Search

Individuals with high-level spinal cord injuries often face significant challenges in performing essential daily tasks due to their motor impairments. Consequently, the development of reliable, hands-free human-computer interfaces (HCI) for assistive devices is vital for enhancing their quality of life. However, existing methods, including eye-tracking and facial electromyogram (FEMG) control, have demonstrated limitations in stability and efficiency. To address these shortcomings, this paper presents an innovative hybrid control system that seamlessly integrates gaze and FEMG signals. When deployed as a hybrid HCI, this system has been successfully used to assist individuals with high-level spinal cord injuries in performing activities of daily living (ADLs), including tasks like eating, pouring water, and pick-and-place. Importantly, our experimental results confirm that our hybrid control method expedites the performance in pick-place tasks, achieving an average completion time of 34.3 s, which denotes a 28.8% and 21.8% improvement over pure gaze-based control and pure FEMG-based control, respectively. With practice, participants experienced up to a 44% efficiency improvement using the hybrid control method. This state-of-the-art system offers a highly precise and reliable intention interface, suitable for daily use by individuals with high-level spinal cord injuries, ultimately enhancing their quality of life and independence.

Interest point detection and matching play an extremely important role in many computer vision applications. In recent years, deep learning-based algorithms are being used for interest point detection, of which SuperPoint is the most notable algorithm. Although SuperPoint has achieved good results, its interest point detection accuracy is limited to the pixel level; additionally, the simple rounding of noninteger ground truth coordinates in this algorithm results in the loss of decimal information, introducing quantization errors. To overcome these limitations, this study introduces the subpixel module and randomized round encoding methods to reduce the prediction error and quantization error in SuperPoint, respectively. First, we propose a differentiable decoder, soft-random to achieve subpixel-level accuracy of interest point detection. Additionally, to further reduce the interest point localization error, we propose the reprojection homography adaptation of the training step based on SuperPoint’s homography adaptation that is, weincorporate the reprojection error into the algorithm’s training loss. The optimized algorithm proposed in this study, called SRSP, is tested on the HPatches dataset and Euroc dataset. The results showed that SRSP performed better than SuperPoint regarding all indicators, indicating the effectiveness of SRSP on both image matching dataset and real-world dataset.

Most of the previous research about unmanned surface vehicle (USV) control system just focuses on reaching the target but not considers berthing. Moreover, computation cost is rarely concerned. For automatic target arriving and berthing of USV in presence of multi-shape convex obstacles and disturbances, this paper designs a composite control system including trajectory optimization and trajectory tracking. In order to generate a smooth and optimal trajectory, two objectives including minimum running time and energy consumption are proposed, and their priority requirement is presented. Fuzzy satisfactory optimization is introduced to reformulate the objective functions, and the membership degree difference is used to model priority between two objectives. The obstacle avoidance constraint is developed based on the improved P-criterion. The comprehensive optimization model is established to obtain the maximum membership degree. Further, the priority optimization model is proposed to expand the feasible domain by relaxing the comprehensive optimization result. Two models are discretized by Gaussian pseudospectral method to find the optimal solution. To achieve good trajectory tracking under disturbances, two first-order active disturbance rejection control (ADRC) controllers are designed to follow the desired longitudinal velocity and course angular velocity. All the velocities of USV can converge. The simulation proves the effectiveness of the proposed co-designed system by comparing with other methods.

Underwater vehicles are crucial for advancing underwater geological exploration, military training, and the marine industry. However, the accuracy of lithium-ion battery SOC estimation algorithms for these vehicles deteriorates with depth changes due to variations in seawater temperature, leading to limited generalization. This paper proposes a SOC estimation framework that combines the UKF with the MLP-Mixer neural network for lithium-ion batteries in underwater vehicles. The proposed framework employs the MLP-Mixer neural network to predict battery terminal voltage, incorporating hierarchical temperature compensation based on temperature data from the BMS. This straightforward and efficient compensation method enhances prediction accuracy. Subsequently, the framework combines the MLP-Mixer neural network with the UKF to form a new estimator, MM-UKF. The MM-UKF estimator’s prediction accuracy, temperature adaptability, and generalization ability underwent comprehensive examination and evaluation using a public dataset under varying temperatures and complex operating conditions. Results indicate that: 1) The superior ability of the MLP-Mixer neural network to observe battery terminal voltage plays a significant role in the precise SOC estimation of MM-UKF; 2) The proposed method shows robustness and performs well under different driving cycles. The RMSE of SOC estimation does not exceed 0.01 under intense driving cycles, showcasing superior generalization ability; 3) While the estimator’s prediction accuracy deviates slightly from the reference value at low temperatures and SOC levels, it outperforms prediction methods based on MLP and LSTM neural networks in various temperatures and low SOC scenarios. Overall, the proposed SOC estimation method based on MM-UKF holds significant practical value for underwater vehicles operating under complex conditions.

Against the backdrop of the partially idle resources of conventional railways and the continuous growth of suburban leisure tourism demand, it is required to develop railway tourism by utilizing the disused resources of conventional railways in China. Therefore, restoring, analyzing, and evaluating the visual landscapes seen when traveling on conventional trains are extremely important. The innovation of this study is proposing a dynamic visual landscape interface for disused railways, which establishes a connection between the “seeing” and “perceiving” for passengers and describes the visual landscape they observe. By analyzing the dynamic visual factors of passengers and combining DEM data, the temporal-spatial relationship of the visual landscapes along the railway is described as a dynamic visual landscape interface model. Based on this, the study proposes criteria and indicators for analyzing and evaluating the dynamic visual landscapes of railways, using typical landscapes along the Jingmen railway as an example for evaluation and validation. The research results provide analytical and decision-making tools for repurposing disused railways as tourist railways.

This work proposes a sub-optimal method based on a two-layer structured meta-deep reinforcement learning (MDRL) approach to address the hardware impairment (HWI) optimization issue in large intelligent surface (LIS) systems. This method, designed for distributed LIS systems with reflection matrices, effectively enhances the system capacity and performance despite HWIs. Building upon existing techniques of dividing large-area LIS systems into multiple small-area subsystems, the simulated results demonstrate that sub-optimal LIS performance can be achieved with fewer samples in diverse dynamic wireless environments. This innovative approach enhances the adaptability of distributed LIS systems and offers an effective HWI management strategy, paving the way for future LIS system optimization.

