In this paper, we investigate an air-to-ground (A2G) wireless network system where multiple uncrewed aerial vehicles (UAVs) provide downlink communication coverage for mobile ground users (GUs). This system accounts for UAVs progressively depleting their energy during coverage provision, ceasing operations when their energy reserves fall below a predefined threshold. We aim to maximize cumulative system throughput over the task period while satisfying the minimum fairness requirement through joint trajectory control and resource allocation (JTCRA) optimization. To meet the fairness requirement, enhancing system robustness is critical; energy-sufficient UAVs must autonomously assist GUs that lose connectivity when their serving UAVs terminate operations. Therefore, we propose a multi-agent deep reinforcement learning (MADRL) framework with a parameter-sharing architecture to solve this problem. As conventional parameter sharing is restricted to homogeneous agents with identical observation-action spaces, we design a dual-agent structure: a trajectory agent (Traj-agent) and a communication agent (Comm-agent) are deployed for each UAV. This separation organizes the heterogeneous tasks of trajectory control and resource allocation into distinct homogeneous agent groups, facilitating effective parameter sharing within each type. Based on this framework, we apply two alternative algorithms: an MAPPO-based JTCRA algorithm and a QMIX-based JTCRA algorithm. Simulation results demonstrate the superiority and effectiveness of our proposed JTCRA algorithms, which maintain service continuity for GUs through intelligent trajectory control, thereby minimizing the adverse impact of coverage gaps.

To address the challenges of fire and smoke detection in complex environments, including small early-stage targets, large scale and shape variations, weak visual features, and lightweight real-time deployment requirements, this paper proposes LMFSNet, a lightweight multi-scale collaborative detection network. LMFSNet consists of several coordinated components. The lightweight multi-scale feature extraction module enhances target representation under scale variations with low computational cost. The efficient multi-scale context fusion module strengthens contextual discrimination, improving the recognition of ambiguous fire and smoke features while suppressing background interference. The lightweight fusion path based on the Simplified Feature Transfer Module enables efficient multi-scale information exchange under real-time resource constraints. In addition, the SlimHead detection head with a high-resolution P2 detection layer and the multi-scale adaptive loss improves sensitivity to small early-stage targets and enhances regression stability. To support model training and evaluation, we construct a high-quality multi-scenario fire and smoke dataset. Experimental results show that the proposed method improves detection accuracy while maintaining high inference efficiency, providing an effective solution for real-time fire early warning in complex environments.

Optimizing unmanned aerial vehicle (UAV)-assisted wireless networks to serve mobile users (MUs) via beamforming presents significant challenges, mainly due to the dynamic and complex environments. Traditional single-modal data-based modeling methods are often insufficient for capturing the varying environmental characteristics, leading to inaccurate UAV trajectory design and beamforming. To address these issues, we propose a multi-UAV-assisted integrated sensing, communication, and computation (ISCC) framework that processes multi-modal data to enhance environmental awareness and improve communication performance. We then formulate an optimization problem to maximize the average sum rate by jointly optimizing the UAV trajectory and beamforming vectors. Given the non-convex nature of the problem, traditional optimization techniques are inadequate. To this end, we introduce a fine-tuned multi-modal large language model (M2LLM)-driven deep reinforcement learning (DRL)-based joint optimization framework. Specifically, a pre-trained M2LLM is first fine-tuned to predict future MU positions by leveraging historical multi-modal data, including texts, images, and wireless sensing data. The fine-tuned M2LLM is then employed to extract environmental features, where the output of the fine-tuned M2LLM’s last hidden layer is regarded as the environment state vector to eliminate the output uncertainty of the M2LLM. Subsequently, we use a DRL agent to optimize the UAV trajectory and beamforming in a coordinated manner. Extensive simulation results demonstrate that the proposed framework can significantly enhance network performance by enabling environment-aware and adaptive trajectory design and beamforming. The code is available in https://huggingface.co/blYin/MmllmDrlUavTdBf.

In Wi-Fi mmWave networks, blockage and mobility cause frequent beam switching, risking link stability. In 802.11ad, neighboring beam search method is used to select optimal beams, which only tests the limited candidate directions near the current beam, possibly leading to suboptimal communication performance. To improve the link stability, in this paper we propose a sub-7GHz and mmWave integrated Wi-Fi network architecture, where the omnidirectional sub-7GHz band is leveraged to improve the network coverage. To enhance the beam tracking performance, we further propose a CNN-LSTM based beam tracking scheme, where the CNN-LSTM is leveraged to extract features from both sub-7GHz and partial mmWave channel state information (CSI). Simulation results show that our proposed scheme can improve beam tracking accuracy, reduce beam search time and achieve higher system capacity.

