Search

Accurately estimating global soil heterotrophic respiration (RH) is crucial in evaluating whether terrestrial ecosystems act as carbon sources or sinks. However, current global RH estimates were significantly restricted by the scarcity of in situ RH observations and their biased distribution, leading to considerable uncertainties. This study developed a novel data-driven model of global RH based on the environmentally similar zones of global in situ RH sites with daily and subdaily observations and remote sensing data with high spatial and temporal resolutions. Compared with the unified modeling method using all available data as the training samples of data-driven models, our zone modeling method was more accurate. The relationship between observed and predicted RH was improved, with the R2 value increasing from 0.41 to 0.53 and the RMSE decreasing from 0.87 to 0.78 g C m−2 d−1. Our study effectively improved the problem that the data-driven models were highly affected by the spatial representativeness of in situ RH observations and achieved a significantly improved accuracy for global RH estimation entirely based on remote sensing data. Future research focusing on improving the sparse sampling of in situ RH sites and the availability of remote sensing data will help to reduce the uncertainties of our study.

With the rapid proliferation of drones across various domains, aerial target detection has become increasingly crucial. However, the targets in aerial images present challenges such as scale variation, small size, and density, leading to suboptimal performance of current detectors on aerial images. Based on the aforementioned challenges, we design an efficient aerial target detection algorithm called DMFF-YOLO. Specifically, to address the issues of small target size and scale variation, we design the DMFF neck structure, adding a small target detection head to tackle the small target size problem, using the DMC module to fuse different scale features for enriching detailed information, and employing the DSSFF module to construct a scale sequence space to solve the target scale variation problem. In the network backbone, we employ RFCBAMConv modules as downsampling layers, which interact with receptive-field features to mitigate the information disparity caused by positional changes and outperform traditional convolutional layers. Finally, we design the Soft-NMS-CIoU module to address the issue of suppressing adjacent boxes due to dense targets. On the VisDrone dataset, compared to the original algorithm, our method reduces the number of parameters by 31.1% while achieving an 11.7% improvement in mAP50. Extensive experiments on the VisDrone, DOTA, and UAVDT datasets demonstrate that the proposed algorithm performs well in aerial image detection tasks.

The fault characteristics of large-capacity wind farms affect the sensitivity of transmission-line protection, resulting in protection devices refusing to operate or mal-operate. A phase-to-ground fault-locating method with single-end transient information for wind farm transmission lines was proposed in this study. The proposed method employs a time-domain transient quantity integration equation matrix with the transmission line equivalent resistance, inductance, and fault path resistance to solve the fault location. The effectiveness of the scheme was evaluated using Real-Time Lab (RTLAB). The impact of factors such as the fault resistance, location, and distributed capacitance is considered. Under different fault locations, the distance measurement error of the proposed algorithm is within 3%, and there is no rejection or mal-operation. The results demonstrate that the proposed algorithm is characterized by high speed and reliability.

Multiple scattering of laser ray leads to distance calculation error and accumulated intensity error in LiDAR waveform. This study quantitatively assesses multiple scattering in LiDAR canopy waveform and explores its relationship with sensor configurations, forest structure, and leaf biochemistry. Airborne LiDAR measurements were conducted to verify the presence of multiple scattering in spaceborne LiDAR waveforms. A Monte Carlo-based radiative transfer model, DART, was applied to simulate laser multiple scattering. The simulated results were compared with the actual measurements and further revealed how forest characteristics and sensor parameters influence multiple scattering. Results show that multiple scattering contributes significantly (∼30% of canopy subwaveform and ∼18% of total waveform) when there are sufficient scatterers in a large LiDAR footprint, and the scatterers possess a reasonable spatial distribution and a strong albedo at the laser wavelength; contribution of multiple scattering is typically less than 5% for small-footprint LiDAR; multiple scattering of large-footprint LiDAR increases with forest leaf area index and fractional coverage, while also varying with laser pointing angle, crown shape, and leaf angle distribution; leaf biochemical contents affect multiple scattering only at their sensitive wavelength. These findings help to quantitatively explain errors in the simulation, validation, and interpretation of LiDAR signal.

In this article, a partially isolated dual work function (PIDWF) gate In-Ga-Zn-O (IGZO) thin-film transistor (TFT) is proposed to reduce the off-state current (Ioff) obviously, which also provides a feasible integration method for stacking IGZO TFT on Si-based devices. It is found that compared with the general back gate IGZO TFT structure, the Ioff of the proposed IGZO TFT reduces from $2.57\times 10{^{-}14 }$ A/ $\mu $ m to $7.57\times 10{^{-}16 }$ A/ $\mu $ m, achieving two orders of magnitude improvement. This breakthrough has the potential to increase the retention time of DRAM applications by nearly 100 times. Moreover, the pronounced novel structure has mitigated parasitic capacitance, thereby leading to a notable 47.7% reduction in write latency within dynamic-random-access-memory (DRAM) circuits. The relevant operation mechanism is carefully demonstrated and verified by the simulation of the electric field and potential barrier results by technical computer-aided design (TCAD). Furthermore, the impacts of the dual gate work function level, the length, and the type of isolation dielectric between dual work function gates are systematically investigated. The results show that the off-state leakage is further reduced by increasing the difference of the work function levels between in dual gates, the dielectric length (LD) and using the isolation layer with a lower dielectric constant. The PIDWF gate IGZO TFT exhibits scalability and is capable of achieving an 84.6% reduction in leakage current even with ultra-short channel lengths, which offers a promising application for future 3D DRAM applications with little extra cost.

Amorphous oxide semiconductor-thin film transistors (AOS-TFTs) have attracted considerable attention due to their impressive performance in various applications. However, there is a limited amount of study available on the reliability of AOS-TFTs. This work investigates the endurance and reliability mechanisms of Indium Gallium Zinc Oxide (IGZO) and Indium Tin Oxide (ITO) TFTs. The devices underwent a range of test conditions to evaluate their endurance properties. The study utilized Zero-Bias Endurance Tests (ZBET) to examine the fluctuating behaviors of threshold voltage, revealing valuable insights into the causes of electrical instability. The study highlights the crucial importance of electron depletion and restoration dynamics in affecting the reliability of TFTs. Additionally, the study found differences in the performance of IGZO-TFTs and ITO-TFTs, suggesting that the differing features of the materials have a significant impact on the endurance and reliability of TFTs.

Due to the distinctive fault characteristics of wind power converters in large-capacity wind farms (WFs) with power converter-based systems, protective relays for the connected transmission lines can malfunction, which poses a serious challenge to the stable and reliable power transmission. This article focuses on addressing this issue by proposing a phase-to-phase fault protection method that utilizes single-end transient information for wind farm transmission lines. Conventional solutions often encounter difficulties such as the negative influence of fault resistance and the uncertainty of remote parameters in single-ended data locations. The proposed method employs time-domain differential equations to accurately solve the fault location problem. It takes into account the equivalent resistance and inductance of the transmission line, as well as the fault path resistance. This helps minimize the adverse effects of fault resistance and parameter uncertainties, making the method insensitive to fault impedance and system parameters. The effectiveness of the proposed scheme is comprehensively evaluated using the Real-Time Lab (RTLAB). Various factors like fault resistances, locations, and distributed capacitance are considered. The results clearly demonstrate that the proposed algorithm not only operates at high speed but also exhibits remarkable reliability, offering a promising solution for enhancing the protection of wind farm transmission lines.

