of the measurement tube is a key factor affecting and constraining the measurement accuracy, reliability, and safety of Coriolis flow meters. Wall attachment failure is one of the forms of measurement tube failure that can easily occur during the service life of Coriolis flow meters. The occurrence of wall attachment failure changes the physical rigidity of the Coriolis flow meter, leading to a shift in the calibration factor, which directly affects the measurement accuracy of mass flow and other fluid information. Furthermore, if wall attachment failure is not promptly warned, its progression may result in pipeline blockage and, in severe cases, even explosions, posing significant industrial safety risks. Therefore, detecting the service status of Coriolis flowmeters and identifying wall attachment faults in the measuring tube are urgent needs to improve the measurement accuracy, reliability, and safety of Coriolis flowmeters. To address this issue, this paper proposes a wall attachment fault detection method based on a combination of band-stop filters and sample entropy. Since vibration response signals typically contain multiple modal characteristic signals as well as interference signals, the combined band-stop filter effectively eliminates interference while preserving target modal characteristic signals. By calculating the sample entropy of modal characteristic signals under different states, the method fully utilizes its high sensitivity to dynamic signal changes. When a fault occurs, the complexity of the signal increases significantly, causing the sample entropy value to change accordingly, providing a reliable basis for fault diagnosis and evaluation. By comparing the sample entropy values of normal and faulty states, it is possible to quantitatively analyze the severity of faults, thereby achieving effective monitoring of the flow meter′s fault conditions. The experimental results indicate that this method can effectively identify the wall-hanging faults of the Coriolis flowmeter measurement tube, outperforming existing methods in fault detection accuracy and reliability.

Abstract:A hybrid TV-KF slurry level impedance imaging method combining total variation regularization and Kalman filtering is proposed to address the challenges of accurate detection of slurry level, low visualization level, and overly smooth imaging results in flotation processes. Firstly, a field model for measuring the flotation slurry level is constructed to obtain the boundary voltage of the field, and a flotation process error function based on total variation regularization is established to calculate the initial conductivity value of the slurry level. Secondly, based on the total variation regularization algorithm, the conductivity value is calculated as prior information for the prediction equation of the Kalman filter algorithm. The updated equation and prediction equation of the Kalman filter algorithm are iteratively updated using the measured voltage values over time. Finally, based on the proposed TV-KF impedance imaging algorithm, the conductivity distribution of the slurry level is solved to obtain accurate slurry level detection results. The simulation and experimental results show that the proposed algorithm has higher resolution and better edge characteristics of the interface between slurry and froth, providing more comprehensive and accurate slurry level information. In various slurry level simulation models, the Pearson correlation coefficient (PCC) exceeds 85% , while the image reconstruction error (IRE) is lower compared to other algorithms, resulting in better reconstruction performance. The maximum measurement error of the slurry level on the on-site experimental platform is less than 2. 4 cm, meeting the accurate detection requirements of the flotation industry′s on-site liquid level. Compared with the existing methods, the proposed algorithm exhibits stronger visualization of froth layer information, better adaptability to slurry variations, higher sensitivity to froth fluctuations, and sustained, stable measurement of slurry levels, making it highly valuable for practical flotation applications.

Abstract:Research on the electrical characteristics of rock and ore specimens is a fundamental aspect of geophysical exploration. Aiming at the problems of low efficiency and poor anti-interference ability in the measurement of electrical parameters of rock and ore specimens by frequency conversion method, a method for measuring the broadband electrical parameters of rock and ore specimens based on invert-repeated m-pesudo random binary sequence(IRmPRBS) is proposed and the corresponding measurement system is designed. The system adopts the method of coded constant current signal excitation, and uses the frequency distribution characteristics of the coded signal source along with the multi-frequency data processing method. By transmitting a set of coded rectangular wave currents, it efficiently measures the response information at multiple frequency points of rock and ore specimens. This paper introduces the signal characteristics of the inverse-repetitive m-sequence and presents the workflow of the multi-frequency impedance measurement algorithm. Additionally, it investigates the key factors affecting the performance of the Howland voltage-controlled current source through both theoretical analysis and simulation. The main sources of high-frequency impedance measurement errors under experimental conditions are clarified and the calibration scheme is given. Finally, the measurement system is built and the impedance measurement comparison experiment is carried out with the Zurich MFIA impedance analyzer. The results show that: Using a fifth-order signal improves measurement efficiency by 2. 3 times compared to the frequency conversion method, and the measurement of different frequency points can be realized by adjusting the signal coding parameters; The system measurement frequency range covers 1 mHz ~ 100 kHz. The maximum measurement error of the impedance modulus of the resistance-capacitance model is about 0. 5% , and the phase maximum error is about 23 mrad. The comparison of the results before and after calibration shows the effectiveness of the calibration scheme in high frequency measurement. The system can realize fast and accurate measurement of broadband impedance spectrum, which provides a broadband fast measurement method and technical support for the measurement of electrical parameters of rock and ore specimens. This research holds significant theoretical and engineering application value.