High-speed maglev trains represent a key direction for the future development of rail transportation. As operating speeds increase, they face increasingly severe aerodynamic challenges. The streamlined aerodynamic shape of a maglev train is a critical factor influencing its aerodynamic performance, and optimizing its length plays a significant role in improving the overall aerodynamic characteristics of the train. In this study, a numerical simulation model of a high-speed maglev train was established based on computational fluid dynamics (CFD) to investigate the effects of streamline length on the aerodynamic performance of the train operating on an open track. The results show that the length of the streamlined section has a pronounced impact on aerodynamic performance. When the streamline length increases from 8.3 to 14.3 m, the aerodynamic drag of the head and tail cars decreases by 16.2% and 32.1%, respectively, with reductions observed in both frictions drag and pressure drag-the latter showing the most significant decrease in the tail car. Moreover, the extended streamline length effectively suppresses flow separation on the train body surface. The intensity of the positive pressure region on the upper surface of the head car streamlined section is reduced, directly leading to a 38.2% reduction in lift. This research provides a theoretical basis for the parametric design of aerodynamic shapes for high-speed maglev trains and offers guidance and recommendations for drag and lift reduction optimization.

To address the limitations of traditional algorithms in detecting small UAV targets, An improved YOLOv8-based UAV detection network is proposed. First, a re-parameterization. RepVGG-style backbone replaces the original YOLOv8 backbone to enable more detailed multi-scale feature extraction and fusion. Second, a content-guided attention fusion module (CGAFusion) is integrated into the neck network, dynamically adjusting the feature weights to enhance the focus on UAV-specific features. Finally, the normalized Wasserstein distance (NWD) is proposed to enhance regression loss optimization and minimize sensitivity to minor target position deviations. Experimental results show that the proposed network achieves a 2.7% improvement in mean average precision (mAP) over compared models.

In millimeter wave (mmWave) communication systems, beam alignment is doomed to play a vital role in ensuring directional link performance. In this paper, we propose a novel transformer-based angle prediction scheme to achieve fast and effective beam alignment. Transformer is one of the hottest seq2seq models in recent times, which is utilized to build the mapping relationship between the geographic position and beam alignment angles of users in this paper. Simulation results demonstrate the performance of the proposed scheme in terms of prediction accuracy and achievable sum rate.

The introduction of Integrated Sensing and Communication (ISAC) technology in high-speed railway mobile networks (HSRMNs) addresses reliability concerns within existing railway operation control systems. However, current ISAC research often focuses on overall performance assessment across extended periods, potentially overlooking practical sensing requirements, resulting in degraded or even unstable communication performance. To mitigate these challenges, we propose an Integrated On-Demand Sensing and Communication (IDSAC) mechanism for HSRMNs. IDSAC optimizes train communication energy efficiency (EE) by jointly optimizing target sensing selection and transmit beamforming, while adhering to Age of Information (AoI) constraints and ensuring beam pattern gain for designated targets. To tackle the non-convex mixed-integer programming challenge, we decompose the problem into two parts. First, we employ iterative algorithms using successive convex approximation (SCA) and semidefinite relaxation (SDR) to address the EE problem. Second, we introduce a dynamic control algorithm employing the Lyapunov drift-plus-penalty method to manage AoI constraints and sensing selection. Our joint optimization approach achieves optimal EE under dynamic operational constraints typical of high-speed rail systems. Simulation results validate the effectiveness of IDSAC in optimizing ISAC performance, enhancing railway safety and communication efficiency.

Real-world systems frequently encounter data loss caused by external interference or channel congestion. Such unreliable transmission can significantly impact system dynamics. To mitigate its effects on Boolean control networks (BCNs), this paper investigates the asymptotic feedback set stabilization of BCNs with missing data. Firstly, an augmented system is constructed to handle the data loss. Subsequently, two methods are proposed to design the feedback control, with Q-learning employed to address situations where a model-free approach is required. Finally, an illustrative example is presented to demonstrate the effectiveness of the proposed results.