The problem of closed branches and leaves in orchards can cause satellite navigation systems to experience star loss, leading to safety issues such as collisions and loss of control. To address this issue, this study develops a multi-mode navigation control system for satellite navigation and sensor tracking, aiming to achieve multi-mode intelligent path judgment and ensure the accuracy and safety of operations. This method utilizes point cloud iterative nearest point algorithm LiDAR technology to gain 3D information of the surrounding environment when satellite signals are unavailable. Through perception and analysis of the environment, real-time path planning and obstacle avoidance decision-making are achieved. At the same time, through micro-inertial navigation technology, accurate perception and control of the position and attitude of navigation equipment are achieved, thereby achieving precise control of the operation process. According to relevant experimental data, the experimental distance accuracy based on the vehicle mounted daytime environment dataset was 84.88%, and the duration accuracy was 97.63%. Accordingly, the proposed system is capable of accurately determining the current position and heading, as well as making path planning and obstacle avoidance decisions based on real-time environmental information. This ensures the accuracy and safety of operations. The research method exhibits higher stability and reliability in orchard operations, providing an effective solution for the accuracy and safety of orchard operations.

Synthetic aperture radar (SAR) image change detection is currently a popular topic, but the existence of speckle noise renders it challenging. Making full use of the neighborhood information can reduce noise interference and improve accuracy. To this end, this study proposes a graph-guided method for SAR image change detection. First, we establish the local, nonlocal, and global connections according to three types of neighboring rules. A novel heterogeneous graph is then constructed by assigning different weights to these multilevel connections. On the basis of heterogeneous graph, a composite random walk matrix (CRWM) is presented to quantify the similarity of multilevel connections. Thereafter, the heterogeneous graph shift system consisting of the multiorder CRWM is designed to aggregate the attributes of neighboring vertices along the multiorder and multilevel connections. The change measure can be implemented by comparing the output signals from the heterogeneous graph shift systems, resulting in the generation of difference image with good separability. Finally, change analysis is carried out using a binary classification algorithm. Experiments conducted on six real SAR datasets confirm that the proposed M2HG method successfully strikes a balance between speckle suppression and change enhancement. This strategic balance translates into improvements in performance metrics, with OA, κ, and F1 indicators surpassing those of the suboptimal method by 0.87%, 3.34%, and 3.84%, respectively. Overall, the proposed method emerges as a promising contender for SAR image change detection tasks.

This research addresses the critical challenge of classifying surface defects in lithium electronic components, crucial for ensuring the reliability and safety of lithium batteries. With a scarcity of specific defect data, we introduce an innovative Cross-Domain Generalization (CDG) approach, incorporating Cross-domain Augmentation, Multi-task Learning, and Iteration Learning. Leveraging a steel surface defect dataset as foundational knowledge, our approach compensates for the limited lithium-specific data and enhances model generalization. We also introduce the Lithium Electronic Surface Defect Classification (IESDC) dataset, demonstrating significant accuracy improvements over baseline methods. Our comprehensive evaluation covers model interpretability, robustness, and adaptability. Beyond battery technology, this methodology offers a framework for data scarcity challenges in various industries, emphasizing the importance of adaptable learning methods.

Landslides, as a major geological hazard, caused the significant threats to human life and property. Therefore, the identification of potential landslides is a crucial concern. This study combines synthetic aperture radar interferometry phase-gradient stacking with deep learning to achieve efficient and accurate identification of large-scale potential landslides. By stacking phase gradients to highlight local surface deformation information and refining the YOLOv8 model based on small target features of local deformations, this study introduces improvements. This involves adding a convolutional block attention module layer to the backbone, replacing the C2f modules with GhostNetV2 to suppress information loss during long-distance feature transmission, and enhancing the network's perception and detection capabilities for small targets. In addition, a 160 × 160 small target detection head is added to the detection module to specifically handle small target detection tasks, improving accuracy and performance. A new loss function, FocalSIoU loss, is introduced based on the characteristics of the dataset, combining SIoU with the gamma bias option to make the model more targeted during training. The improved model achieved a maximum mAP50 value of 93.4% on the validation set. Finally, using mask factors to identify region deformation points, this study reduces misjudgments of non-landslides, identifying 378 potential landslides in the study area with a false positive rate of only 10.2%.

In order to solve the problem of limited driving speed of piezoelectric actuators when the piezoelectric components are small and the driving frequency is low. This paper proposes a bi-directional piezoelectric stick-slip actuator, which mainly utilizes the combination of flexible drive foot and lever amplification mechanism to realize the displacement amplification function, thus improving the driving performance. Firstly describe the structure and working principle of the proposed actuator, and carries out theoretical analysis and mathematical modeling of the flexible stator. Then the driving process of the proposed actuator is simulated and analyzed with COMSOL Multiphysics simulation software, and the feasibility of the driver structure is preliminarily verified. Finally, establish an experimental system to test the basic characteristics of the prototype. The experimental results confirm that under the piezoelectric stacks elongation of $5\mu $ m, the driving frequency of 610Hz, the maximum drive speed of the actuator is 11.21mm/s. The maximum load capacity is 0.78N and the forward and backward displacement deviation within 10 cycles is $0.99\mu $ m under the locking displacement of $10\mu $ m. The proposed stick-slip actuator is capable of realizing a large driving speed when the elongations of the piezoelectric stacks are small and the driving frequency is low.

Semantic segmentation is a crucial step in the intelligent interpretation of high-resolution remote sensing images (HRSIs). Convolutional neural networks and transformers are widely used for semantic feature extraction in remote sensing images, but the former inevitably has limitations in modeling long-range spatial dependency information, while the latter lacks the ability to learn local semantic features. Existing remote sensing image segmentation methods are optimized and modified based on the backbone networks used in natural image processing. Despite achieving relatively good results, the complexity of their network structures leads to high computational costs and limited improvements in accuracy. These methods have limited boundary distinction for ground objects in complex environments, especially for small targets. In this article, we propose an efficient semantic segmentation architecture for HRSIs called MCAT-UNet, which utilizes multiscale convolutional attention (MSCA) and the cross-shaped window transformer (CSWT) to reconstruct UNet. The encoder stacks a sequence of MSCA to exploit the advantages of convolution attention to encode context information more effectively and enhance hierarchical multiscale representation learning. The proposed U-shaped decoder integrates three skip connections using the CSWT block to further capture long-range spatial dependency and gradually restore the size of the feature map. We benchmark MCAT-UNet on three common datasets, Potsdam, Vaihingen, and LoveDA. Comprehensive experiments and extensive ablation studies show that our proposed MCAT-UNet outperforms previous state-of-the-art methods with remarkable performance.