Abstract:The right heart perfusion is essential for the normal functioning of the pulmonary circulation. Assessing right heart perfusion provides critical insights into the health status of the pulmonary circulation. Pulmonary blood flow regulation refers to the adjustment of blood supply by the pulmonary artery during changes in right ventricular perfusion to maintain proper right ventricular-pulmonary artery coupling. This study aims to investigate cardiovascular blood flow regulation during right ventricular perfusion using bioimpedance technology and to assess changes in arterial elasticity and flow resistance parameters in the pulmonary circulation. Regional blood flow is assessed using bioimpedance technology, characterizing arterial elasticity and flow resistance parameters through the ratio of diastolic wave amplitude to the depth of the rebound wave and the rate of change of the maximum systolic wave. The discussion focuses on changes in pulmonary blood flow impedance parameters under acute exercise conditions, comparing them with changes in pulmonary blood flow impedance parameters at rest. Compared with the resting state, the changes in arterial elasticity and flow resistance under acute exercise conditions were 77. 78% and 11. 46% , respectively. Statistical analysis revealed significant differences in both parameters before and after exercise (P < 0. 001). Bioimpedance technology effectively detects changes in pulmonary arterial blood flow regulation during pulmonary circulation. By analyzing the pulmonary blood flow impedance waveform to quantify arterial vascular elasticity and flow resistance, facilitating the assessment of right ventricular-pulmonary artery coupling and flow regulation. Bioimpedance technology effectively detects changes in right ventricular flow regulation during pulmonary circulation. By analyzing the pulmonary blood flow impedance graph to quantify the indices of elasticity and resistance, the function of blood flow regulation can be assessed. This method establishes a foundation for non-invasive bioimpedance assessment of pulmonary circulation dynamics.

Abstract:To enhance the guided wave amplitude excited by the single-magnet electromagnetic acoustic transducer (EMAT), a simple methodology using a semi-closed yoke structure is proposed in this study, which is based on both simulation and experiment analyses. The proposed EMAT structure consists of one rectangular-shaped magnet, a semi-closed yoke, a meander-line coil, and a metal sample. Through simulation analyses, three different possible yoke structures are compared, and the feasibility of the semi-closed yoke structure for waveform enhancement is confirmed. The parameters of the semi-closed yoke structure are carefully studied, including side plate spacing, top plate spacing, plate thickness, and relative magnetic permeability of the material used. Therefore, their optimized combination achieves the highest possible waveform amplitude. Compared with the original yoke-free solution, the semi-enclosed yoke EMAT with optimized parameters achieves an excitation efficiency improvement of nearly 50% in simulation. Accordingly, the optimized EMAT units are manufactured. Experimental results on a 1 mm thick aluminum plate show that with a lift-off distance of 2 mm, the 10- cycle 1 MHz S0-Lamb wave excited by the proposed solution exhibits a 40. 42% amplitude enhancement compared to the original design. When the lift-off distance is 6 mm, the improvement is 31. 32% . Then, a 5-cycle 415 kHz A0-Lamb wave is implemented on a 2 mm thick aluminum plate, and the amplitude enhancement brought by the proposed scheme is more than 30% at different lift-off distances. Hence, the general applicability of the proposed method is verified. The used yoke is made of common cold-rolled steel sheet (SPCC), which is readily available and suitable for practical applications.

Abstract:Line-structured light vision 3D measurement technology is widely used for its high precision and non-contact 3D reconstruction. However, existing methods face calibration coupling issues and are highly sensitive to background noise and lighting changes in complex environments, leading to reduced accuracy in stripe extraction and 3D measurements. To address these challenges, a robust 3D measurement method is proposed, which is based on convolutional neural networks (CNN). First, we design an innovative Residual U-shaped block feature pyramid network (RSU-FPN) to suppress background noise and achieve high-precision extraction of the structured light stripe center. Second, we develop a new line-structured light sensor and introduce a decoupled calibration model that separates camera and light plane calibration, enhancing system flexibility and scalability. Experimental results show that our method achieves high-precision stripe extraction with root mean square errors of 0. 005 mm, 0. 009 mm, and 0. 097 mm in the x, y, and z directions, respectively. It also provides high-precision 3D reconstruction on different surface types, demonstrating its robustness and excellent performance in real-world applications.