We propose and demonstrate a high-quality ($Q$) bound state in the continuum (BIC) cavity based on a hexagonal lattice photonic crystal (PhC) slab with $C_{6}$ symmetry, which carries high-order topological charges of $q=-\text{2}$. By combining the lateral photonic bandgap and high-order topological charge, we realize a $Q$ of $\text{7}\times \text{10}^{4}$ in the experiment which is several times higher than traditional BIC cavities with topological charge of $q=\pm\text{1}$. Specifically, we found that the boundary orientation would significantly influence the performance of lateral confinement. In contrast to an X direction boundary, a gradient J direction boundary can best alleviate the shortcoming of lateral leakage due to incomplete bandgap and the scattering because of momentum mismatching, and relax the strict constraints on wafer thickness. Our findings could broaden the application of BIC cavities from bio-sensing to on-chip telecommunication.

In the global precipitation measurement (GPM) era, the integrated multisatellite retrievals for GPM (IMERG) stands as a pivotal precipitation algorithm. This study aims to evaluate the accuracy of IMERG in capturing precipitation frequency and volume in the Yellow River Basin, China. Four satellite-based precipitation estimates (SPEs) from three algorithmic versions of IMERG were analyzed on an hourly scale using gauge observations. An improved error component procedure was employed to identify error sources. Results showed that all four IMERG products effectively captured the spatial distribution patterns and seasonal changes of precipitation, IMERG_F demonstrated the best overall performance, followed by IMERG_L, while IMERG_E performed the worst. However, they tended to overestimate precipitation. IMERG_F_Cal performed the best for precipitation frequency detection, with the highest probability of detection (POD = 52.7%) and the lowest missed events (MIS = 47.3%). Error component analysis highlights false bias as the main source of error, followed by missed bias. In winter, missed bias was the primary error source. Notably, a significant overestimation of precipitation was observed along the Yellow River. In detail, false bias dominated IMERG_E, IMERG_L, and IMERG_F_UnCal below 800 m in spring, summer, and autumn. However, in winter, missed bias became the primary error source for these three products at elevations above 200m and for IMERG_F_Cal above 500 m. IMERG_F_Cal exhibited false bias as the primary error source in all seasons. Suggested algorithm developers focus on improving IMERG SPEs' identification capabilities for light precipitation events and rainstorms. Findings can provide a reference for improving the IMERG product algorithms and enhancing users’ understanding of the error characteristics and sources of IMERG products.

The gas pipeline network constitutes a critical component of urban infrastructure. Emergencies, such as leaks or explosions, can lead to severe casualties, extensive property damage, and long-lasting effects on societal and environmental safety. The genesis of such emergencies typically involves a complex interplay among various risk factors. Therefore, the meticulous identification of these risk factors and the effective modeling of their interrelations are crucial for assessing the risks associated with such emergencies and enhancing urban safety. In this research, we initially developed a specialized dataset tailored specifically to analyze emergencies within gas pipeline networks. Recognizing the significance of unstructured domain-specific knowledge, we introduced an innovative approach Fusing expert knowledge and pre-trained model. This approach facilitates the fusion of domain experts’ knowledge into the process of risk factor identification and extraction in different steps, ensuring a thorough and precise determination and filtration of risk factors. Subsequently, a network depicting the associations between risk factors was established, drawing on their co-occurrence relationships, followed by a quantitative analysis of this network. Supported by real-world gas network management and emergency cause investigations, the methods proposed in this study provide strong support for gas network safety management and emergency decision-making.

Terahertz (THz) radar is well-suited for dynamic mobile target surveillance due to its high frame rates, minimal delay, and superior resolution. However, the defocusing problem caused by the motion of targets in the scene and the nonideal motion of the airborne platform severely affect synthetic aperture radar (SAR) imaging, particularly in the THz wave band. To address this issue, this article proposes a high-resolution imaging algorithm for THz-SAR ground-moving targets using the equalized alternating direction method of multipliers (ADMM). First, the range-Doppler algorithm is employed to generate a coarse image and extract the moving target region as the region of interest. Then, the range profile can be obtained by applying the inverse, which serves as input for the ADMM solving algorithm. Finally, the 2-D image of the target is reconstructed iteratively. The algorithm fully exploits the low-rank and sparse characteristics of moving targets in the SAR image to separate them from the background. It updates the azimuthal matched filter iteratively by minimizing image entropy and designs the equalization factor to retain target details while maintaining image sparsity. This approach significantly improves focusing accuracy compared with conventional algorithms while maintaining high imaging speed without estimating Doppler parameters.

Name Entity Recognition (NER) aims to recognize entities in the engine room domain from unstructured engine room domain text. But in the engine room domain, the entities are diverse and complex, and there is a nesting phenomenon, resulting in a low entity recognition rate. In this paper, a deep learning method incorporating language models is proposed to enhance the entity recognition performance within the engine room. domain. Firstly, the Bidirectional Encoder Representation from Transformers (BERT) language model is employed to train text feature extraction, acquiring a matrix of vector representations at the word level. Secondly, the trained word vectors are fed into the Bidirectional Gated Recurrent Unit (BiGRU) for contextual semantic entity feature extraction. Finally, the global optimal sequence is extracted by combining with the Conditional Random Field (CRF) model to obtain the named entities in the ship cabin semantics. The experimental results show that the proposed algorithm can obtain better F1 values for all three types of entity recognition. Compared with BERT-BiGRU, the overall accuracy of entity identification, recall rate and F1 value are improved by 1.35%, 1.45% and 1.40%, respectively.