Traffic accidents frequently lead to fatal injuries, claiming millions of lives every year. To mitigate driving hazards and ensure personal safety, it is crucial to assist vehicles in anticipating the objects in the traffic scene (treated here as important objects) which may pose a threat during the driving task. Previous research on important object detection primarily assessed the importance of individual participants, treating them as independent entities and frequently neglecting the interconnections among these participants. Unfortunately, this approach has proven less effective in detecting important objects in complex scenarios. In this work, we introduce Driving scene Relationship Understanding transformer (DRUformer), designed to enhance the important object detection task. The DRUformer is a transformer-based multi-modal important object detection model that takes into account the relationships between all the participants in the driving scenario. Recognizing that driving intention also significantly affects the detection of important objects during driving, we have incorporated a module for embedding driving intention. To assess the performance of our approach, we conducted comparative experiments on the DRAMA dataset, comparing our model against other state-of-the-art (SOTA) models. The results demonstrated a noteworthy 16.2% improvement in mIoU and a substantial 12.3% boost in ACC compared to SOTA methods. Furthermore, we conducted a qualitative analysis of our model’s ability to detect important objects across different road scenarios and classes, highlighting its effectiveness in diverse contexts. Finally, we conducted various ablation studies to assess the efficiency of the proposed modules in our DRUformer model. Through extensive experimentation, it has been demonstrated that our model performs exceptionally well in the task of driving scene important object localization. The code is publicly available on the following link: https://github.com/oniu-uin0/DRUformer

The task of background suppression in infrared small-target scenarios aims to eliminate irregular noisy backgrounds while preserving targets with high-frequency features. In infrared small-target scenes at long distances, the backgrounds become complex and the target features are degraded, highlighting a significant disparity between the detailed and realistic background and the limited features of the targets. To address these challenges, we propose a patch-based semantically enhanced generative adversarial network (GAN) named PSEnet for background suppression in infrared small-target scenarios. First, we introduce a patch-scale GAN that allows the model to concentrate on local background suppression. This shift from a global to local perspective simplifies the complexity of background suppression. Second, we employ the PSE module consisting multiscale dilated convolution and adaptive weight fusion to extract local semantic information. Third, by segmenting the infrared image into smaller patches and resampling them, we create a more balanced dataset for adversarial training. Experimental results demonstrate that the proposed algorithm significantly improves the signal-to-noise ratio of dim and small targets, reduces the missing detection rate, and achieves a precision of almost 91%. In conclusion, this approach effectively uses GANs for background suppression in complex environments.

Building extraction from remote sensing images (RSIs) requires exploring multiscale boundary detailed information and extracting it completely, which is challenging but indispensable. However, existing solutions tend to augment feature information solely through multiscale fusion and apply attention mechanisms to focus on feature relationships within a single layer while ignoring the multiscale information, which affects segmentation results. Therefore, enhancing the capability of the network to adaptively capture multiscale information and capture the global relationship of features remains a pivotal challenge in overcoming the aforementioned hurdles. To address the preceding challenge, we propose a Multiscale Direction Context-aware network with Global Attention (MDCGA-Net), employing a classic encoder–decoder architecture enhanced with direction information and global attention flow. Specifically, in the encoder part, the multiscale layer is used to extract contextual information from the interlayer. In addition, the multiscale direction context-aware module is adopted to adaptively acquire multiscale information. In the decoder part, we propose a global attention gate module to capture discriminative features. Furthermore, we construct an operation of attention feature flow to obtain the global relationship among the different features with long-range dependencies, which guarantees the integrity of results. Finally, we have performed comprehensive experiments on three public datasets to showcase the efficacy and efficiency of MDCGA-Net in building extraction.

Synthetic aperture radar automatic target recognition (SAR-ATR) models based on deep neural networks (DNNs) are vulnerable to attacks of adversarial examples. Universal adversarial attack algorithms can help evaluate and improve the robustness of the SAR-ATR models and have become a research hotspot. However, current universal adversarial attack algorithms have limitations. First, considering the difficulty in obtaining information on the attacking SAR-ATR models, there is an urgent need to design a universal adversarial attack algorithm under a black-box scenario. Second, given the difficulty of acquiring synthetic aperture radar images, the effectiveness of attacks under small-sample conditions requires improvement. To address these limitations, this study proposed a black-box universal adversarial attack algorithm: transferable universal adversarial network (TUAN). Based on the idea of the generative adversarial network, we implemented the game of generator and attenuator to improve the transferability of universal adversarial perturbation (UAP). We designed loss functions for the generator and the attenuator, respectively, which can effectively improve the success rate of black-box attacks and the stealthiness of attacks. In addition, U-Net was used as a network structure of the generator and the attenuator to fully learn the distribution of examples, thereby enhancing the attack success rate under small-sample conditions. The TUAN attained a higher black-box attack success rate and superior stealthiness than up-to-date UAP algorithms in non-targeted and targeted attacks.

Aircraft object detection in remote sensing images is a challenging task, especially for small objects, complex backgrounds, and aircraft objects in flight. Due to the lack of contextual relationship between the aircraft object and the surrounding background during flight and the small number of pixels in the object itself, the use of regular rectangular convolution for feature extraction results in many background pixels being sampled. This article proposes YOLO-FRS for aircraft object in flight detection, which uses a new module based on deformable convolution (DCN), the feature response separation deformable convolution (FRS-DCN) module. The FRS-DCN module adds semantic segmentation supervision on the basis of stacked DCNs, so that the background and object in the input of DCNs in this module are separated as much as possible. We design a soft label annotation and loss calculation method for semantic segmentation supervision. In addition, we propose a flight-state aircraft object dataset that includes multiple backgrounds and cloud interference. YOLO-FRS was tested on the proposed dataset, and the results showed that the FRS-DCN module improved the performance of aircraft object detection. Compared with multiple mainstream deep convolution models, the YOLO-FRS model also exhibits competitive performance.

Wetlands play an important role in supporting biodiversity conservation and helping sustainable development. Time series wetland classification requires a series of sample sets acquired at different times. To meet this need, we proposed a method for automatically producing global wetland samples based on 13 global and regional wetland-related datasets and millions of images from Landsat 8 Operational Land Imager (OLI), Moderate Resolution Imaging Spectroradiometer (MODIS), Sentinel-1 Synthetic Aperture Radar Ground Range Detected, and Sentinel-2 Multispectral Instrument sensors. On this basis, we analyzed the classification capabilities of wetland features in the above four sensors and obtained a multisensor, full spectral band, multi-index wetland classification feature dataset at a global scale. The analysis results show that the shortwave infrared band and land surface water index can distinguish nonwetlands from other land cover types well, while the modified normalized difference water index, normalized difference vegetation index, normalized difference water index, and thermal infrared bands could help distinguish water bodies and wetlands. C bands in Sentinel 1 perform well in identifying water bodies, but the ability to distinguish wetlands is limited. The mapping coefficients of the corresponding spectral bands and indexes between Sentinel-2, MODIS data, and Landsat were obtained to help realize the collaborative use of multisensor image data. This article could provide guidance for large-scale wetland classification and mapping research.