Abstract:To achieve rapid and accurate estimation of traffic noise in urban road video surveillance scenarios, a real-time noise estimation method based on computer vision is proposed. First, starting with an analysis of the mechanisms behind road traffic noise, a series of computer vision-based methods for extracting traffic flow information related to urban road noise are introduced, improving the convenience of traditional methods for extracting traffic flow data. Secondly, to address the low accuracy of traditional noise estimation algorithms, an analysis of the factors influencing urban road traffic noise is conducted. By combining traffic flow features with environmental factors, a machine learning-based model for traffic noise estimation is developed, enhancing the accuracy of urban road noise estimation. Finally, the short-term variation patterns of urban road traffic noise are analyzed, and a variable-scale feature extraction time window is determined. A complete real-time noise estimation solution is proposed, improving the real-time performance of noise estimation. Experimental results show that the proposed computer vision-based traffic flow information extraction method outperforms commonly used object detection and tracking algorithms in accurately extracting traffic noise-related information. The developed model for traffic noise estimation offers better real-time performance and accuracy compared to traditional models and provides more accurate estimates in various scenarios compared to existing machine learning-based noise estimation methods. The noise estimation methods with time scales of 3 and 10 minutes are validated, demonstrating practical application value. Keywords:computer vision; traffic noise; feature extraction; noise estimation; machine l

Abstract:To solve problems such as limited detection field of view, easy occlusion of target and weak image information of single light source in the current process of single unmanned aerial vehicle (UAV) target detection, and improve the reliable and efficient perception and computing capability of UAV, a multi-modal target detection method for multi-UAV cooperation is proposed in this article. Firstly, a multimodal object detection algorithm based on visible light and infrared fusion is studied, and a dual light fusion model based on a convolutional fusion network driven by visual tasks is proposed. The fused image is fed back to the fusion network through the semantic segmentation network, and a dual light image fusion model with a smaller loss is iteratively trained on the fusion network parameters. Then, the fused image is input into the visual perception enhancement module for image enhancement, eliminating the impact of poor lighting conditions on image quality and improving the preservation of target detail features. The effectiveness of the algorithm is verified on the MSRS dataset. In addition, an active perception process based on distributed biosensing processing is proposed for multi-drone collaborative detection. By calculating the detection confidence of the drone at the location of the sensed target and allocating the detection priority of the host and slave through the release of pheromones, a guidance strategy for multi-drone collaborative detection tasks is completed, achieving target detection in unstructured ground scenes under different lighting conditions. Experimental results show that the algorithm has a 56. 55 ms delay and 45. 84 fps reasoning speed on the unmanned aerial intelligent edge computing platform RK3588, and can accurately detect typical military targets deployed in ground scenes, with an average detection accuracy of 78. 5% .

Abstract:In view of the small and unobvious target of the UAV and the re-tracking problem after the target is occluded, an anti-UAV recognition and tracking algorithm, YOLO-G-KCF, which integrates improved YOLO and improved KCF models, is proposed. This algorithm introduces multi-channel gradient features and original image features into the YOLOv10 algorithm by means of feature concatenation in feature processing. Therefore, the improved algorithm has a better detection effect on targets under strong light, shadow, and other complex lighting conditions. Meanwhile, the multi-channel gradient features are introduced into the KCF target tracking algorithm, and a multi-scale detection is designed to make the KCF algorithm have good scale adaptability. The KCF tracking results are introduced after the Head, and the new loss function Inner-IoU is calculated to more accurately identify the tracking target. The experimental results show that the YOLO-G-KCF algorithm achieves a 95. 3% accuracy rate when tested on the dataset comprising multiple open-source UAV video target tracking. This is in comparison with the original model of YOLOv10, wherein the improved model′s mAP@ 0. 5 has an increase of 1. 37% , and the average precision mAP@ 0. 5 reaches 94. 28% and the recognition speed reaches 112 FPS, which can operate at more than 100 FPS to satisfy the real-time requirements of UAV target recognition and tracking. Compared with other algorithms without sacrificing speed, introducing recognition mechanisms for tracking and improving them has better recognition and tracking effects. YOLO-G-KCF algorithm realizes the recognition and tracking of low-speed, small-sized, and low-altitude unmanned aerial vehicles in situations where the target is small, not prominent, and occluded. It has high recognition accuracy, strong anti-interference ability, good real - time hardware development, and certain theoretical research and engineering application value.