Pedestrian re-identification (Re-ID) leverages cross-camera data acquired by the Internet of Things (IoT) devices and sensors to identify, monitor, and analyze pedestrians, allowing IoT applications to provide more intelligent, secure, and tailored services. Current pedestrian Re-ID research faces many challenges, such as low image resolution, perspective changes, posture changes, light changes, and occlusions, resulting in models trained on other datasets being unable to be directly applied and showing poor generalization capabilities. In addition, IoT edge devices are often limited by processing power and memory capacity and cannot withstand complex and large deep learning models. Therefore, designing a lightweight and generalizable pedestrian Re-ID model is better suited for implementation on edge devices. Considering these issues, this study presents MetaGON, a lightweight model with cross-domain generalization capabilities, which combines the lightweight omni-scale network (OSNet) with the meta-learning method and Cycle Generative Adversarial Network (CycleGAN) to perform domain generalization. The model’s generalization is enhanced through the simulation of the two stages of domain generalization in the meta-learning pipeline, where the obtained losses from meta-training and meta-testing are utilized for model optimization. Moreover, CycleGAN is employed to enhance and introduce style variations to the source data. The proposed MetaGON model is tested on a railway station re-identification dataset, and the model is deployed to edge devices for evaluation, which verifies the effectiveness of the algorithm.

Accurate prognostic prediction in patients with disorders of consciousness (DOC) is a core clinical concern and a formidable challenge in neuroscience. Resting-state EEG has shown promise in identifying electrophysiological prognostic markers and may be easily deployed at the bedside. However, the lack of brain dynamic modeling and the spatial mixture of signals in scalp EEG have constrained our exploration of biomarkers and comprehension of the mechanisms underlying consciousness recovery. Here, we introduce EEG source space analysis and brain dynamics to investigate the brain networks of patients with DOC (n = 178) with different outcomes (six-month follow-up), followed by graph theory and high-order topological analysis to explore the relationship between network structure and prognosis, and finally assess the importance of features. We show that a positive prognosis is associated with large-scale lower levels of low-frequency hypersynchrony. Moreover, we provide evidence that this pattern is driven not by all brain states but only by specific states. Analyses reveal that the positive prognosis is attributed to the network retaining lower segregation, higher integration, and stronger stability compared to the negative prognosis. Furthermore, our results highlight the importance of brain networks derived from brain dynamics in prognosis. The prognosis models based on clinical and neural features can achieve acceptable and even excellent performance under different outcome definitions (AUC = 0.714–0.893). Overall, our study offers new perspectives for the identification of prognostic biomarkers and provides avenues for profound insights into the mechanisms underlying consciousness improvement or recovery.

An electromagnetic railgun is a new type of weapon with ultrahigh speed based on electromagnetic thrust. It is used in important military domains, such as long-range strikes, strategic air defense, and ballistic interception. It is also used in aerospace fields, such as space debris cleanup, microsatellite launch, and space station relay launch, so its future prospects are excellent. We analyzed 1324 publications on electromagnetic railgun using CiteSpace and identified information including authors, institutions, journals, and popular research topics and trends. Our analysis found that the University of Texas is the research institution that has published the most documents on electromagnetic railgun, and French researcher Markus Schneider has published the most documents. The journals of IEEE Transactions on Magnetics and IEEE Transactions on Plasma Science have played an important role in promoting the development of electromagnetic railgun. The research direction of electromagnetic railgun has a distinct multidisciplinary characteristic because it touches on physics, engineering science, mathematics, materials science, energy and fuel, communication technology, instrumentation, and computer science. The development process of electromagnetic railgun can be grouped into three stages: basic theoretical research, engineering research, and system optimization research. The major research topics include electromagnetic force, armature design, rail materials, pulsed power supply, armature–rail contact surface characteristics, and coupling analysis of multiple physical fields. Future electromagnetic railgun will be scalable in terms of armature–rail contact characteristics and life span, energy storage density of power supply, current-carrying capability and thermal management of launcher materials, optimization design of armature structure, and complex data processing of control systems.

The radiometric calibration technique of spaceborne synthetic aperture radar (SAR) aims to establish a constant relationship between the SAR target pulse response energy and the actual radar cross-section (RCS) of the ground. It is an essential technology for achieving quantitative observations of the Earth using radar. In this article, the process of in-orbit radiometric calibration campaign for the SuperView Neo-2 SAR satellite is introduced, which is the distributed X-band SAR satellite in the new generation of Chinese SiWei commercial remote sensing satellite system. More specifically, the initial step involves deriving the total transfer function of the SAR system using the radar equation and establishing a stable calibration constant model for the SAR system. Further, by using the obtained calibration constant to invert the RCS value of the corner reflectors in the Otog Banner calibration field, while validating and analyzing the absolute radiometric accuracy. Considering that the elevation antenna pattern (EAP) predominantly affects the consistency of image radiometric correction, the method of extracting EAP and beam center bias angle from the rainforest is presented. Finally, the quality validation of impulse response function is performed over the California calibration field, which can effectively guarantee the quality of SuperView Neo-2 SAR satellite products.

We propose a cost-effective and memory-aware end-to-end learning scheme utilizing bi-directional gated recurrent unit (bi-GRU) for geometric constellation shaping (GCS) under the first-order regular perturbation (FRP) auxiliary channel. The performance of the proposed system has been numerically verified at a 32 GBd 5-channel wavelength division multiplexing (WDM) 64 quadrature amplitude modulation (QAM) 800 km optical coherent communication system. Results show that the proposed bi-GRU based GCS scheme can achieve a performance gain over square 64QAM in mutual information (MI) with 0.12 bits/symbol and a Q-factor gain of 0.4 dB at optimal launched optical power. When transmission distance is extended to 1280 km, a generalized mutual information (GMI) gain of 0.136 bits/symbol is observed. Additionally, compared with the bi-directional long short-term memory (bi-LSTM) based GCS, the proposed bi-GRU scheme has lower computation complexity with similar system performance.

Segmentation of island waterline contributes to shoreline movement analyzing, environmental monitoring, and integrated coastal zone management. To achieve high efficiency and high accuracy of island waterline segmentation in remote sensing images, we proposed a model for island waterline segmentation based on DANet (DANet-SMIW). In DANet-SMIW model, different indexes (Normalized Difference Water Index and OTSU) were taken as new channels added to input dataset, which enhanced waterline's spectral information. DANet backbone network was improved by dense connection of DenseNet. Loss function, consisting of binary cross entropy loss and Dice loss, was used to resolve the sample imbalance problem, and then, rough results of island waterline segmentation were refined by boundary refinement module (BRM). In total, 2042 island images were taken as experiment dataset, which were cropped from Landsat-8 images and divided into 1634 images for training, 100 images for testing, and 308 images for validation, and DANet-SMIW model was compared against other models, including FCN-32s, DeepLabv3+, PSPNet, Dense-ASPP, PSANet, ICNet, DuNet, and PIDNet. Results demonstrated that DANet-SMIW model achieved the highest values with pixel accuracy and Mean Intersection over Union and possessed higher segmentation efficiency than most other models. Collectively, DANet-SMIW model was an integrated accurate and efficient model for island waterline segmentation in remote sensing images.