Facial paralysis refers to the abnormal behavior of facial muscles caused by a disorder of the facial nerve, mainly manifested as facial asymmetry. In recent years, deep learning has found extensive applications in facial paralysis detection research. However, most existing methods are constrained to assessing the severity of facial paralysis, thereby concealing crucial symptoms within black-box models. Compared to the severity of facial paralysis, the symptoms of facial paralysis are of greater significance to both physicians and patients. To address this issue, this paper proposes a facial paralysis symptom detection model based on facial action units (AUs). To enhance the accuracy of AU intensity prediction, a novel Difference Ensemble Method (DEM) is introduced. This method leverages differential information between frames within the same video to improve the accuracy of predictions for the current frame. Building upon the predicted AU intensity sequences for keyframes in a video, an interpretable model for detecting facial paralysis symptoms is designed. This model employs an active means to describe the asymmetry in facial muscle strength and utilizes co-occurrence matrices to detect synkinesis. It is noteworthy that DEM is exclusively trained on a dataset of normal faces but exhibits excellent performance when transferred to a facial paralysis dataset. Additionally, DEM exhibits higher accuracy in predicting AU intensity compared to existing methods. The F1 scores for detecting facial muscle function in the eyebrow, eye, and mouth regions with our proposed model are 80.0%, 79.23%, and 90.91%, respectively. To demonstrate the model’s performance, a synkinesis detection experiment is conducted, further validating its applicability in facial paralysis detection.

Training task-oriented dialog agents based on reinforcement learning is time-consuming and requires a large number of interactions with real users. How to grasp dialog policy within limited dialog experiences remains an obstacle that makes the agent training process less efficient. In addition, most previous frameworks start training by randomly choosing training samples, which differs from the human learning method and hurts the efficiency and stability of training. Therefore, we propose Scheduled Curiosity-Deep Dyna-Q (SC-DDQ), a curiosity-driven curriculum learning framework based on a state-of- the-art model-based reinforcement learning dialog model, Deep Dyna-Q (DDQ). Furthermore, we design ed learning schedules for SC-DDQ and DDQ, respectively, following two opposite training strategies: classic curriculum learning and its reverse version. Our results show that by introducing scheduled learning and curiosity, the new framework leads to a significant improvement over the DDQ and Deep Q-learning (DQN). Surprisingly, we found that traditional curriculum learning was not always effective. Specifically, according to the experimental results, the easy-first and difficult-first strategies are more suitable for SC-DDQ and DDQ. To analyze our results, we adopt ed the entropy of sampled actions to depict action exploration and found that training strategies with high entropy in the first stage and low entropy in the last stage lead to better performance.

Change detection (CD) methodologies have garnered substantial attention owing to their capability to monitor alterations in geographic spaces across temporal intervals, especially with the acquisition of high-resolution remote sensing images. However, challenges persist due to dissimilar imaging conditions and temporal windows. Although deep-learning architectures have shown promise in addressing challenges in CD, many existing methods struggle to capture long-range dependencies and local spatial information effectively. The current CD methods rely heavily on pure CNNs and transformers, which employ only single-pass forward propagation. This approach leads to inadequate utilization of feature information, resulting in inaccurate CD maps, particularly when discerning edges. To overcome these limitations, we propose a transformer-based conditional generative diffusion method for CD, named TransC-GD-CD, tailored for RS data. This approach leverages the numerous sampling iterations of the DDPM, contributing to the generation of high-quality CD maps. In addition, the frequency cross transformer mechanism seamlessly amalgamates CD condition with the noise feature within the DDPM. The innovative mechanism effectively bridges diffusion noise and conditional semantic terrains. Moreover, a novel multitype difference extraction module, named appear–disappear–concat, is devised to partition the CD task to optimize both segmentation extraction and CD classification, overcoming the persistent challenge of information loss endemic to conventional CD algorithms, such as simple subtraction. We demonstrate the superiority of TransC-GD-CD by comparing the experiment results against various algorithms across three widely used CD datasets, namely CDD, WHU, and LEVIR.

Precipitation nowcasting refers to the prediction of small-scale precipitation events at minute and kilometer scales within the upcoming 0 to 2 h, which significantly impacts both human activities and daily life. However, prevailing deep learning models have primarily focused on a single radar echo data source, limiting their ability to effectively capture intricate and rapidly evolving precipitation patterns. Thus, meteorological satellite is considered to supplement radar echo data. To achieve a comprehensive integration of multisource data with enhanced details, a two-stage fusion satellite and radar GAN-based prediction network (named FsrGAN) is proposed. In the first stage, we design a satellite-radar fusion prediction network known as FsrNet. This network employs an encoder-fusion-decoder architecture, where a novel spatial-channel attention (SCA) is proposed to enhance the filtering and fusion of multisource and multiscale features. In the second stage, we introduce a GAN-based network (FusionGAN) that also mines the complementary information of satellite images to sharpen the first-stage predicted radar maps with more details. Experiments are conducted on meteorological dataset in the Yangtze River Delta (YRD) region. The test results exhibit the notably superior performance of our model in terms of image quality and precipitation forecasting metrics in comparison to traditional optical flow-based methods and some well-known deep learning methods (ConvLSTM, ConvGRU, TrajGRU and PredRNN++). More importantly, our fusion model using satellite and radar data demonstrates the ability to predict convective initiation.

Vertical extrinsic photoconductive semiconductor switches (PCSSs) are presented with initial characteristics comparison between V-doped 4H-SiC and Fe-doped GaN PCSS under axial triggering such as dark resistance, photoconductivity, power output, and breakdown behavior. Experiments are carried out under the 532-nm-wavelength laser with mJ-level energy and a pulse width of 30 ns. Photoconductive experiments show that the photoelectric conversion efficiency of GaN PCSS is 2.27 times higher than 4H-SiC PCSS with the same electric field strength under different laser energies from 1 mJ to 5 mJ. 4H-SiC PCSS with a dark-state resistance of $10^{12} \Omega \cdot $ cm can withstand a bias voltage of 8 kV (16 kV/mm) and laser energy of 8 mJ and the maximum output power is up to 428.7 kW, while that of GaN can only stand a bias voltage of 1 kV (2.9 kV/mm) because of low dark resistance and defect. Obvious cracks of 4H-SiC PCSS can be observed from the breakdown image after breakdown occurs, while the dark-state resistance of GaN PCSS drops from $10^{6} \Omega \cdot $ cm to $10^{4} \Omega \cdot $ cm under high DC voltage.

The variability in the propagation pathway in epilepsy is a main factor contributing to surgical treatment failure. Ways to accurately capture the brain propagation network and quantitatively assess its evolution remain poorly described. This work aims to develop a dynamic step effective network (dSTE) to obtain the propagation path network of multiple seizures in the same patient and explore the degree of dissimilarity. Multichannel stereo-electroencephalography (sEEG) signals were acquired with ictal processes involving continuous changes in information propagation. We utilized high-order dynamic brain networks to obtain propagation networks through different levels of linking steps. We proposed a dissimilarity index based on singular value decomposition to quantitatively compare seizure pathways. Simulated data were generated through The Virtual Brain, and the reliability of this method was verified through ablation experiments. By applying the proposed method to two datasets consisting of 29 patients total, the evolution processes of each patient’s seizure networks was obtained, and the within-patient dissimilarities were quantitatively compared. Finally, three types of brain network connectivity patterns were found. Type I patients have a good prognosis, while type III patients are prone to postoperative recurrence. This method captures the evolution of seizure propagation networks and assesses their dissimilarity more reliably than existing methods, demonstrating good robustness for studying the propagation path differences for multiple seizures in epilepsy patients. The three different patterns will be important considerations when planning epilepsy surgery under sEEG guidance.