Abstract:Forward-looking sonar is a crucial sensor in ocean exploration, widely used for target detection and tracking over long distances. However, sonar data acquisition is often compromised by environmental noise in the ocean, which is unevenly distributed and reduces the accuracy of target detection in sonar images. Traditional convolutional neural networks (CNNs) for tracking forward-looking sonar targets often fail due to the low frame rates of sonar image sequences and unclear target features. To address the issue of noise pollution in forward-looking sonar images, this paper proposes an enhanced BM3D (Block Matching and 3D Filtering) algorithm tailored to the specific characteristics of sonar images. The Manhattan distance is utilized in place of the Euclidean distance to compute similar block-matching distances, improving noise handling across different types and intensities. Additionally, to mitigate target loss, we introduce a forward-looking sonar image target detection algorithm based on an improved YOLOv8-s network. This enhancement includes modifications to the ConvNeXt-based C2N algorithm, the addition of a shallow feature detection head, and improvements to the normalized Wasserstein distance (NWD) loss function. Experimental results from sonar image data acquisition show that the accuracy of the improved model is 87. 2% , with an mAP@ 0. 5 of 85. 4% . Compared to the original YOLOv8-s model, the modified model′s size increased by only 4. 6 MB, while precision improved by 5. 1 percentage points, and mAP @ 0. 5 rose by 4 percentage points. The improved YOLOv8-s outperforms other detection models, significantly enhancing target detection accuracy in sonar images.

Abstract:Magnetic flux leakage detection in oil and gas pipelines is a crucial method for evaluating pipeline integrity. However, traditional manual analysis methods suffer from low efficiency and high false detection rates. This study proposes an intelligent recognition method for pipeline magnetic flux leakage detection based on the YOLOv8 deep learning algorithm, achieving automated detection of pipeline defects. The research innovatively constructs an automated training dataset generation method based on manual annotation experience, effectively inheriting expert domain knowledge, and significantly improving dataset construction efficiency. Through preprocessing and image enhancement of magnetic flux leakage signals, raw data are converted into standardized grayscale images, and an adaptive image enhancement strategy is adopted to effectively improve image quality and feature distinguishability. In this study, a dataset of 36,098 high-quality magnetic flux leakage images is collected and generated from real-world engineering projects, including 3, 403 defect-containing images used as the training set. The defects exhibit a relatively uniform distribution in the axial-circumferential plane of the pipeline, with localized high-density regions near weld seams. The defect sizes are predominantly within a smaller range, exhibiting a long-tailed distribution, providing a solid data foundation for model training. During training, the model′s precision and recall metrics stabilized at 0. 66 and 0. 60, respectively, with an mAP@ 0. 5 of 0. 57 and an mAP@ [0. 5:0. 95] of 0. 27. Testing on real-world engineering data achieves a precision of 63. 17% , recall of 65. 24% , and an F1 score of 64. 19% . The feasibility and excellent detection performance of the YOLOv8 model for pipeline inspection tasks are verified. This method not only significantly improves detection efficiency and reduces manual costs but also effectively avoids judgment bias caused by human factors. The results show that deep learning-based intelligent recognition methods have broad application prospects in pipeline magnetic flux leakage detection.