As a typical application of intelligent transportation systems (ITS), the electronic toll collection (ETC) system has been widely used on highways due to its excellent toll efficiency. In the actual traffic environment, there usually exists more than one vehicle in the radiation area of the road-side units (RSU) at the same time, and the front vehicle will affect the ray paths between the RSU and the on-board units (OBU) of the back vehicle. Here we studied the front vehicle’s impact on the ETC system’s path propagation loss by ray tracing technology and Uniform Theory of Diffraction (UTD). Firstly, we simplified the vehicle body structure into two equivalent geometric models. Four situations were considered according to the vehicles’ front and rear positions. We analyzed each scenario’s propagation loss models under line-of-sight (LOS) and non-line-of-sight (NLOS) conditions based on the distance change among RSU, OBU, and the front vehicle. Finally, we developed an ETC comprehensive test equipment to measure the propagation loss of the four scenarios at a toll station. Both simulation and experiment results indicate that the propagation loss models proposed in this paper are valid.

The 4-wheel independent drive (4WID) electric vehicle has better safety and stability. However, the multi-motor drive system would lead to an increased fault probability in the vehicle. In particular, the uncertainty of fault motors and parameter disturbances could lead to uncertainties in control. This study presents a fault-tolerant hierarchical control approach that utilizes an improved Model Predictive Control (MPC) to address the problem. The proposed method employs Sliding Mode Control (SMC) in the upper layer to generate the vehicle’s yaw moment, which could inhibit system parameter disturbance and improve the system’s robustness. Incorporating the fault matrix induced by motor malfunctions, the state-space equation of the vehicle is modified to establish a vehicle dynamics model under motor faults. In the lower layer, the torque reconstruction allocation strategy is designed to coordinate the four motors under the motor fault condition by the MPC rolling optimization online, which could reduce the impact of the motor fault uncertainty for fault-tolerant control. The multi-constraint conditions of MPC are set up according to the vehicle state parameters. To address the additional yaw moment caused by the MPC torque reconfiguration allocation control strategy, the torque transfer method is used as the input allocation for MPC. Finally, the proposed control strategy is verified by online simulation. The simulation outcomes demonstrate the effectiveness of the proposed hierarchical fault-tolerant control approach in achieving fault-tolerant control of the 4WID electric vehicle, with the improved MPC outperforming the conventional MPC in terms of performance.

Recently, unmanned aerial vehicles (UAVs) have shown immense potential in military and civilian applications. However, the limited endurance of UAVs poses a crucial challenge to their widespread adoption and development. Fortunately, wireless power transfer (WPT) technology has emerged as a promising solution to address this issue. In contrast to the previous papers on the topic, this review aims to provide a holistic top-down view of WPT technology for UAVs. More specifically, it introduces the fundamental principles of WPT technology firstly. Then, the design and optimization methods of coupling mechanisms, especially for UAVs, are introduced and compared. Numerous different and significant designs are reviewed and discussed. Furthermore, the compensation topology and control strategies are elaborated in detail subsequently. Finally, the key technical challenges faced by WPT for UAVs are put forward and future directions for development is proposed. Therefore, this review can be considered as an expedient reference for researchers conducting studies in the field of WPT technology for UAVs. Keeping this in mind, we present meaningful perspective for each of the analyzed key aspects, providing inspiration for what follows.

Accurate detection of target location and type is crucial for treating ocular trauma caused by foreign bodies intrusion. However, the traditional method of manually marking CT image targets has slow recognition speed and poor detection accuracy, which cannot meet the real-time and accuracy requirements for detecting foreign bodies in clinical diagnosis. To address this issue, we propose a lightweight detection and recognition model based on feature extraction and fusion. Firstly, the normalization-based attention module and the sigmoid linear unit activation function are introduced into the inverted residual block of the backbone network to enhance the model’s attention to salient features and improve the detection accuracy. Then, the path aggregation feature pyramid network is utilized to fuse multiscale features, enabling the information interaction between different levels of the network and enhancing the accuracy foreign bodies classification. In particular, the incorporation of the space-to-depth convolution and convolutional mixing modules into the feature pyramid network significantly reduce the computational overhead while effectively capturing the key semantic features in both space and channel directions, thereby improving the lightweight level of the model. Finally, the location and type information of the foreign intraocular bodies are obtained by this model. The experimental results demonstrate the superior performance of the proposed model in terms of mAP@0.5, accuracy, sensitivity and specificity, achieving 97.2, 93.5, 98.0 and 88.0, respectively. Furthermore, the smaller number of parameters and faster detection time allow the proposed model run in real-time on poorly configured hardware, making it more suitable for clinical applications.

Compared with English Named Entity Recognition (NER), Chinese Named Entity Recognition (CNER) has a high difficulty in word segmentation, and accurate extraction of contextual semantic feature information is a key work of CNER. For that, we propose a CNER model to extract both local and global contextual semantic feature information. First, we propose to apply the dynamic convolutional kernel to the convolutional layer of TCN to enhance the local features of contextual semantic feature information. Second, we define a dynamic scaling factor computation method to compute the correlation between named entity characters in the multihead attention, which to process the problem of sparse distribution of named entities, and can efficiently extract the global features of contextual semantics. We validated the effectiveness of the proposed model on the Weibo dataset with an F1 value of 89.24%, which is better than commonly used models.

Robots capable of robust, real-time recognition of human intent during manipulation tasks could be used to enhance human-robot collaboration for activities of daily living. Eye gaze-based control interfaces offer a non-invasive way to infer intent and reduce the cognitive burden on operators of complex robots. Eye gaze is traditionally used for “gaze triggering” (GT) in which staring at an object, or sequence of objects, triggers pre-programmed robotic movements. We propose an alternative approach: a neural network-based “action prediction” (AP) mode that extracts gaze-related features to recognize, and often predict, an operator’s intended action primitives. We integrated the AP mode into a shared autonomy framework capable of 3D gaze reconstruction, real-time intent inference, object localization, obstacle avoidance, and dynamic trajectory planning. Using this framework, we conducted a user study to directly compare the performance of the GT and AP modes using traditional subjective performance metrics, such as Likert scales, as well as novel objective performance metrics, such as the delay of recognition. Statistical analyses suggested that the AP mode resulted in more seamless robotic movement than the state-of-the-art GT mode, and that participants generally preferred the AP mode.