This article introduces an innovative method to improve the accuracy of estimating the Direction of Arrival (DOA) by utilizing non-uniform linear arrays (NULAs) operating at harmonic frequencies. Common DOA techniques face challenges in maintaining sufficient degrees of freedom (DOFs) while minimizing the number of sensors. Our approach takes advantage of the extended array aperture of NULAs and the diverse frequencies involved in harmonics to address these challenges. We explore various NULA structures, such as nested arrays, coprime arrays, and coprime arrays with compressed inter-element spacing (CACIS), each with unique characteristics and requirements for enhancing DOFs at the third and fifth harmonics. Simulation results demonstrate a substantial increase in DOFs, emphasizing the effectiveness and adaptability of our method in improving DOA estimation across different NULA configurations. This research contributes not only to theoretical advancements in antenna array processing but also offers practical insights, showcasing the potential of our method to significantly enhance DOA estimation accuracy in complex signal environments.

Runway friction coefficient is a crucial parameter that reflects the airworthiness of airport runway and measuring the runway friction coefficient is a prerequisite for ensuring flight safety. Many studies have been conducted on the measurement methods for runway friction coefficient, but most of them are still at the theoretical research level. Although new methods and ideas are constantly being proposed, they are still in the conceptual design and exploration stage. A method based on numerical simulation for runway friction coefficient measurement is proposed in this paper. Firstly, the numerical analysis method is used to explore the interaction mechanism between tire and runway. Based on finite element analysis (FEA), multi-physics field coupling analysis is carried out to reveal the influence of different working conditions on friction coefficient, and to obtain the variation law. Secondly, a friction coefficient measurement method is proposed based on the analysis results, and the friction coefficient estimation models are constructed to achieve a new measurement method that only relies on the tread frictional stress to estimate the friction coefficient with specific working condition. Finally, a friction coefficient measurement system is constructed and used to validate the method. The experimental results indicate that the error between the estimated and actual measurements is from 2.47% to 4.13%. The experiments have verified the effectiveness and accuracy of the estimation model based on numerical analysis method, providing ideas for researching runway friction coefficient measurement methods.

The wide applications of wide-bandgap (WBG) devices in power electronics systems bring benefits in higher switching speed, frequency, and power density, but also cause severe electromagnetic interference (EMI) issues. While many EMI mitigation methods are available to reduce the noise generated from the source (i.e., switching WBG devices), it is equally important to analyze and control the EMI propagation paths. In WBG systems, the parasitic and coupling effects become dominant due to the high switching frequency and the compact system size, which creates uncontrolled propagation paths that degrade the systems’ EMI performance significantly. Therefore, it is crucial to investigate the parasitic and coupling effects and develop corresponding mitigation techniques. This review focuses on analyzing the EMI propagation paths to provide a comprehensive guideline to identify possible parasitics and couplings in the power module package level, EMI filter level, and the power electronics system level. The corresponding mitigation techniques for common-mode and differential-mode conducted EMI as well as radiated EMI are summarized, and the remaining challenges and future research topics are also discussed for all three levels in the conclusion.

This article introduces a research method aimed at achieving optimal control timing for the zero-voltage quasi-square wave switch (ZVS-QSW) in a silicon carbide buck converter across a wide range. The method establishes the correlation function between optimal control timing parameters and load current through composite segmented curve fitting to the plane of the ZVS-QSW state over the entire operating range. It also compensates for the charge change on the filter capacitor to determine the load current at the corresponding operating point. By collecting the input/output voltages and load currents and substituting them into the correlation function, the optimal timing parameters for achieving ZVS-QSW can be determined online. The proposed method is validated on a silicon carbide buck converter prototype with an input voltage of 400 V. ZVS-QSW can be achieved in the conversion ratio range of 0.1-0.9. Consequently, the proposed method not only reduces the difficulty of inductance current measurement but also simplifies and visualizes the timing control of the converter.

Seeds are the most basic and important means of production for agriculture. During the production and processing of seeds, they may undergo potential mechanical damages and mildew alterations, which might jeopardize their germination viability. Hence, checking the quality of seeds before sowing is of paramount importance for the benefit of the sower and the safety of agricultural production. In order to achieve an efficient detection of maize seed quality, our experiment assembled a dataset composed of 2,128 seeds with four different health statuses of maize: healthy, broken, moth-eaten, and mildewed. In this paper, we proposed a lightweight maize seed quality detection model for small objects based on improved YOLOv8: I-YOLOv8. Firstly, we introduced a multi-scale attention mechanism called EMA to efficiently retain information across channels and reduce computational load. Next, we chosen the SPD-Conv module for low-resolution images and small objects, and applied it to the backbone, which addressed the loss of fine-grained information and the less efficient learning of feature representations present in YOLOv8. Lastly, we reduced the large detection layer, which directed the network to pay more attention to the location, channel, and dimensional information of smaller objects, and we also replaced the loss function with WIoUv3. We validated our model using ablation studies and compared it with YOLOv5, YOLOv6, and YOLOv8. The mAP (Mean Average Precision) of the improved model I_YOLOv8 reaches 98.5%, which is 6.7% higher than YOLOv8. The average recognition time per image was 163.9fps, a boost of 5.2fps compared to YOLOv8. This study lays a theoretical foundation for the efficient, convenient, and rapid detection of maize quality, while also offering a technical basis for advancing automated maize quality detection means.

The application of multisampling technology in voltage-source inverters (VSIs) systems can significantly reduce the control delay and break the bandwidth limitations of traditional synchronous sampling schemes. However, when sampling the capacitor voltage, the high-frequency ripple is introduced into the control loop, leading to the local vertical crossing (VC) phenomenon. Previous studies have demonstrated that a moving average filter (MAF) or improved repetitive filter (IRF) can effectively remove the high-frequency ripple. However, they introduce additional phase lag, which undermines the advantages of multisampling technology in terms of phase boost. The approach of spectrum analysis can be used to study the impact of voltage ripple. The results indicate that after modulation, the output signal contains low-frequency harmonics, and the superposition of the ripple makes the modulation process more susceptible to saturation. Therefore, a 20k ripple notch filter (RNF) is proposed to attenuate the ripple signal while introducing a more minor phase lag. Additionally, a 90° phase shift filter is proposed to suppress the aliasing signals generated in the system. Finally, the effectiveness of the proposed methods is verified through the experimental test of a 10kW single-phase VSI system.

Considering the significant contribution of the transport sector to carbon emissions, the importance of battery electric vehicles (BEVs) as environmentally friendly vehicles is self-evident. Due to the rapid expansion of the BEV market in recent years, a comprehensive evaluation of BEV options from the consumer perspective has become an important issue. This paper proposes a data-driven decision aids for purchasing BEVs based on a multiple criteria decision-making methodology (i.e., PROMETHEE-II). A hierarchical evaluation criteria system of BEVs is constructed and correlation analysis between indicators is performed to eliminate duplicate indicators. Then, a comprehensive weighting method by integrating large-scale group decision making method and the Entropy-based method is proposed to identify the weights of criteria. Based on which, the ranking of candidate BEVs can be obtained based on the PROMETHEE-II with a hierarchical evaluative criteria, which can help consumers make BEV purchase choices. Furthermore, the robustness and reliability of the results are tested by applying the sensitivity analysis and contrastive analysis.