Abstract:This article proposes an improved UAV target detection algorithm, BG-YOLO, to address the limitations of existing UAV detection models in terms of model size, computational resource requirements, and the detection performance of small targets. Based on YOLOv8, BG-YOLO adds detection heads to the high-resolution feature layers, effectively reducing information loss during image downsampling and significantly enhancing the model′ s ability to detect small targets. The introduction of the Biformer attention mechanism enables precise capture of long-range dependencies in images, thereby strengthening the model′s perception of targets at different scales. Additionally, the incorporation of the NWD loss function overcomes the issue of traditional loss functions being sensitive to positional deviations in small target detection, thereby significantly improving the model′s robustness. The model′s lightweighting based on GhostNetV2 replaces traditional convolutional modules, reducing model parameters and computational load while maintaining detection accuracy. Experimental results show that BG-YOLO achieves a 10. 3% improvement in mAP@ 0. 5 on the Det-Fly dataset compared to YOLOv8, with a 33. 18% reduction in model parameters, and a 7. 9% improvement compared to YOLOv9. Moreover, on the self-collected dataset, BG-YOLO demonstrates excellent performance in detecting low, slow, and small targets in various scenarios, including sky, mountain, and urban backgrounds, achieving average precisions of 96. 2% , 88. 1% and 86. 2% , respectively, with detection speeds of 150. 36, 128. 21 and 112. 53 fps. These results meet the real-time requirements of high detection accuracy and speed. In summary, BG-YOLO significantly enhances the detection accuracy and real-time performance for low, slow, and small UAV targets through the design of detection heads, incorporation of attention mechanisms, refinement of the loss function, and model lightweighting, thereby offering broad application prospects.

Abstract:Ultra-high-voltage overhead transmission lines are crucial in power systems. But, they often face accidents triggered by external factors, such as construction activities and wildfires. These incidents not only damage the national economy and affect grid stability, but also pose a threat to the safety of power workers. Deep learning-based object detection methods offer a novel solution for detecting external force damage objects. However, existing methods often rely on local neighborhood information for sampling operations, which limits their perceptual range and expressive capabilities. To address this issue, a real-time global awareness-enhanced method, GAE-YOLO, based on YOLOv10, is proposed to improve the detection accuracy of external force damage objects in ultra-high-voltage overhead transmission lines. To overcome the limitations of local perception in traditional methods, two novel upsampling and downsampling modules are designed, including the global awareness downsampling module ( GADM) and the global awareness upsampling module (GAUM). GADM enhances perceptual performance by learning global spatial information from the feature map and generating global perception weights to optimize the downsampling process. GAUM dynamically enhances the membership relationship of sampling points by utilizing channel information from deep feature maps, effectively highlighting object boundaries. To evaluate the effectiveness of GAE-YOLO, a large-scale dataset for detecting external force damages in ultra-high-voltage overhead transmission lines is constructed. The model achieves mAP of 93. 05% , a mAP 5. 13% improvement over the baseline model. Experimental results show that GAE-YOLO significantly improves the detection accuracy of external-force damage objects, offering substantial application value and providing new technical support for the safe operation of power grids.

Abstract:In the process of data sampling and fitting reconstruction of optical system point spread function, the background noise and external interference will cause the error measurement results of instrument. The traditional fixed objective function point spread function fitting algorithm exhibits poor adaptability, making it difficult to accurately restore the morphology of light spots during optical instrument testing. These error terms will affect the analysis results of image quality to some extent. To solve this problem, a calculation method of point spread function for optical imaging system is proposed. The energy acquisition and reconstruction are carried out according to different spot morphology, and the original shape of spot is preserved effectively. Through the cubic spline interpolation of spot image data, the subpixel matrix is constructed. By performing scattered light suppression and centroid correction on different positions of light spots, we can obtain a three-dimensional surface closer to the real spot shape. Considering sensor accuracy variations, the method adjusts step size to control the spot radius corresponding to encircled energy, accommodating different testing requirements. Gaussian spot simulation analysis and laboratory test results show that, compared with Gaussian fitting method, nearest neighbor interpolation method and blind deconvolution method, the method proposed in this paper is closer to the actual situation. The error of algorithm based on surface interpolation is only, ε = 0. 000 2, the deviation rate within the observation area of image quality is less than 5% . This method can provide more accurate capability concentration results. The algorithm is sensitive to the morphological changes of the point spread function and can effectively distinguish the presence of multiple peaks in small size spot. It also has high accuracy under normal temperature test conditions. The algorithm has engineering applications in the field of interference source location and optical focusing, and provides a theoretical foundation for the development and performance analysis of optical imaging systems.