As an important part of the high-temperature superconducting (HTS) maglev system, the performance and stability of the permanent magnet guideway (PMG) is directly related to the loading capacity and safe operation of the HTS maglev train. Therefore, it is necessary to carry out optimization research on the performance of PMG. At present, PMG optimization studies are mostly from the perspective of magnetic circuit, and seldom considered from the perspective of statistical analysis. In this paper, a PMG optimization method based on sensitivity analysis is discussed. First, the PMG dataset is established by simulation with parametric sweeping method. Then, the influence of permanent magnet (PM) dimension and grade on the electromagnetic force of HTS maglev system is investigated using sensitivity analysis method. And the optimization suggestions for the HTS maglev system are given. Finally, the optimization method is applied to present a PMG optimization case, and the optimization results show that the maximum vertical magnetic field is increased by 16.57 %, the maximum lateral magnetic field is increased by 10.97 %, the levitation force of the HTS bulks is increased by 3.23 %, and the guidance force is increased by 11.78 %.

High-temperature superconducting (HTS) maglev technology offers significant potential for high-speed transportation due to its zero magnetic resistance. However, the inherent low damping characteristics of HTS maglev systems make them highly vulnerable to external disturbances, leading to reduced system stability, passenger discomfort, and thermal instability of superconducting magnets. Therefore, it is essential to simultaneously address vibrations in both the vehicle body and the suspension frame. To address this challenge, an electromagnetic damper is introduced between the vehicle body and the suspension frame, along with a skyhook-groundhook control strategy to collaboratively regulate vibrations of both components. The parameters of control algorithm are optimized using the entropy weight method, focusing on multiple dynamic performance indicators. Simulation results indicate that the proposed approach significantly reduces vibrations in the suspension frame and vehicle body simultaneously compared to traditional skyhook-only or groundhook-only control strategies. Consequently, the stability and ride quality of HTS maglev systems are enhanced, making them more suitable for high-speed transportation applications.

Hybrid electric vehicles (HEVs) are perceived as transitional products bridging the gap between fueled vehicles and electric vehicles (EVs) because people intuitively believe that EVs are more environmentally friendly than HEVs. But is this perception true in the context of carsharing services (CSSs)? This paper pioneers a general large-scale multi-objective simulation-optimization (MOSO) method to explore the values of deploying HEVs in CSSs. We firstly develop a physically logical simulation model, emulating operations of CSSs and capturing mesoscopic dynamics of shared vehicles in a link-based traffic network. This model adopts an event-driven discrete-event mechanism, enhancing efficiency while maintaining high fidelity. Subsequently, we design a simulation-optimization framework aimed at achieving Pareto optimality by jointly optimizing station capacity, fleet size, and trip pricing. The goal is twofold: to maximize operational profits and to minimize carbon emissions, thereby quantitatively analyzing the potential of shared HEVs (SHEVs). To tackle the high-dimensional MOSO problem, we introduce the multi-objective optimization into stochastic approximation field by proposing a general algorithm that incorporates the multiple gradient descent algorithm with the simultaneous perturbation stochastic approximation algorithm. Furthermore, we derive its analytical expression for bi-objective optimization problems. We theoretically prove and practically demonstrate its strong global convergence. The efficiency of this method was validated through large-scale computational experiments conducted in Chengdu, Sichuan Province, involving 66,710 decision variables. These experiments showcased the method's superiority over existing MOSO algorithms. Several groups of sensitivity experiments focusing on vehicle types and traffic scenarios reveal some interesting findings. (1) Regardless of the increase in travel distances, SHEVs, which can be viewed as shared EVs (SEVs) without range anxiety (RA), continue to primarily rely on electricity rather than fuel for their operational mileages. This high utilization of electricity results in lower carbon emissions compared to SEVs. (2) Under any traffic condition, the dual-engine feature of SHEVs significantly reduces the number of failed pickups. (3) As travel demand increases, the state of charge for SEVs may rapidly fall below the threshold that triggers RA, whereas SHEVs maintain a more reliable power supply.

To avoid the impact of hard samples on the training of the Flying Bird Object Detection model (FBOD model), the Self-Paced Learning strategy with Easy Sample Prior Based on Confidence (SPL-ESP-BC) is proposed. First, the loss-based Minimizer Function is improved and a confidence-based one is introduced, making it more suitable for one-category object detection. Second, to enable early-stage judgment of easy and hard samples using SPL, an SPL strategy with Easy Sample Prior (ESP) is put forward. The FBOD model is initially trained with easy samples via the standard strategy and then continues with all samples using SPL. Combining ESP and the confidence-based Minimizer Function, the SPL-ESP-BC strategy is formed. Training the FBOD model with this strategy helps it learn flying bird object features in surveillance videos from easy to hard. Experimental results show that compared to the non-distinguishing standard training strategy, the Average Precision at 50 AP_50 ) of the FBOD model trained by SPL-ESP-BC increases by 2.0 https://github.com/Ziwei89/FBOD-BSPL .