A transformer can capture long-term dependencies through an attention mechanism, and hence, can be applied to various vision tasks. However, its secondary computational complexity is a major obstacle in vision tasks that require accurate predictions. To address this limitation, this study introduces linear-window attention (LWA), a new attention model for a vision transformer. The transformer computes self-attention that is restricted to nonoverlapping local windows and represented as a linear dot product of kernel feature mappings. Furthermore, the computational complexity of each window is reduced to linear from quadratic using the constraint property of matrix products. In addition, we applied the LWA to feature matching to construct a coarse-to-fine-level detector-free feature matching method, called transformer with linear-window attention for feature matching TRLWAM. At the coarse level, we extracted the dense pixel-level matches, and at the fine level, we obtained the final matching results via multi-head multilayer perceptron refinement. We demonstrated the effectiveness of LWA through Replace experiments. The results showed that the TRLWAM could extract dense matches from low-texture or repetitive pattern regions in indoor environments, and exhibited excellent results with a low computational cost for MegaDepth and HPatches datasets. We believe the proposed LWA can provide new conceptions for transformer applications in visual tasks.

In this paper, we proposed a cavity containing an intracavity medium of five-level double-ladder-type atoms with electromagnetically induced transparency to enhance Goos-Hänchen shifts of reflected and transmitted light beams. The dependence of the Goos-Hänchen shifts has been analyzed. It is shown that Goos-Hänchen shifts can be controlled by modifying the intensity and detuning of the coherent control field without changing material and the structure of the dielectric interface. This work has considerable potential for applications such as optical devices in information processing, flexible optical-beam steering and alignment, optical sensors and optical switches.

Accurate and efficient forecast of PM2.5 concentration is the primary prerequisite for promoting urban green development and improving residents’ well-being. In this study, a hybrid model based on secondary decomposition ensemble and weight combination optimization is presented to materialize exact PM2.5 concentration prediction. First, the empirical wavelet transform (EWT) is adopted to disassemble the primeval PM2.5 concentration sequence to get high and low-frequency components. Considering the intricacy of high-frequency components and the difficulty of direct prediction. Therefore, it is further decomposed into a collection of modes with significant discrepancies by adaptive variational mode decomposition (AVMD). Second, the prediction network and meteorological data are determined respectively according to Hurst exponent. Then support vector regression (SVR) model and bidirectional long short-term memory (BILSTM) network are used to model each sequence separately. In addition, the weights of each forecast network were optimized by improved sparrow search algorithm (ISSA) to correct decomposition errors. Finally, all prediction results were weighted and integrated to receive the ultimate prediction values. The test results show that whether it is 1-step prediction, 3-step prediction or 5-step prediction, the proposed model has the best prediction effect in Beijing, Handan and additional Shanghai cases.

The output feedback robust control design is required for switched systems (SSs) without the state feedback information. For the sliding mode control (SMC) based on state observers, the unmatched uncertainty and disturbance cannot be rejected completely. In this paper, an output feedback sliding mode control (SMC) based on a robust integral sliding mode (RISM) is presented for switched systems with unmatched uncertainties. First, the control task of the output feedback stabilization based on the state observer is given. Then, in order to reject the unmatched uncertainty, the RISM is constructed on the space of the estimated state, whose parameters are selected ensuring that the system in the sliding mode is robust exponentially stable. Linear matrix inequality conditions for the parameter design and the stabilization switching rule are achieved, which make the sliding mode can reject the unmatched uncertainty and suppress the unmatched disturbance both at the designed control switching rule and the arbitrary switching rule. Consequently, the corresponding SMC controller is designed based on the state estimation and the propsoed RISM. Finally, the application simulation results to a one-link manipulator with the load change validated the effectiveness and feasibility.

The traditional approach to ship pipeline programming often involves describing and calculating based on text files. This method tends to be error-prone and time-consuming, especially for complex systems with a large amount of data related to pipelines. In case of system failures, troubleshooting becomes inconvenient. Additionally, matrix calculations can lead to issues such as extreme values, and visualizing the results is challenging. To address these challenges, this paper proposes an intuitive, user-friendly, and efficient ship pipeline programming tool. The tool utilizes Microsoft’s WPF (Windows Presentation Foundation) graphical interface technology and the C# programming language. A graphical interface for ship pipeline networks is constructed, allowing users to visually build the physical model of pipelines through intuitive drag-and-drop and connection operations. Furthermore, the paper adopts Excel spreadsheets as the input method for data and combines it with domain knowledge of marine engineering pipelines. This approach establishes logical models for various pipelines, valves, and other equipment on ships. The models incorporate attributes from the spreadsheet to control characteristics like maximum flow rates and flow directions. By combining graphical modules, pipeline connections, and Excel spreadsheets, the method proposed in this paper offers a more convenient way to construct pipeline network models. The resulting models closely resemble actual ship pipelines, leading to improved accuracy and efficiency in data processing and calculations.

For high-resolution spotlight synthetic aperture radar, Doppler aliasing is always introduced in the long synthetic aperture observation time. The existence of signal Doppler aliasing will greatly affect the accuracy of motion compensation. To avoid Doppler aliasing, it is an effective strategy to use the subaperture (SA) approach. However, it will introduce the long synthetic aperture time stitching problem of signal and phase error. Compared with the SA approach, the two-step approach (TSA) based on the full aperture not only eliminates the Doppler aliasing but also avoids the signal splicing problem. Therefore, a two-dimensional (2-D) frequency-domain autofocus combined with the TSA is adopted in this article. First, TSA is used to perform dealiasing and coarse focusing imaging on full-aperture (FA) data. Second, the phase error history and the phase error form in the TSA process are analyzed and deduced, and an accurate prior structure of the phase errors is established. Then, the accurate inversion of the common 2-D phase errors is realized based on this prior structure. Finally, the autofocus process of the FA signal combined with the TSA is realized in the nonaliasing 2-D frequency domain. The validity of the recommended algorithm is verified by simulation and experimental results.

