In complex maritime environments, formation intention recognition is fundamental to situation awareness and threat assessment tasks. Traditional methods rely on handcrafted rules and expert knowledge, making it difficult to jointly model inter-ship cooperative behaviors and dynamic sensing data. To address these challenges, this paper proposes a Static-Dynamic Collaborative Heterogeneous Graph Transformer network (SDC-HGT) for formation intention recognition. A static knowledge graph models ship attributes, onboard equipment, and inter-ship relationships, and a formation-level heterogeneous graph is constructed. A graph convolutional network aggregates static structural features, while a long short-term memory network extracts temporal features from radar and maneuvering behaviors. A gated fusion mechanism adaptively integrates static and dynamic features at the ship node level. Furthermore, a Heterogeneous Graph Transformer aggregates ship representations to produce a formation-level representation for intention classification. Experimental results demonstrate that SDC-HGT effectively fuses heterogeneous information, captures inter-ship cooperative behaviors, and achieves superior accuracy, robustness and anti-interference performance in extreme noise scenarios.

With the advancement of 3D printing technology, there is a growing trend toward employing intricate selective laser melted (SLM) lightweight lattice structures as hypervelocity impact-resistant devices, potentially replacing traditional Whipple shield configurations. However, systematic analysis of the hypervelocity mechanical performance of SLM-manufactured materials—particularly the widely used AlSi10Mg aluminum alloy—remains insufficient. To investigate the dynamic response mechanisms of SLM AlSi10Mg aluminum alloy under hypervelocity impact, this study systematically quantifies the material's mechanical behavior and pore defect effects through integrated porosity-incorporated numerical simulations and hypervelocity shock compression experiments. A quantitative predictive model correlating porosity with shock wave propagation was established through micro-CT-based pore reconstruction. The study identifies dual attenuation mechanisms mediated by pore networks, involving both energy dissipation through pore collapse and impedance mismatch effects at pore-matrix interfaces. These coupled mechanisms reduce shockwave velocity, attenuate pressure amplitude, and ultimately decrease the equation-of-state (EOS) parameters compared to those of defect-free theoretical values. Hypervelocity shock compression experiments were then conducted at pressures of 14.76 GPa-58.45 GPa, with maximum velocities exceeding 5 km/s, validating the reliability of numerical simulations and enabling the pioneering experimental determination of Hugoniot EOS parameters for SLM AlSi10Mg under hypervelocity conditions. The experimental results demonstrate that compared to conventional wrought aluminum alloys, the SLM material exhibits slight reductions in EOS parameters (1%-10%) alongside systematic degradation of compressive resistance. The scientific innovations of this work include quantitative elucidation of additive manufacturing (AM) defect-shockwave interactions through energy redistribution mechanisms; the pioneer experimental acquisition of Hugoniot EOS parameters for SLM aluminum alloys under extreme dynamic loading.

This paper addresses the problem of on-orbit inspection, where multiple small deputy spacecraft collaboratively maneuver around a chief spacecraft to acquire high-quality multi-view information under stringent safety and time constraints. The key challenges arise from inspection waypoint selection, transfer trajectory optimization, and collision avoidance in close-proximity operations. To enhance observation stability, a teardrop hovering configuration is adopted to constrain relative motion. An inverse-proportional approximation is further introduced to efficiently model the trade-off between velocity increment and transfer time, significantly simplifying trajectory planning. Building on these formulations, a multi-objective optimization model is established to jointly minimize informationin cost and mission duration. A two-stage solution framework is developed, combining a greedy-enhanced tabu search for waypoint selection with an improved NSGA-II for waypoint allocation and sequence. Simulation results demonstrate that the proposed method can effectively generate collision-free trajectories for multiple deputy spacecraft across diverse scenarios. The greedy strategy substantially improves initial solution quality and accelerates convergence, achieving a 1.59% performance improvement over standard tabu search. Moreover, the proposed multi-objective optimization algorithm consistently outperforms representative benchmark methods by 2%-6% in optimization effectiveness.