Currently, incidents of personal accidents occur frequently in industrial fields, and falling is one of the most common safety hazards. This makes fall detection a research area of significant importance. Timely responses to fall events can significantly reduce the harm caused. However, most of the available fall detection models often suffer from issues such as insufficient detection accuracy or high parameter and computational requirements, making them challenging to deploy on local devices. In response, this paper introduces an enhanced convolutional neural network model, YOLOv7-fall, aimed at promptly detecting fall incidents. Firstly, the paper proposes a novel attention module, SDI, based on the Coordinate Attention and Shuffle Attention. This module enhances the feature extraction capabilities for detecting targets. Secondly, the inclusion of GSConv and VoV-GSCSP modules in the model’s head section is in order to reduce model parameters and computational complexity, making it more suitable for deployment. Thirdly, by replacing the conventional $3\times 3$ convolution in the final ELAN(Efficient Layer Aggregation Networks) module of the Backbone with the DBB (Diverse Branch Block), the model captures features from different layers or types in the image, increasing the network’s diversity. Experimental results demonstrate that YOLO-fall improves mAP by 2.7% compared to YOLOv7-tiny while reducing model parameters by 3.5% and computational requirements by 5.4%. In comparison to existing detection algorithms under similar conditions, YOLO-fall achieves more precise and lightweight capabilities.

The inadequate automation level in the transit of oil drilling tubular columns has led to significant inefficiencies and safety issues. To address these challenges, a real-time detection algorithm, ECS-YOLOv5s, has been proposed. This algorithm aims to improve the accuracy of drill pipe identification during operational processes, facilitating the automation of tubular column handling It has the potential to reduce drilling cycles and overall drilling costs. ECS-YOLOv5s enhance the detection accuracy of drill pipes by incorporating a Bidirectional Feature Pyramid Network (BiFPN) architecture with an improved multi-scale feature fusion network. The use of EfficientNet as the backbone network reduces the number of parameters and computations while effectively merging features from different layers. Additionally, the Spatial Pyramid Pooling (SPP) structure in the Neck is replaced with SPPF, and a Convolutional Block Attention Module (CBAM) is introduced to improve model robustness, reduce parameters and computations, and enhance the model’s ability to detect dense targets. The ECS-YOLOv5s algorithm exhibits superior performance in drill pipe inspection, achieving a mean Average Precision (mAP) of 90.2%, a frame rate of 125 FPS, and a parameter count of only 37%. It achieves an accuracy of 98.6%, outperforming the original model by 9.2%. The comparative analysis demonstrates that the improved algorithm surpasses traditional models such as YOLOv5s, SSD, Faster-RCNN, and YOLOv7-tiny in both performance and accuracy. These findings provide valuable insights for the research on automated processing of tubular columns in intelligent oil drilling platforms.

Landslides are catastrophic geological events that can cause significant damage to properties and result in the loss of human lives. Deep-learning technology applied to optical remote sensing images can enable effective landslide-prone area detection. However, conventional landslide detection (LD) models often employ complex structural designs to ensure detection accuracy. The complexity often hampers the detection speed, rendering these models inadequate for the swift emergency monitoring of landslides. To address these problems, we propose a new lightweight deep-learning-based framework, BisDeNet, for efficient LD. To improve the efficiency of the proposed BisDeNet, we replaced the context path in the original BiSeNet with DenseNet due to its strong feature extraction ability, few required parameters, and low model complexity. Two sites with different and representative landslide developments were selected as the study areas to verify the performance of our proposed BisDeNet. Additionally, we introduced landslide causative factors to enhance the sampling dataset. To evaluate the effectiveness of our approach, we compared the performance of our BisDeNet with the performances of three other BiSeNet-based methods and an advanced transformer-based model data-efficient image transformer (DeiT). Our experimental results indicate that the F1-scores of BisDeNet in the two study areas are 0.9006 and 0.8850, which are 26.22% and 1.86% higher than the scores of BiSeNet, respectively, but slightly lower than that of the DeiT model. Furthermore, our proposed BisDeNet requires the fewest number of parameters and the least memory out of the five models.

Given the widespread occurrence of obesity, new strategies are urgently needed to prevent, halt and reverse this condition. We proposed a noninvasive neurostimulation tool, ultrasound deep brain stimulation (UDBS), which can specifically modulate the hypothalamus and effectively regulate food intake and body weight in mice. Fifteen-min UDBS of hypothalamus decreased 41.4% food intake within 2 hours. Prolonged 1-hour UDBS significantly decreased daily food intake lasting 4 days. UDBS also effectively restrained body weight gain in leptin-receptor knockout mice (Sham: 96.19%, UDBS: 58.61%). High-fat diet (HFD) mice treated with 4-week UDBS (15 min / 2 days) reduced 28.70% of the body weight compared to the Sham group. Meanwhile, UDBS significantly modulated glucose-lipid metabolism and decreased the body fat. The potential mechanism is that ultrasound actives pro-opiomelanocortin (POMC) neurons in the hypothalamus for reduction of food intake and body weight. These results provide a noninvasive tool for controlling food intake, enabling systematic treatment of obesity.

Alzheimer’s Disease (AD) is a significant cause of dementia worldwide, and its progression from mild to severe affects an individual’s ability to perform daily activities independently. The accurate and early diagnosis of AD is crucial for effective clinical intervention. However, interpreting AD from medical images can be challenging, even for experienced radiologists. Therefore, there is a need for an automatic diagnosis of AD, and researchers have investigated the potential of utilizing Artificial Intelligence (AI) techniques, particularly deep learning models, to address this challenge. This study proposes a framework that combines a Vision Transformer (ViT) and a Gated Recurrent Unit (GRU) to detect AD characteristics from Magnetic Resonance Imaging (MRI) images accurately and reliably. The ViT identifies crucial features from the input image, and the GRU establishes clear correlations between these features. The proposed model overcomes the class imbalance issue in the MRI image dataset and achieves superior accuracy and performance compared to existing methods. The model was trained on the Alzheimer’s MRI Preprocessed Dataset obtained from Kaggle, achieving notable accuracies of 99.53% for 4-class and 99.69% for binary classification. It also demonstrated a high accuracy of 99.26% for 3-class on the AD Neuroimaging Initiative (ADNI) Baseline Database. These results were validated through a thorough 10-fold cross-validation process. Furthermore, Explainable AI (XAI) techniques were incorporated to make the model interpretable and explainable. This allows clinicians to understand the model’s decision-making process and gain insights into the underlying factors driving the AD diagnosis.

Though the forearm is the focus of the prostheses, myoelectric control with the electrodes on the wrist is more comfortable for general consumers because of its unobtrusiveness and incorporation with the existing wrist-based wearables. Recently, deep learning methods have gained attention for myoelectric control but their performance is unclear on wrist myoelectric signals. This study compared the gesture recognition performance of myoelectric signals from the wrist and forearm between a state-of-the-art method, TDLDA, and four deep learning models, including convolutional neural network (CNN), temporal convolutional network (TCN), gate recurrent unit (GRU) and Transformer. It was shown that with forearm myoelectric signals, the performance between deep learning models and TDLDA was comparable, but with wrist myoelectric signals, the deep learning models outperformed TDLDA significantly with a difference of at least 9%, while the performance of TDLDA was close between the two signal modalities. This work demonstrated the potential of deep learning for wrist-based myoelectric control and would facilitate its application into more sections.