Abstract:In simultaneous localization and mapping ( SLAM), the loop closure detection is a critical step to improve localization accuracy. By identifying loop closures and correcting accumulated errors, the accuracy and robustness of localization can be significantly enhanced. However, most existing LiDAR-based loop closure detection methods primarily rely on low-level features such as coordinates and reflectivity to construct descriptors, failing to fully utilize semantic information within the scene. As a result, these methods often face challenges in terms of accuracy and reliability in complex scenarios. To address these limitations, this article proposes a semanticassisted intensity scan context method to overcome the insufficiencies of existing approaches. First, the proposed method employs the iterative closest point (ICP) algorithm for coarse registration of two-point clouds, reducing the impact of angular and translational errors on loop closure detection. On this basis, semantic features are integrated with the three-dimensional coordinates and reflectivity information of the point clouds to generate a global descriptor that incorporates multi-level features. Finally, loop closures are determined by calculating the similarity of the descriptors, enabling more reliable detection. Experimental results on the publicly available KITTI dataset show that the proposed method achieves a maximum F1 score improvement of 19. 71% compared with the Scan Context algorithm, while reducing the average root mean square error (RMSE) by 36% compared with the lego-loam algorithm. Additionally, real-world experiments in a campus environment show that the proposed method improves the maximum F1 score by 19. 23% compared with the LIOSAM algorithm and by 70. 62% compared with the lego-loam algorithm. Furthermore, the average RMSE is reduced by 56. 68% compared with LIO-SAM and by 20. 7% compared with lego-loam. These results show that the suggested method not only greatly improves the accuracy of loop closure detection but also exhibits greater robustness in diverse scenarios. By incorporating semantic information, this method markedly improves the discriminative capability of descriptors in complex environments, providing new perspectives and methodological support for the development of SLAM technologies.

Abstract:Marine electric field measurement is of significant importance for monitoring marine activities and exploring seabed resources. Due to the low-frequency characteristics of marine electric field signals, weak signal measurement is inevitably affected by device 1 / f noise and environmental frequency-conversion harmonic noise. Notably, the frequency of environmental frequency-conversion harmonic noise may closely approach that of the target electric field signals, making it difficult to suppress using existing chopper amplification techniques. To address the challenge of existing measurement circuits in suppressing environmental frequency-conversion harmonic noise in conventional measurement circuits, this article proposes a marine electric field measurement method incorporating switched-frequency conversion quartz crystal resonance narrowband noise suppression. The approach converts marine electric field to 32. 768 kHz, and the up-converted signal undergoes ultra-narrowband filtering via a high-Q quartz crystal resonator. Theoretical analysis is conducted on the frequency-conversion tuning process of the marine electric field and the frequency-selective characteristics of the proposed measurement circuit, leading to the derivation of the expression for the output signal after frequency-conversion tuning narrowband processing. An experimental platform for electric field measurement is established, and the designed frequency-conversion tuning narrowband measurement circuit is tested. Experimental results show that when the marine electric field signal frequency ranges from 0. 01 to 0. 2 Hz, the sensitivity of the frequency-conversion tuning narrowband measurement reaches 2. 78 times that of direct measurement, with a circuit bandwidth of less than 0. 4 Hz. For a marine electric field signal at 0. 1 Hz, the signal-to-noise ratio of the frequency-conversion tuning narrowband measurement shows a 19. 82 dB improvement compared to direct measurement. This method not only enables the detection and measurement of weak marine electric field signals in strong noise environments but also provides a foundation for future deployments of marine electric field measurement arrays.

Abstract:To enhance sensor resolution and accuracy without increasing size, while enabling absolute angular displacement measurement for space-constrained industrial applications, this paper proposes a high-resolution absolute time grating angular displacement sensor using secondary modulation dual-channel synchronization. The sensor consists of a fixed ruler, a moving ruler, and a driving circuit board. The fixed ruler includes two fine machine code channel excitation coils, two coarse machine code channel induction coils, and the secondary winding of an electromagnetic coupling coil. The moving ruler has two fine machine code channel induction coils, two coarse machine code channel excitation coils, and the primary winding of the electromagnetic coupling coil. The driving circuit board contains a driving signal generator, two induction signal processors, and an FPGA core circuit. The fine machine code channel induction coils on the moving ruler are connected in series with the coarse machine code channel excitation coils for secondary modulation. When the fine machine code channel excitation coil is powered, the angular displacement signals from the fine machine code channel induction coils are modulated onto the coarse machine code channel, enhancing resolution by effectively integrating both channels. Additionally, one precision machine′s induction signal is transmitted back to the fixed ruler via the electromagnetic coupling coil as a full-cycle positioning signal. Both signals are processed synchronously by the FPGA for absolute angular displacement measurement. A prototype with an outer diameter of 140 mm was fabricated using PCB technology. Experiments show that the sensor achieves absolute angular displacement measurement with a single drive circuit, improving resolution from 0. 38″ to 0. 2″ (a 47% increase) and reducing measurement error from ±34. 14″ to ±16. 06″ (a 53% reduces).