Due to the distinctive communication environment in high-speed railway (HSR) scenarios, propagation paths are mostly in line-of-sight (LoS) environments, leading to the problem of insufficient channel matrix ranks in MIMO systems in scenarios like viaducts with weak multipath effects. In this letter, to address the issue, we propose a reconfigurable intelligent surface (RIS) assisted MIMO transmission scheme for HSRs. To optimize the MIMO transmission performance, we employ the deep deterministic policy gradient (DDPG) algorithm to dynamically adjust the phases of RIS in response to varying wireless channel conditions. Simulation results demonstrate that our proposed scheme can significantly improve the transmission capacity compared to traditional schemes without RIS.

Despite its significant application potential,low-vacuum ultra-high-speed rail transit faces several pressing scientific and technical challenges.This paper presents a comprehensive review of low-vacuum tube ultra-high-speed rail transit technology,covering its origins,development,and current international research trends.The necessity and feasibility of developing this technology are then an-alyzed,followed by a detailed discussion on the adaptability of magnetic levitation technology for trains operating in low-pressure envi-ronments,as well as the primary technical challenges,such as heat dissipation and insulation issues.Furthermore,the paper proposes technical solutions and measures for rescue and evacuation specific to low-vacuum tube transportation systems.

High-Frequency (HF) communication is widely used for long-distance transmission in remote and disaster areas. However, the dynamic nature of the ionosphere and multipath propagation in the HF channel pose significant challenges to designing efficient and robust communication systems. In this paper, we propose a Fixed Station (FS) and frequency matching method, as well as a power allocation method, to improve the sum-rate of air-to-ground HF communication networks. We derive optimal power allocation among users that share the same frequency, based on which a modified water-filling algorithm is used to solve the power allocation problem in multi-user scenarios, while a low-complexity algorithm is proposed to solve the integer optimization problem of frequency-FS matching. Simulation results demonstrate that the proposed algorithm outperforms the naive algorithm, indicating its effectiveness.

Through human-machine knowledge interaction, the interpretability of model decisions is improved, and the reliability and optimization problems of intelligent models are overcome by knowledge enhancement guided by expert feedback and evaluation information, so as to achieve continuous optimization of decision level and improvement of model generalization ability. Effectively improve the reliability level of the action scheme of the decision-making model, realize the recommendation of control actions under various complex operation scenarios, reduce the work burden of dispatchers, and improve the safety, stability and economy of power grid operation. Based on the feedback knowledge of power experts and the improved strategy gradient reinforcement learning algorithm, how to combine artificial intelligence model with expert logic is proposed. By constructing the memory reasoning framework of autonomous decision-making agents, the autonomous decomposition of regulatory tasks and automatic invocation of various executable small models are realized, and the decision-making level of artificial intelligence models is improved.

High-temperature superconducting (HTS) pinning magnetic levitation (maglev) vehicle technology offers stable levitation and guidance without active control or directional magnetoresistance, offering a groundbreaking solution for the rail transport sector. However, HTS bulks are sensitive to changes in critical current, magnetic field, and temperature during vehicle operation, potentially leading to levitator failure. This article examines how such failures affect the dynamics of an HTS pinning maglev vehicle. First, we derive a model of levitation force decay by analyzing the reduction in levitation force when HTS bulks lose superconductivity. Using this model, we then develop a dynamic model of the maglev vehicle that accounts for the levitator failure. Finally, we analyze vehicle dynamics in scenarios that involve one or multiple levitator failures. Our findings indicate that when a single levitator at a bogie corner fails, the vehicle's minimum levitation height decreases from 8.78 mm to 5 mm, the maximum roll angle reaches 0.09(degrees), and the maximum yaw angle reaches 0.16(degrees). When all four corner levitators fail, the minimum levitation height decreases from 8.78 mm to 5.22 mm, with a maximum roll angle of 0.03(degrees) and yaw angle of 0.04(degrees). These results provide valuable information to ensure the future safety of HTS maglev vehicles.