Existing technologies can achieve relative geometric correction and stabilization of geostationary satellite image sequences through fixed land scene matching or homonymous point adjustment. However, these methods heavily rely on fixed land areas, rendering them completely ineffective in vast ocean regions with only ship targets. Additionally, the trajectories of ship targets after processing still exhibit noticeable jitter, hindering motion information analysis. To address these issues, this paper proposes a joint image adjustment and stabilization method based on multi-target trajectories in marine environments: (1) An optimized target detection algorithm based on a multi-scale heterogeneous convolution module is introduced, which extracts background and target features through convolutions of different scales, enabling accurate detection and tracking of weak small targets in the image sequence frame by frame. (2) Curve fitting is performed on the detected positions of the same ship across multiple frames to simulate its motion trajectory under stabilized conditions. Combined with the prior assumption of uniform motion, an equal-division strategy is adopted to determine the corrected positions of the target in the image sequence. (3) The deviation correction values of multiple targets within the same frame are obtained, and based on the principle of intra-frame deviation consistency, precise image stabilization is achieved under multi-target constraints. Experiments based on Gaofen-4 satellite image sequences demonstrate that this method reduces the average position deviation of ship targets in the original images from 8.5 pixels (425 m) to 3.4 pixels (170 m), a decrease of approximately 59.41%, effectively improving the relative geometric accuracy of the image sequence and significantly eliminating target trajectory jitter.

Small object detection from panchromatic (PAN) and hyperspectral imagery (HSI) remains challenging because the two modalities differ substantially in spatial resolution and spectral representation. Most existing approaches process image fusion and object detection as separate stages, which increases computational cost and may propagate errors from the fusion stage to the detector. To address this issue, we propose a multi-level supervised framework for small object detection using PAN and HSI. First, an interactive loss function is introduced to enable joint training and to improve both detection performance and interpretability by injecting task-related supervision into the network. Second, a mechanism-guided image-fusion component combines ratio-transform-based fusion with deep feature learning. Finally, a CNN-transformer-based detection component with multiple spatial-spectral attention projection modules is constructed to extract spatial and spectral information effectively. Experiments on datasets collected by the Tiangong-1 and EO-1 satellites show that the proposed method outperforms several state-of-the-art approaches.

The Wankel rotary engine (WRE), renowned for its high power density and compact structure, faces challenges in starting performance due to periodic load torque fluctuations and sensor reliability issues in harsh operating environments. To address these limitations, this study proposes an integrated control strategy combining sensorless control and an improved sliding mode observer (SMO) with an optimized reaching law design. The sensorless control integrates the extended Kalman filter (EKF) for rotor position estimation and I-F control to achieve smooth start-up of the surface-mounted permanent magnet synchronous motor (SPMSM) under zero and low-speed conditions, eliminating dependence on physical position sensors. A sliding mode observer with an optimized reaching law is designed to estimate periodic load torque disturbances, enabling real-time feedforward compensation. Simulation results demonstrate that the proposed strategy ensures a smooth starting and seamless transition from standstill to target speed while reducing speed overshoot by 7.6% and suppressing steady-state speed fluctuations by 76.5% compared to uncompensated systems. These advancements significantly enhance the SPMSM-driven WRE's dynamic response accuracy, steady-state stability, and operational robustness under harsh conditions. The proposed framework provides a scalable solution for high-performance WRE control, particularly suited for compact, weight-sensitive systems requiring high power density and reliability.

In future sixth-generation (6G) communication systems, it is foreseen that complex communication scenarios and critical performance requirements will necessitate more flexible air interface configurations. Traditional air interface adaptation will no longer be applicable to 6G due to issues such as high computational complexity, sub-optimal trade-offs among multi-objective performance metrics, outdated configurations due to fast-varying channels, etc. In this paper, the relevant user behaviors, communication environment, and system are virtualized via the digital twinning technique. Then, a knowledge graph-based multi-objective recommendation framework is proposed to configure the digital twinning air interface to adapt to channel conditions, while balancing various service requirements. First, the knowledge graph is applied to reveal complex dependencies between the air interface and the service requirements, and more importantly, to reconcile possibly contradictory performance targets. Furthermore, the air interface configuration, empowered by the digital twin technique, is able to exploit predicted prior knowledge about user behavior and the channel characteristics, thus improving the utilization efficiency of wireless resources promptly. Moreover, the digital twin technique allows the candidate air interfaces to be virtually verified and compared with little effort. Finally, two case studies are presented to demonstrate the potential of the knowledge graph-based recommendation method for the digital twinning air interface.