To mitigate the temperature dependence of the resonance frequency in microcavities, athermal photonic devices have been developed by incorporating materials with an opposite thermo-optical coefficient (TOC). Here, we conduct an experimental demonstration of the athermal effect in a microbubble resonator, employing an aluminum (Al) holder for its mounting. By converting the thermal expansion of the holder into controlled geometry adjustments in the microbubble resonator, we achieve athermal optical modes at a specific temperature. The athermal compensation is accomplished by counterbalancing the pulling force resulting from the thermal expansion of the Al holder against the thermal characteristics of the material, specifically $\mathrm{SiO_{2}}$. Especially, the optical modes located around the equator are particularly susceptible to the influence of the pulling force exerted by the Al holder. This temperature-insensitive feature of the resonance establishes a new avenue towards athermal microresonator through this special structure.

Recurring green tides (also called Ulva prolifera) cause significant damage to marine ecosystems in the Yellow Sea of China, especially in the settlement region. The settlement region is a critical area for the natural decay and decomposition of green tides, which obviously has ecological consequences. Recent studies in relation to this topic are mostly based on point observation data, which prevents to quantitatively analyze the ecological impact patterns of green tide biomass at large spatial scales. Therefore, we used remote sensing time-series of Geostationary Ocean Color Imager satellite data combined with in situ data to invert marine chlorophyll-a (Chl-a, main indicators representing phytoplankton biomass) concentrations by machine learning methods. Finally, based on a cross-satellite model, we quantified the green tide biomass that occurred in Haiyang City during 2015 and 2016. The main results found are as follows. First, the green tide biomass reveals negative correlations on Chl-a concentration in the settlement region. The Chl-a concentration showed a decreasing trend and remained at a low level (0.1–0.2 mg/m3) when the biomass of the green tide increased. After the disappearance of the green tides, the Chl-a concentration accreted rapidly and then began to gradually decrease. Second, combined with the pixel-based statistics grid of the green tide and the average Chl-a concentration, the time cycle of green tides in the settlement region is about 30–35 d. Finally, some special cases (such as typhoon) can change the pattern and temporal cycle in the settlement region. This article provides support for marine ecosystem monitoring.

Mesh division plays an important role in applications of the finite element method (FEM). The proposed research shows that under the same order, the equilateral triangular meshes have the most uniform dispersion distribution. The isosceles triangles with equal base and height have more uniform dispersion error than the square meshes, while the maximum phase error is similar. Taking the rectangular waveguide as an example, the relative errors in the cut-off frequency are analyzed based on different meshes. The numerical results show that under the same interpolation order and node numbers, the relative error of isosceles triangles with equal base and height for TE10 mode is the smallest. The results are useful in choosing appropriate element order, node density and mesh shape when applying FEM.

To address the issue of low predictive accuracy in complex trajectory forecasting for civilian rotorcraft unmanned aerial vehicles (UAVs), this paper presents a method that utilizes the SVM-BTS technique for recognizing and predicting these intricate trajectories. Initially, the Support Vector Machine-Binary Tree Support Vector Machine model (SVM-BTS) is employed to segmentally recognize the complex trajectories of civilian UAVs. Based on this identification, five distinct flight states are identified: vertical, pitch, transverse, roll, and transitional. To assess the predictive performance of these states, a combination of Sliding Window Polynomial Least Squares, Unscented Kalman Filtering, and Long Short-Term Memory neural network methods is utilized. Consequently, the most suitable prediction algorithm is determined for each flight state. Experimental results demonstrate that the SVM-BTS recognition method, compared to SVM, achieves a 10.4% increase in recognition accuracy. Across different flight datasets, this prediction method exhibits the lowest mean squared error values compared to SWPLS, UKF, and LSTM. Therefore, this study accurately predicts the complex flight trajectories of civilian rotorcraft UAVs, enhancing the precision and efficiency of UAV flight prediction.

In wireless communication wiretap channel, for the eavesdropper to obtain the legitimate receiver decoding rules situation, this paper proposes a LT-SLT fountain code anti-eavesdropping channel coding design. This method targets the Luby Transform (LT) code transmission for some of the original symbols of the Shifted LT (SLT) code and utilizes the fountain code to receive the correct symbols in different noise channels with differential characteristics so that the decoding process of the eavesdropper changes and cannot be decoded in synchronization with the legitimate receivers, and then the partial symbols of the recovered source are different from those of the legitimate receivers. When these symbols continue to participate in SLT decoding, increasing the untranslated rate of the eavesdropper. Experimental results show that although the method proposed in this paper increases the number of decoded symbols by a small amount, the eavesdropper untranslated rate of this scheme gets improved by about 15% when the main channel or the wiretap channel is varied individually, compared with LT code and SLT code. When both the main channel and the wiretap channel are varied simultaneously, the untranslated rate of the eavesdropper in this scheme gets approximately 30% higher compared to LT code and SLT code, and the untranslated rate of the eavesdropper in this scheme gets approximately 14% higher compared to SLT-LT fountain code. When the main channel is worse or slightly better than the wiretap channel, the untranslated rate of eavesdroppers of this scheme is better than that of SLT-LT fountain code, which effectively ensures secure transmission of information.

The lightweight, small diameter, and irregular shape of small vegetable seeds create difficulties for online monitoring of sowing quality. We propose a machine vision-based online monitoring method with a sowing test bench designed to address the challenges. Vision devices and image processing systems are employed to detect the quality of seed sowing. Firstly, the seed segmentation image is obtained by completing the steps of median filtering, graying and image segmentation. We then implement the Circumscribed circle method to detect the position of the seed. Afterward, the coordinate system is converted using calibrated results to eliminate non-seed impurities. Finally, we count the number of identified seeds to evaluate the recognition accuracy. The trial compared three algorithms: the image segmentation algorithm OTSU, the critical point localization algorithm SIFT, and the algorithm designed in the experiment. The algorithm we designed outperformed the others regarding recognition accuracy and processing time. The experimental method employed in the study encompasses various functionalities, including seeding counting, understanding detection, replaying, and monitoring deviations from seed bands during sowing. Cabbage seeds (1.50 mm–2.00 mm), tomato seeds (1.00 mm–1.50 mm), and radish seeds (0.50 mm–1.00 mm) were selected as the experimental subjects due to the uniform particle size distribution. The results demonstrate that the relative error between the online image recognition algorithm and the system’s seeding rate monitoring is below 3.0%. Moreover, the accuracy of missed seeding monitoring is 92.5%, while the accuracy of deviation monitoring during seeding is 92.0%. We observed that the image recognition algorithm employed in the system achieved a processing time of 0.29 seconds, with a seed band recognition rate of 96.8%, fulfilling the monitoring requirements for small seed sowing experiments. The processed images and collected data are presented in real-time on the upper computer terminal. This study significantly contributes to the advancement of small-grain vegetable seed sowing monitoring technology.