With continuous advancement of policies regarding to energy conservation, emission reduction and clean heating, the market for air source heat pumps is rapidly growing. In the current situation, the need to reduce operating costs of heat pump heating has become urgent. This study proposes a distributed air source heat pump heating system based on phase change heat storage. The system utilizes a phase change heat storage device to transfer heat load from high electricity price periods to low electricity price periods. In order to reduce operating costs, control schemes are formulated based on peak-valley electricity pricing. The system has been tested in Qinghai Province, a high-altitude area in China. Results demonstrate that the system helps to achieve the decoupling between heating and power. The proportion of low-valley electricity consumption in the system increased to an average of 57.45%, while the proportion of peak electricity consumption decreased to an average of 12.41%. The daily electricity costs were reduced by 5.28% compared to direct heat pump heating. With the same thermal storage capacity, application of suitable operation scheme increased COP by 21.76%, reduced electricity consumption and operating costs by 5.69% and 13.50% respectively. Increasing the capacity to 150 kW $\cdot \text{h}$ can further reduce operating costs by 10.04%.

Automatic Identification System (AIS) can obtain a large amount of data on ship trajectories and movement characteristics, which provides data support for ship route planning, navigation safety, and maritime traffic control. Therefore, it is of great importance to analyze and cluster the AIS data to obtain an understanding of the movement behavior of ships. The Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is capable of discovering arbitrarily shaped clusters and is suitable for clustering ship trajectories. However, the traditional DBSCAN algorithm suffers from the following shortcomings. First, the DBSCAN algorithm requires the input of a similarity matrix between the trajectories, and the similarity matrix obtained from the traditional Hausdorff distances does not identify the trajectory directions. Second, the DBSCAN algorithm is sensitive to the input parameters, and the clustering results obtained from different datasets with the same parameters vary widely. Finally, the DBSCAN algorithm is poor at clustering multi-density distribution datasets. To address these shortcomings, this study proposes a multi-density adaptive trajectory clustering (MDA-Traclus) algorithm, which considers the multi-density distribution trajectory dataset, adaptively determines the input parameters, and adds a trajectory direction identification mechanism to realize the clustering of trajectory clusters with different directions and different densities. Here, the classical DBSCAN algorithm, KANN-DBSCAN algorithm, and MDA-Traclus algorithm are compared and analyzed using four manually labeled trajectory datasets and a real trajectory dataset. The results show that the MDA-Traclus algorithm and KANN-DBSCAN algorithm are able to automatically determine the input parameters when clustering trajectory datasets with uniformly distributed densities and achieve the same clustering effect as the DBSCAN algorithm. When the trajectory density distribution is not uniform and the directions of the trajectory clusters are not consistent, the MDA-Traclus algorithm can effectively identify the low-density regions, differentiate the different directions of trajectories, and improve the clustering of the trajectories.

Vehicle localization plays a crucial role in ensuring the safe operation of autonomous vehicles and the development of intelligent transportation systems (ITS). However, there is insufficient effort to compare the performance and challenges of different vehicle localization algorithms. This paper aims to address this gap by analyzing the comprehensive performance of existing advanced vehicle localization techniques and discussing their challenges. Firstly, we analyze the self-localization methods based on active and passive sensors. The results show that, the light detection and ranging (LiDAR) and vision-based localization techniques can reach high accuracy. However, they have high computational complexity. Only using the inertial measurement unit (IMU), global positioning system (GPS), radar, and ultrasonic sensors may not realize localization result with high accuracy. Then, we discuss V2X-based cooperative localization methods, analyze the multi-sensor based localization techniques and compare the comprehensive performance among all methods. Although the artificial intelligence (AI) techniques can effectively enhance the efficiency of vision-based localization algorithms, the high computational complexity still should be considered. In addition, since the IMU, GPS, radar, and ultrasonic sensors have good performance in terms of the availability, reliability, scalability, and cost-effectiveness, they can be used as auxiliary sensors to achieve good comprehensive performance through data fusion techniques. Finally, we propose the challenges of different techniques and look forward to future work.

Accurate grasp of chicken body temperature can effectively improve the success rate of caged chicken breeding, by monitoring the number of open-mouthed chickens as a percentage of the total number of chickens and can directly determine whether the chicken body temperature is appropriate. There is no relevant solution to this requirement at present, so this paper proposes a joint feature target detection algorithm based on YOLOv5 to detect the opening and closing state of the chicken mouth. The algorithm improves the YOLOv5 network in the following ways: 1) The improved ResC module is used to reconstruct the backbone network of YOLOv5, which diversifies feature scales and enhances the ability of target feature extraction; 2) Integrate the Transformer module with the four-layer feature pyramid to expand the range of feature fusion and improve the accuracy of feature extraction; 3) The joint feature verification(JFV) module is designed to improve the detection accuracy of small targets by adopting the idea of joint verification of small targets and large targets. Finally, the improved network is used to detect the opening and closing state of the chicken beak on the test set, which is derived from the actual cage chicken breeding environment. The results show that the average accuracy (mAP) of the improved RJ-YOLOv5 algorithm is 85.6%, and the detection accuracy is 7.1% higher than the YOLOv5 algorithm; The video detection frame rate reaches 69 FPS, which can meet the requirements of real-time monitoring of chicken farms.

In modern smart cities, the scale of urban backbone networks used to provide Internet communication environment are constantly increasing. When faults occur, it usually takes lots of efforts to detect and locate the faults. As a result, automatic detection and positioning of faults with use of intelligent algorithms have been a practical demand in this area. In this paper, the complicated whole urban backbone network is viewed as a graph-level object, in which massive nodes and edges are involved. On this basis, a two-stage graph deep learning-based fault detection and positioning method for Internet communication networks. For the first stage, the graph neural network is employed to extract graph-level features from Internet communication networks. This is expected to obtain proper feature representation for core characteristics of backbone networks. For the second stage, the fault detection and positioning algorithm is formulated to output final results. At last, experiments are conducted to assess performance of the proposal. The results show that the proposed method has good performance in abnormal node detection as well as high accuracy in fault positioning. The accuracy of the two-stage graph deep learning algorithm proposed in this chapter is much higher than that of KNN algorithm, reaching 96.5% in the end, slightly lower than that of pure graph deep learning algorithm, while the accuracy of IRBFG algorithm can only reach 92%.