Abstract:In the early diagnosis of submucosal tumors, using endoscopy to obtain tactile feedback information can help improve the accuracy of elastic modulus detection of biological tissues, thereby accurately locating blood vessels, determining the health status and type of biological tissues, and improving the quality of treatment. In this paper, a novel micro piezoelectric tactile sensor ( PTS, ϕ= 2. 0 mm) suitable for installation on an endoscope to detect the elastic modulus of biological tissue is designed. This device mainly consists of two components with different stiffness ( internal and packaging components) and a polyvinylidene fluoride ( PVDF) piezoelectric film. Simultaneously, based on the series spring model and piezoelectric transfer equation, a numerical model for PTS / biological tissue contact sensing has been established, and the sensing law of biological tissue elastic modulus and its energy conversion law of PTS have been deeply analyzed via the Comsol Multiphysics. Meanwhile, a PTS prototype was prepared using a MEMS manufacturing process, and a PTS / biological tissue dynamic load testing platform was conducted to test the soft / hard characteristics of different artificial tumors in the pig stomach for validating the above model. Moreover, the calculation results show that most of the axial load is transmitted by the internal components-copper balls, and the deformation of the PDMS encapsulation layer is minimal when PTS comes into contact with harder biological tissues. On the contrary, the PDMS encapsulation layer undergoes greater deformation when PTS comes into contact with softer biological tissues. The calculation results also indicate that the response voltage (V) and stress (σ) generated in the contact area of the PVDF layer near the internal component ( copper ball) are significantly higher than those in the contact area with the packaging layer, which reveals the mechanical-electric field coupling and energy transfer process of PTS. Furthermore, the experimental and computational results confirm that the PTS device, with a 2. 0 mm diameter, was successfully installed in the endoscopic biopsy channel. It is noteworthy that the ratio of the response voltage (V1 / V2 ) between the internal and the packaging component is linearly related to the elastic model of biological tissue ( Et ). when Et increases, V1 / V2 incr 0. 2~ 3. 5 MPa, which aligns well with the model calculation results. The above proposed PTS device effecti eases within vely identifies elastic parameters of tissue, and this measurement method providing a new way for solving the measurement problem of biomechanical information of animal tissue Ke

Abstract:The thermal expansion and contraction of the mechanical structure, temperature drift of circuit components, and variations in the magnetic induction intensity of the permanent magnet all contribute to indication drift in micro electromagnetic force weighing sensors. Investigating the mechanisms of hardware-induced drift, conducting appropriate temperature tests, and implementing effective temperature compensation methods are crucial for mitigating temperature-related drift issues. For a sensor with a range of 200 g and a resolution of 0. 1 mg, this study employs mathematical modeling to analyze key factors including the mechanical lever force transmission ratio, voltage reference of the driving circuit, acquisition resistor, and the temperature drift model of the permanent magnet. This analysis identifies the primary contributors to temperature drift and determines the optimal installation position for the temperaturecompensated sensor. A linear temperature rise test is conducted, collecting and recording indication drift data at each 10℃ interval. Quadratic fitting is then applied to derive the temperature drift compensation function. The study proposes an interval temperature compensation method where the zero reference point and half-scale reference point proportionally follow the maximum scale reference point. This approach compensates for both mechanical and circuit drifts without altering the length of the scale range interval. Additionally, the concept of dynamic compensation sensitivity is introduced, updating the compensation sensitivity in real time based on the ratio of the compensated interval length to the graduation number. This addresses asymmetry issues in temperature compensation amounts across different scale intervals and enhances compensation accuracy. Experimental results show that the proposed method achieves dynamic temperature compensation for a 200 g sensor with a resolution of 0. 1 mg within the 5℃ ~ 35℃ range, with a compensation error absolute value less than 0. 5 mg, thereby improving the sensor′s adaptability in environments with significant temperature fluctuations.