The surface electromyography (sEMG) signals contain a large amount of physiological information reflecting the body’s motor intention. The use of sEMG signals for gesture recognition has received a lot of attention in both robotics and rehabilitation fields. Most of the current studies on gesture recognition based on sEMG signals obtain discrete gestures by classification, ignoring the continuous natural motion. In this paper, a continuous gesture recognition and force estimation method is proposed based on sEMG signals. To establish a foundation for this approach, sEMG sensors are thoughtfully positioned on the forearm’s surface, guided by considerations of physiological structure and muscular function. The finger curvature is proposed to describe the gesture state, and the gesture changes at different moments can be represented by the set of finger curvatures of different fingers, thus achieving continuous gesture recognition. Muscle force estimation was performed while recognizing gestures under different force partterns. A multi-stream convolutional neural network (MSCNN) is used to model finger curvature with sEMG to achieve gesture recognition, and Long short-term memory (LSTM) is used for muscle force estimation due to it is able to capture the temporal relationship during muscle force generation. The experimental results show that the average Pearson correlation coefficient (CC) and root-meansquare error (RMSE) of gesture recognition are 0.84 and 0.11, respectively, and the coefficient of determination $(R^{2}) $ of muscle force estimation is 0.92. The whole scheme achieves the simultaneous estimation of gesture and muscle force, which can well meet the needs in the field of prosthetic control and rehabilitation assistance.

Self-supervised learning (SSL) has emerged as a promising approach for learning representations from unlabeled data. Momentum-based contrastive frameworks such as MoCo-v3 have shown remarkable success among the many SSL methods proposed in recent years. However, a significant gap in encoder representation exists between the online encoder (student) and the momentum encoder (teacher) in these frameworks, limiting the performance on downstream tasks. We identify this gap as a bottleneck often overlooked in existing frameworks and propose “residual momentum” that explicitly reduces the gap during training to encourage the student to learn representations closer to the teacher’s. We also reveal that a similar technique, knowledge distillation (KD), to reduce the distribution gap with cross-entropy-based loss in supervised learning is useless in the SSL context and demonstrate that the intra-representation gap measured by cosine similarity is crucial for EMA-based SSLs. Extensive experiments on different benchmark datasets and architectures demonstrate the superiority of our method compared to state-of-the-art contrastive learning baselines. Specifically, our method outperforms MoCo-v3 0.7% top-1 in ImageNet, 2.82% on CIFAR-100, 1.8% AP, and 3.0% AP75 on VOC detection pre-trained on the COCO dataset; it also improves DenseCL with 0.5% AP (800ep) and 0.6% AP75 (1600ep). Our work highlights the importance of reducing the teacher-student intra-gap in momentum-based contrastive learning frameworks and provides a practical solution for improving the quality of learned representations.

In the context of generative artificial intelligence (AI), AIGCP (content generation-based AI products), represented by ChatGPT, have attracted extensive attention in the field of education. This study focuses on the discipline of university operating systems and adopts the Rasch model as the theoretical foundation. By combining ChatGPT with existing question banks and using the bidirectional fine-grained table method, it compiles questions that match the corresponding abilities for three different levels of student groups. This aims to explore personalized question matching and student cognitive ability analysis methods to support personalized teaching. The research findings indicate that ChatGPT is capable of matching exercises of similar difficulty under the Rasch model, but its accuracy in generating exercise content is relatively low, and the variety of exercise content is limited. Students’ performance in overall competency requires improvement. This study aims to leverage the combined strengths of ChatGPT and traditional educational assessment methods to introduce an innovative approach to support personalized instruction. It aims to establish the routine utilization of exercise creation by ChatGPT and personalized analysis of student cognitive abilities, thereby better fulfilling the demands of education within the classroom setting.

Owing to the increased demand for autonomous driving and advanced driver assistant systems, autonomous parking configurations have been widely researched. In order to solve the problem of path tracking accuracy degradation, caused by ignoring the external interference in the actual parking process and the uncertainty of vehicle steering modeling, a parking path tracking algorithm, combining Sliding Model Control (SMC) and Extended State Observer (ESO), is proposed. First, the vehicle kinematics model is established, which includes many external interference factors, such as lateral position, speed, heading angle, uncertainty parameters and time delay of steering mechanism. Considering the vehicle physical constraints and boundary condition constraints, during the parking process, the parking reference path is designed according to reverse outbound driving, while the parking collision analysis and path smoothing are further carried out. Next, an ESO is designed to observe and compensate the external disturbances and model uncertainties, treating these as the total disturbances of the system. On this basis, a SMC for parallel parking path tracking is designed. The observed value of ESO is used as the compensation in the SMC, to weaken the influence of external interference. Finally, using Matlab simulation, the feasibility and effectiveness of the proposed path planning and tracking control method are verified. The simulation results show that, the maximum tracking error of lateral position is less than 0.01m, whereas the maximum tracking error of heading angle is less than 2.5°. Comparing two control strategies, the control effect of the designed SMC with ESO is better than that of the traditional SMC controller, while the ability of resisting external interference is stronger. In addition, real vehicle tests are carried out, to verify the effectiveness of the proposed control method. The test results show that, the vehicle can safely park in the parking space, quickly and accurately, even if there is external interference. The proposed method can produce the parking trajectory according to the given constraints and control the vehicle to complete the parking operation accurately along the planned path.

Target detection technology has been greatly improved for synthetic aperture radar (SAR) images recently, due to the advancement in the deep learning domain. However, because of the existence of clutter in the SAR images, it is still a challenge to detect small targets with high accuracy and low computational complexity. To solve this problem, a detection algorithm based on a feature fusion and cross-layer connection network is proposed in this article. First, attention feature fusion is applied to improve the feature fusion ability for the small targets by allocating weights to various feature maps adaptively. Meanwhile, the depthwise separable convolution (DW-Conv) is used to reduce the computational complexity caused by the increasement of network layers. Then, a cross-layer connection (Cross-Connect) submodule is proposed to fuse shallow features with deep features further. Finally, a multiscale target detection (Multi-Detect) submodule is designed to improve the detection ability for small targets. We compare the proposed algorithm with the other representative methods on the SAR-Ship-Dataset and SSDD, quantitative evaluations show that our proposed algorithm can reach the highest computational efficiency. Therefore, because of the superior performance in terms of accuracy and efficiency, the algorithm proposed in this article is more suitable to detect small targets for the SAR images.

It is generally accepted that the vibration and noise of electrical motors mainly depend on radial electromagnetic forces. However, the relationship between electromagnetic forces and system modes becomes special when integrating the electrical motor into a coaxial electric axle (eAxle). This paper proposes a phase-based tangential electromagnetic force array analysis method to explain and optimize the noise spikes in a coaxial eAxle which is driven by a permanent magnet synchronous motor (PMSM) with a step-skewed rotor. Its electromagnetic force on each stator node is decomposed into radial and tangential forces as inputs for structural finite element simulation. Based on their contributions to the noise, the torsional vibration and noise of the eAxle are correlated with the tangential electromagnetic force array. Thus, the coupling effect of excitation and response is illustrated, and the electromagnetic force array could be optimized more targeted.