Legal judgment prediction (LJP) applies Natural Language Processing (NLP) techniques to predict judgment results based on fact descriptions automatically. The present work addresses the growing interest in the application of NLP techniques to the task of LJP. Despite the current performance gap between machines and humans, promising results have been achieved in a variety of benchmark datasets, owing to recent advances in NLP research and the availability of large-scale public datasets. To provide a comprehensive survey of existing LJP tasks, datasets, models, and evaluations, this study presents the following contributions: 1) an analysis of 43 LJP datasets constructed in 9 different languages, together with a classification method of LJP based on three different attributes; 2) a summary of 16 evaluation metrics categorized into 4 different types to evaluate the performance of LJP models for different outputs; 3) a review of 8 legal-domain pretrained models in 4 languages, highlighting four major research directions for LJP; 4) state-of-the-art results for 11 representative datasets from different court cases and an in-depth discussion of the open challenges in this area. This study aims to provide a comprehensive review for NLP researchers and legal professionals to understand the advances in LJP over the past years, and to facilitate further joint efforts towards improving the performance of LJP models.

This paper analyzes the OBL strategy’s impact on optimizing the GWO algorithm and identifies three shortcomings. the specific limitations of the OBL optimization approach. To address these three shortcomings and enhance both global optimization and local exploration capabilities of GWO, this paper introduces a follow-controlled opposition learning strategy. then, the paper analyzes the control parameter C of the grey wolf algorithm to investigate its impact on global optimization and local exploration. Based on these properties, a new control parameter C is proposed. The proposed learning strategy and control parameter C are introduced into the traditional grey wolf algorithm to obtain the FCGWO algorithm. Finally, this paper conducts a comparative analysis of the FCGWO algorithm in comparison to other meta-heuristic algorithms, as well as the enhanced grey wolf algorithm, utilizing 23 benchmark test functions and 2 engineering problems. The results indicate that FCGWO effectively avoids the shortcomings of the traditional OBL, while also outperforming other algorithms significantly in terms of solution quality.

In the secondary utilization of retired batteries in energy storage systems, there exists a significant disparity between batteries, rendering the storage battery pack susceptible to issues such as overcharging and overdischarging during the charging and discharging process. These challenges impose limitations on the available capacity of the storage battery system and may even give rise to safety hazards. To tackle these issues, this study proposes the design of a battery equalization system specifically tailored for retired batteries employed in energy storage systems. A multilayer equalization topology based on a buck-boost chopper circuit is introduced, along with a multi-objective adaptive fuzzy control equalization strategy. This strategy integrates the open-circuit voltage-state of charge (OCV-SOC) characteristic curve of the battery and utilizes voltage and SOC as equalization variables. It governs the magnitude of equalization current based on the battery pack’s SOC state, enabling effective equalization control of the retired battery pack. Simulation results demonstrate that the proposed equalization strategy does not rely on the precise mathematical model of the battery system. When compared with the mean-difference equalization strategy or the single-variable equalization strategy, it exhibits an enhanced equalization rate. Consequently, it fulfills the practical application requirements for equalizing retired battery packs.

The stability of a grid-tied inverter system will be compromised in weak grid conditions. In order to address this problem, researchers have proposed using a power electronics transformer (PET) to connect inverters and the grid. By adjusting the controller parameters or control strategy, the output impedance of the PET low voltage (LV) converter can be varied to improve the system stability. This paper establishes impedance models of the PET-based grid-tied inverter system and evaluates the stability of the system using the generalized Nyquist criterion. The impact of the variety of control parameters on the system stability is then analyzed. To improve the stability of the system, two stabilized control strategies, namely virtual impedance and digital filter-based methods are proposed. Finally, the effectiveness of these strategies is evaluated through simulations and experiments.

The predictive business process monitoring mainly focuses on the performance prediction of business process execution, i.e., predicting the next activity, the execution time of the next activity, and the remaining time, respectively, for an ongoing process instance based on the knowledge gained from historical event logs. Although there is a specific relationship between these three tasks, recent research has focused on training separate prediction models for each task, resulting in high costs and time complexity. Additionally, existing technologies are limited in their ability to capture long-distance dependent features in process instances, further impeding prediction performance. To address these issues, this paper proposes the MTLFormer approach, which leverages the self-attention mechanism of the Transformer network and conducts multi-task parallel training through shared feature representation obtained from different tasks. Our approach reduces the time complexity of model training while simultaneously improving prediction performance. We extensively evaluate our approach on four real-life event logs, demonstrating its capability to achieve multi-task online real-time prediction and effectively improve prediction performance.

Specific emitter identification (SEI) is a powerful technique that identifies different emitters utilizing the features generated from hardware imperfections. The features, also known as radio frequency fingerprints (RFFs), can be extracted via model-based methods or data-based methods. In this paper, we analyze the identification accuracy of SEI system and provide criteria for evaluating different SEI schemes. We use in-phase/quadrature (I/Q) imbalance as an example to examine the identification performance by analyzing the Euclidean distance between radio frequency (RF) signals of different emitters. This derivation can also be generalized to different modulation schemes and distortion models.

Objective: Multivariate signal (MS) analysis, especially the assessment of its information transmission (for example, from the perspective of network science), is the key to our understanding of various phenomena in biology, physics and economics. Although there is a large amount of literature demonstrating that MS can be decomposed into space-time-frequency domain information, there seems to be no research confirming that multivariate information transmission (MIT) in these three domains can be quantified. Therefore, in this study, we attempted to combine dynamic mode decomposition (DMD) and parallel communication model (PCM) together to realize it. Methods: We first regarded MS as a large-scale system and then used DMD to decompose it into specific subsystems with their own intrinsic oscillatory frequencies. At the same time, the transition probability matrix (TPM) of information transmission within and between MS at two consecutive moments in each subsystem can also be calculated. Then, communication parameters (CPs) derived from each TPM were calculated in order to quantify the MIT in the space-time-frequency domain. In this study, multidimensional electroencephalogram (EEG) signals were used to illustrate our method. Results: Compared with traditional EEG brain networks, this method shows greater potential in EEG analysis to distinguish between patients and healthy controls. Conclusion: This study demonstrates the feasibility of measuring MIT in the space-time-frequency domain simultaneously. Significance: This study shows that MIT analysis in the space-time-frequency domain is not only completely different from the MS decomposition in these three domains, but also can reveal many new phenomena behind MS that have not yet been discovered.