. In this paper, for a class of SIR models with saturated incidence, the SIR model is discretized using the modified Euler method to form a form in which the coefficient matrix contains unknown parameters. The augmented error system method is constructed and the future information is simulated as a feed-forward and compensated into the SIR model, and the output regulation method of the linear discrete system is used to design the non-singular terminal sliding mode surface and the exponential convergence law, and the suitable performance index is given to obtain the optimal sliding mode preview controller. Finally, numerical simulation is utilized to verify the effectiveness of the theory and methodology of this paper.

This paper proposes an adaptive neural control strategy based on Stochastic Configuration Networks (SCNs) for a specific class of Caputo-type fractional order nonlinear systems. The overall control system consists of two components: a neural controller and a sliding mode controller. The SCNs are employed to approximate the unknown function terms present in the system. The SC-III algorithm is utilized to update the output weights, followed by the design of a sliding mode controller to ensure the system's stability. Finally, the effectiveness of the proposed control scheme is demonstrated through simulation examples.

A bionic compound eye (CE) vision system is inspired by examples from nature, such as the eyes of dragonflies, mollusks, and other beings. It is used for visual measurements and 3-D reconstruction at close range due to the large number of overlapping miniaturized subeyes, which allow such systems to be applied in robot navigation, autonomous vehicles, medical endoscopy, and others. The calibration of the CE is difficult due to distortions and the large number of optimized parameters. This work proposes a new method for CE modeling based on graph neural networks (GNNs). This model creates a 2-D to 3-D correspondence solving the problem of missing values that appears when an object is not captured in all subeyes. The obtained results verified better performance of the proposed model in the estimation of 3-D object coordinates and in visual measurement of Euclidean distance between objects, compared to a traditional calibration approach based on pinhole camera model as well as a method based on multilayer perceptron (MLP) model, where missing values are filled with zeros. Comparative analysis is done to validate a design of the proposed GNN-based model.

In this article, a neural network (NN) methodology is raised to resolve the optimum control issue for affine nonlinear systems with multi-input constraints over a finite-horizon. Unlike the value and policy iterations in conventional approximation dynamic programming (ADP) technology, which usually require a sufficient amount of iterations and multiple iteration loops to ensure the steadiness of the system and the convergence of system states and control strategies, a finite-horizon constrainedinputs optimum control methodology relying on the actor-critic structure is presented. This method can be utilized as time goes forward. The final cost function is considered, although the system state values converge to zero in the finite-horizon issue. The stability of the presented control algorithm is examined by means of the Lyapunov stability principle. Ultimately, a simulation case is provided to demonstrate the feasibility and effectiveness of the devised algorithm.

Channel knowledge map (CKM) has become a potential technique to enhance communication performance by exploiting actual radio propagation information, especially in communication between uncrewed aerial vehicles (UAVs) and ground base stations (GBSs). However, CKM constructed by existing methods cannot obtain differentiable expressions from locations to the channel information, rendering it unsuitable for the traditional communication design. This work proposes a site-specific differentiable CKM to jointly design UAV trajectories and transmit power. First, assuming sufficient channel samples collected by a GBS, the CKM is constructed for this specific site as a differentiable back propagation neural network (BPNN). To enable CKM migration towards nearby GBSs, we adopt the transfer learning mechanism to set up new CKMs that require significantly less training samples. Next, leveraging CKM-stored channel knowledge, we investigate the multi-UAV trajectory design and power control strategy, while the UAVs are traversing the network coverage area with designated starting and destination points. Specifically, the minimal average rate between UAVs and associated GBSs is maximized along the designed trajectories, which is solved by continuous convex optimization based on the differentiable CKMs. Numerical results show that the BPNN and transfer learning can effectively construct high-accuracy CKMs, while reducing the overall training cost. It is also shown that the proposed joint trajectory and power optimization based on the CKM-assisted architecture achieves improved minimal average rate compared to the alternating optimization method based on distance-dependent path-loss models and existing CKM-based methods with fixed power configurations, since both site-specific environmental information and power optimization are exploited.