Abstract:Accurate and reliable sensor extrinsic calibration methods are crucial for high-precision localization and navigation in radarinertial fusion systems. However, most existing calibration methods rely on the acquisition of triaxial excitation from inertial sensors, and their performance deteriorates or even fails when the radar and inertial sensors are installed on ground robots with restricted movement. To address this issue, a novel calibration method based on planar features in radar point clouds is proposed for radar-inertial odometry on ground robots. The method first constructs residuals using planar features in the radar point clouds and rapidly converges the extrinsic parameters to a smaller error range by minimizing the distance from radar points to the planes. Subsequently, it further optimizes the extrinsic parameters based on the octree structure, incorporating the spatial occupancy information of the radar point clouds. Finally, by integrating ground constraints through a ground segmentation algorithm, the method corrects the errors in the Z-axis direction that cannot be constrained during planar motion, thereby achieving complete 6-DOF (degrees of freedom) extrinsic parameters. Experimental results show that the proposed method significantly outperforms other algorithms in calibration accuracy on two open-source datasets, with average rotational angle errors reduced by 43. 73% and 36. 47% , and average translational errors reduced by 76. 33% and 41. 52% , respectively. In real-world vehicle validation experiments, the method successfully achieves calibration in various scenarios, including flat terrain, rugged terrain, and narrow passages, further demonstrating its reliability and robustness in practical environments. In localization accuracy analysis experiments, the absolute trajectory root mean square error of the FAST-LIO2 algorithm, initialized with the calibration results of this paper, is reduced by approximately 6. 54% , evaluating the practicality and accuracy of the proposed method.

Abstract:To enhance sensor integration and align with the trend of device miniaturization, a compact fiber optic probe based on a fiber Bragg grating-Fabry-Perot (FBG-FP) cascaded structure is proposed and fabricated for monitoring refractive index and temperature in biological fluids. By constructing a cascaded structure reflectivity distribution model, the device cascade sequence is optimized. The fiber FP probe with a single-mode fiber-hollow-core fiber-single-mode fiber structure is prepared using precise cutting and splicing techniques. A FBG is inscribed 100 μm from the splicing interface using femtosecond laser direct writing, achieving a cascaded compact dual-sensing element fiber optic probe. The wavelength / intensity response characteristics of the FBG and FP structures to temperature and refractive index effectively resolved the cross-sensitivity issue in principle. A temperature and refractive index experimental system is set up to analyze the sensing characteristics of four fiber optic probes within a temperature range of 25℃ to 55℃ and a refractive index range of 1. 333 0 to 1. 381 6. Experimental results show that, during temperature cycling, the FBG and FP central wavelengths redshifted with increasing temperature and blueshifted with decreasing temperature, with average temperature 356 仪 器 仪 表 学 报 第 4 6 卷 sensitivities of 9. 36 and 8. 52 pm / ℃ , respectively. As the ambient refractive index increased, the FBG wavelength and intensity are unchanged, while the FP interference wavelength stays constant and the resonance intensity gradually decreases, with the intensity fitting results showing a parabolic trend. Segmented linear fitting with a refractive index of 1. 354 6 as the boundary reveals that within the refractive index range of 1. 333 0 to 1. 354 6, the FP sensitivity averaged 5. 86 dBm / RIU with a maximum of 10. 72 dBm / RIU, and within the range of 1. 354 6 to 1. 361 8, the FP sensitivity averaged 1. 40 dBm / RIU with a maximum of 2. 74 dBm / RIU. This sensor, characterized by its simple fabrication, compact structure, and high sensitivity, shows promising application prospects in the field of biological fluid monitoring.

Abstract:MIS is widely utilized in numerous clinical fields due to its advantages, such as minimal trauma, reduced pain, and rapid recovery. However, the small incisions characteristic of MIS limit direct tactile feedback at the surgical site, significantly increasing the complexity of intraoperative operations. As a result, the real-time and accurate force sensing during surgery is considered essential for ensuring both precision and safety. FBG sensors, due to their high sensitivity, electromagnetic interference resistance, small size, and biocompatibility, are regarded as highly promising for MIS applications. But several challenges still hinder the further development of this technology, necessitating the exploration of solutions. This paper systematically reviews the research progress of FBG sensors in force sensing for MIS over the past decade. Common techniques for decoupling strain and temperature in FBG sensors are summarized, which are crucial for mitigating temperature interference and improving force sensing accuracy. Furthermore, the latest advancements and applications of FBG force sensors in endoscopic MIS, vascular interventional surgery, retinal microsurgery, and other MIS procedures are discussed, with a focus on sensor structure design and force feedback calculation methods. The challenges associated with the practical application of FBG force sensors, including the design of high-precision micro-force sensors, real-time data processing and feedback, sensor system intelligence, multi-modal data fusion, and the commercialization and clinical translation of these sensors, are also examined. Finally, the paper envisions future development directions for FBG force sensors, emphasizing the potential of technological innovations to enable their widespread adoption in the medical field.