Recent developments in electronic countermeasures have heightened the demand for the direction of arrival (DOA) estimation, and it has become more urgent to improve the resolution of closely spaced sources for conformal arrays. In previous studies, distributed sources incident on linear or planar arrays have been the point of focus, and their performance for closely spaced sources has deteriorated dramatically. To address this issue, we design a multi-head attention (MHA) mechanism-based cascaded estimator, which is composed of a signal-to-noise ratio (SNR) classification network and two DOA estimation subnetworks. The former network uses a fully connected neural network for SNR classification, whose output can activate the corresponding estimation subnetwork. In estimating subnetworks, we adopt multiple self-attention mechanisms for the first time to extract more implied and representative information from raw features for effective estimation. Numerical results verify that the proposed estimator not only exhibits better performance in terms of resolution and recovery accuracy even at lower SNR but also provides great robustness with respect to angle intervals and snapshots, which lays the groundwork for further DOA studies for closely spaced sources.

With the continuous maturity of remote sensing technology, the obtained remote sensing images' quality and quantity have surpassed any previous period. In this context, the content-based remote sensing image retrieval (CBRSIR) task attracts a lot of attention and research interest. Nowadays, the previous CBRSIR works mainly face the following problems. First of all, few works can realize one to many cross-modal image retrieval task (such as using optical image to retrieve SAR, optical images at the same time); second, research works mainly focus on small-area, target-level retrieval, and few on semantic-level retrieval of the whole image; last but not the least, most of the existing networks are characterized by massive parameters and huge computing need, which cannot be applied to resource-constrained edge devices with power and storage limit. For the sake of alleviating these bottlenecks, this article introduces a novel light-weighted nonlocal semantic fusion network based on hypergraph structure for CBRSIR (abbreviated as HGNLSF-Net). Specifically, in the framework, using the topological characteristics of hypergraph, the relationship among multiple nodes can be modeled, so as to understand the global features on remote sensing images better with fewer parameters and less computation. In addition, since the nonlocal semantics often involves a lot of noise, the hard-link module is constructed to filter noise. A series of experimental results on typical CBRSIR dataset, i.e., Multi-modal Multi-temporal Remote Sensing Image Retrieval Dataset (MMRSIRD), well show that with fewer parameters, the proposed HGNLSF-Net outperforms other methods and achieves optimal retrieval performance.

The receive antenna gain is needed to accurately calibrate the normalized bistatistic radar cross section measured by the BF-1 mission, which is a global navigation satellite system reflectometry (GNSS-R) constellation of two microsatellites and the first Chinese GNSS-R satellite mission. The instability of the satellite platform is the main cause of receive antenna gain errors. To obtain a high precision gain value, a calibration method that remaps the ocean surface detection location to the receive antenna pattern using satellite platform attitude measurements is proposed in this article. Thirty-two orbits of delay Doppler maps data, which were greatly disturbed by the attitude, are selected to test the effectiveness of the proposed algorithm. The accuracy of wind speed retrieval is analyzed, and results show that the data calibration algorithm is effective in reducing the wind speed retrieval error. Compared with the un-calibrated data, the data subjected to the calibration algorithm show a significant improvement of 19.33% in correlation coefficient and average decreases of 30.91% and 42.57% in root-mean-square error and mean bias error, respectively. Moreover, the comparison highlights that the influence of the satellite platform attitude disturbance on wind speed retrieval is abated significantly. The proposed approach can effectively improve the quality of GNSS-R measurements, allowing for a better understanding of global weather abnormalities and generally improving weather forecasting.

Achieving the coordinated optimization of tie-line reserve and energy storage on two timescales is a key issue in reserve scheduling for active distribution networks. To effectively manage the uncertainty related to renewable energy, we propose risk-based reserve scheduling for active distribution networks based on an improved proximal policy optimization algorithm. The reserve scheduling for active distribution networks is constructed as a two-scale multistage stochastic programming model, in which intraday real-time operation is constructed as a multi-stage stochastic programming model. With the help of more accurate and realistic intraday operation scenario simulation, the energy storage can be fully utilized to improve the system’s ability to climb and peak, reduce the reserve pressure of the main network, and achieve the purpose of improving the flexibility of system operation. An improved proximal policy optimization algorithm is proposed to solve optimization problems in a model-free manner while effectively handling conditional value-at-risk constraints. The effectiveness of the proposed risk-based reserve scheduling for active distribution networks based on an improved proximal policy optimization algorithm is verified by a modified IEEE33 case.

The automatic identification system (AIS) provides a massive database for ocean science. The original AIS data are redundant. Direct use will cause a waste of data storage space and computation costs; hence, data compression must be performed. The Douglas–Peucker algorithm (DP) is an effective trajectory compression algorithm that can well preserve the spatial characteristics of a trajectory but has the following shortcomings: first, it has poor track recovery when compressing multi-turn routes; second, it does not consider the ship speed and heading; and third, it may have the wrong result of the compressed trajectory crossing the obstacle. To address these situations, this study proposes a multi-objective peak DP algorithm (MPDP) that adopts a peak sampling strategy, considers three optimization objectives (spatial characteristics, heading and speed) of trajectory and adds an obstacle detection mechanism to realize a compression algorithm more suitable for curved trajectories. The classical DP algorithm is compared with the MPDP algorithm by simulating trajectory and real trajectory experiments. The results show that the MPDP algorithm optimizes the length loss rate, simultaneous Euclidean distance, and average deviations of the speed and the heading while maintaining a high compression rate similar to that of the DP algorithm. Moreover, it can also successfully avoid obstacles. The optimization effect is most obvious for the multi-turn or hovering trajectory. The optimization rate of length loss, synchronous Euclidean distance, and average deviation of the heading can reach 40%.