Moving ship detection is vital for real-time maritime monitoring. Nevertheless, several challenges arise in this area: 1) wide-area images often need to be sliced into patches to detect tiny targets, which is inefficient; 2) the ships are small with almost no texture, leading to difficulties in accurate detection; and 3) the contrast between the ships and the ocean is relatively low, resulting in weak features. Although moving ships exhibit weak features, they often possess distinct wake trails. Capitalizing on this characteristic, we tailored a dual-head (DH) supervision network for moving ship detection. Initially, a DH supervision architecture is introduced to guide the model in using wake trails for target localization, thereby addressing the inefficiency caused by slicing. Subsequently, the background association head (BAH) and target confirmation head (TCH) are introduced to collaboratively enhance detection accuracy by leveraging the inter-head attention (IHA) mechanism. Finally, to address the issues of weak features, the dynamic feature enhancement module (DFEM) is embedded into the backbone to boost the model's feature extraction capability for moving targets. Experiments on the GaoFen-1 dataset demonstrated that our method significantly improved the efficiency and performance of infrared moving ship detection and reached the state-of-the-art (SOTA) performance. Source codes will be available at https://github.com/KTqizhi/KTqizhi.github.io

Robust kinematic calibration is essential for improving the position accuracy of industrial robot systems. However, traditional calibration methods often treat all measurement samples equally, neglecting the varying reliability of robot poses and measurement data. This paper proposes a confidence-weighted kinematic calibration method that explicitly incorporates both robot model uncertainty and measurement uncertainty into the calibration process for industrial robots. The confidence of each observation is defined as a function of the robot’s geometric sensitivity—derived from Jacobian-based error propagation. And the measurement residuals of the measurement system. The proposed confidence index is integrated into a RANSAC-based progressive sampling framework and a least-squares refinement, enabling robust parameter identification under heterogeneous noise conditions and outlier contamination. Simulations are performed to evaluate the proposed method under various levels of model and measurement uncertainty. The results show that the confidence-weighted calibration framework achieves faster convergence and higher parameter accuracy compared with least-squares approaches. The proposed approach improves the average position accuracy from 4.55 mm to 0.79 mm, offering am interpretable and generalizable paradigm for uncertainty-aware kinematic calibration in robotic systems.

The growing integration of renewable energy sources, especially photovoltaic (PV) systems, plays a vital role in enhancing energy efficiency and promoting sustainability. However, due to their high dependence on weather conditions, PV systems often exhibit significant intermittency and unpredictability, which complicates power system monitoring, control, and overall stability. To address these issues and strengthen situational awareness as well as operational reliability in PV-integrated grids, accurate and real-time Dynamic State Estimation (DSE) has become increasingly critical. In this paper, we propose the application of the Cubature Kalman Filter (CKF) for DSE in PV systems. The CKF offers a powerful nonlinear filtering framework, which is capable of accurately estimating the internal dynamic states of the PV systems. The simulation results demonstrate that the proposed CKF-based DSE approach effectively tracks the dynamic states of the PV system, thereby contributing to the improved accuracy of state estimation of the PV system.

Satellite communication is a key aspect of future 6G networks, and the impact of artificial intelligence technology utilizing deep learning on satellite communications has garnered significant interest. This paper outlines the current research status of deep learning applications in satellite communication from the perspective of the physical layer, data link layer, and network layer. It also examines the limitations of deep learning in satellite communication applications and anticipates potential research directions for the future.

Accurate and efficient ship tracking by geosynchronous orbit (GEO) satellites holds great significance for large-scale maritime surveillance. Nevertheless, ship tracking continues to grapple with a multitude of challenges as follows: 1) the targets are small and often obscured by cloud interference, leading to weakened features; 2) the contrasts between the ships and the background are relatively low, complicating the identification and tracking process; and 3) the frame-to-frame relative positioning accuracy is poor, posing difficulties in reflecting the actual movement trends of ships. In response to these challenges, we proposed TS-Track, a novel framework employing multilevel supervision paradigm to improve tracking performance. Initially, this framework restructured the tracking task into three key sub-modules: image enhancement, object tracking, and trajectory adjustment, inherently fostering a unified training protocol that naturally encompasses all components. Subsequently, a trajectory-based frame fusion strategy was proposed, utilizing consecutive three-frame images to enhance target features and produce consistent motion feature patterns; Last but not least, a trajectory adjustment network was developed to correct the position of ships during tracking, resulting in stable tracking trajectories, and reproduce the actual movement trends of ships. The experimental results on GaoFen-4 dataset validated that our method delivered a significant improvement in ship tracking and achieved state-of-the-art (SOTA) performance. Source codes are available at https://github.com/KTqizhi/KTqizhi.github.io.

To overcome the limitations of a single USV in task execution, this paper investigates the issues of cooperative motion and obstacle avoidance for USV swarm. The leader-follower control approach is chosen as the strategy for coordinating the formation of multiple USVs. Kinematic and kinetic controllers are designed using sliding mode control, with a saturation function replacing the sign function to avoid chattering. The validity of the kinematic and kinetic controllers is demonstrated using the Lyapunov stability theorem. To tackle the problem of goal reachability when obstacles are nearby in the artificial potential field method for obstacle avoidance, the repulsive potential field function is modified by including the distance factor between the USV and the target point. To address the problem of local minimum, the simulated annealing algorithm is incorporated into the traditional artificial potential field method. Multi-condition simulation experiments are conducted using MATLAB. The experimental results demonstrate that multiple USVs can form a formation in a short time and maintain the formation while navigating, exhibiting high robustness. Furthermore, they are able to effectively navigate through complex marine environments, successfully avoiding obstacles and ultimately reaching their target destination.

The article focuses on collaborative on-orbit inspection involving cooperative target space-craft (TSc). Attitude maneuver planning of the TSc is implemented to ensure that its features can be observed by multiple closely accompanying observers. A method for discriminating the visibility of features is proposed based on the spacecraft's structural dimensions. The article introduces a double-loop optimization framework: the outer loop adopts rolling planning to determine the expected attitude, while the inner loop employs the deep deterministic strategy gradient algorithm (DDPG) to quickly solve the maneuvering trajectory and determine the associated cost. The planning process considers the pointing constraints of the optical load and the maneuverability constraints of the executing mechanism. Finally, the feasibility of the optimization framework is rigorously validated through the simulation examples provided.

With the increasing availability of simultaneous panchromatic and hyperspectral images, object detection methods based on them have demonstrated significant application advantages. However, they still face several challenges that limit detection performance: 1) the sizes of small objects remain very small even in the fused images, insufficient texture information and spectral information that is easily confused, leading to lower accuracy in object detection; 2) etection-by-Preprocess (DBP) methods often suffer from spectral and spatial detail distortions, compromising target features; 3) Preprocess-free detection (PFD) methods extract panchromatic and hyperspectral features directly through networks, but the black-box nature of deep network makes it difficult to ensure precise alignment of these two types of features, thereby hindering further improvements in detection accuracy. Therefore, this paper proposed a novel Collaborative Heterogeneous Supervision Network (CHS-Net) for small object detection on panchromatic and hyperspectral images. First, integrating fusion and detection components into a heterogeneous supervision network enhances learning capabilities by incorporating more empirical knowledge. Second, a unified joint regulation strategy was introduced to enhance integrated learning capabilities using optimized feedback loss functions. This approach enhanced the attention of different components to target features, effectively improving weak small target detection performance. Finally, comparative experiments based on EO-1 dataset demonstrate that the proposed method outperforms many start-of-the-art approaches.