Superconducting pinning Maglev, with its passive self-stabilization, low resistance, and noise, holds promise as the next generation of high-speed rail transit. The vehicle dynamic simulation is essential for high-speed operation, and the interaction between the high-temperature superconductor (HTS) and the permanent magnet guideway (PMG), termed the “HTS-PMG Relation,” is fundamental to understanding the system’s dynamics. This paper presents an efficient computation model for the HTS-PMG relation, employing flux penetration characteristics to reduce the dimensionality of the physical field. The model integrates Maxwell’s equations with the Bean E-J constitutive relationship, resulting in reduced-dimensional governing equations and a nonlinear boundary finite difference method (FDM) optimization algorithm. Validated through quasi-static and dynamic experiments, this model accurately simulates the complex hysteresis behaviors of pinning Maglev and significantly enhances computational efficiency compared to traditional finite element methods. The proposed fast HTS-PMG relation computation model enables large-scale dynamic simulations of superconducting Maglev trains and offers a novel approach to studying their dynamic performance and levitation drift.

Objective Ensuring the safety and stability of train operations is essential in the advancement of railway intelligence.The growing maturity of Wireless Sensor Network(WSN)technology offers an efficient,reliable,low-cost,and easily deployable approach to monitoring railway operating conditions.However,in complex and dynamic maintenance environments,WSNs encounter several challenges,including weak signal coverage at monitoring sites,limited accessibility for tasks such as sensor node battery replacement,and the generation of large volumes of monitoring data.To address these issues,this study proposes a multi-Unmanned Aerial Vehicle(UAV)-assisted method for data collection and computation offloading in railway WSNs.This approach enhances overall system energy efficiency and data freshness,offering a more effective and robust solution for railway safety monitoring. Methods An intelligent data collection and computation offloading system is constructed for multi-UAV-assisted railway WSNs.UAV flight constraints within railway safety protection zones are considered,and wireless sensing services are prioritized to ensure preferential transmission for safety-critical tasks.To balance energy consumption and data freshness,the system optimization objective is defined as the weighted sum of UAV energy consumption,WSN energy consumption,and the Age of Information(AoI).A joint optimization algorithm based on Multi-Agent Soft Actor-Critic(MASAC)is proposed,which balances exploration and exploitation through entropy regularization and adaptive temperature parameters.This approach enables efficient joint optimization of UAV trajectories and computation offloading strategies. Results and Discussions(1)Compared with the Multi-Agent Deep Deterministic Policy Gradient(MADDPG),MASAC-Greedy,and MASAC-AOU algorithms,the MASAC-based scheme converges more rapidly and demonstrates greater stability(Fig.4),ultimately achieving the highest reward.In contrast,MADDPG exhibits slower learning and less stable performance.(2)The comparison of multi-UAV flight trajectories under different algorithms shows that the proposed MASAC algorithm enables effective collaboration among UAVs,with each responsible for monitoring distinct regions while strictly adhering to railway safety protection zone constraints(Fig.5).(3)The MASAC algorithm yields the best objective function value across all evaluated algorithms(Fig.6).(4As the number of sensors and the AoI weight increase,UAV energy consumption rises for all algorithms;however,the MASAC algorithm consistently maintains the lowest energy consumption(Fig.7).(5)In terms of sensor node energy consumption,MADDPG achieves the lowest value,but at the expense of information freshness(Fig.8).(6)Regarding average AoI performance,the MASAC algorithm performs best across a range of sensor densities and AoI weight settings,with the greatest improvements observed under higher AoI weight conditions(Fig.9).(7)The AoI performance comparison by sensor type(Table 2)confirms that the system effectively supports priority-based data collection services. Conclusions This study proposes an MASAC-based intelligent data collection and computation offloading scheme for railway WSNs supported by multiple UAVs,addressing critical challenges such as limited WSN battery life and the high real-time computational demands of complex railway environments.The proposed algorithm jointly optimizes UAV flight trajectories and computation offloading strategies by integrating considerations of UAV and WSN energy consumption,data freshness,sensing service priorities,and railway safety protection zone constraints.The optimization objective is to minimize the weighted sum of average UAV energy consumption,average WSN energy consumption,and average WSN AoI.Simulation results demonstrate that the proposed scheme outperforms baseline algorithms across multiple performance metrics.Specifically,it achieves faster convergence,efficient multi-UAV collaboration that avoids resource redundancy and spatial overlap,and superior results in UAV energy consumption,sensor node energy consumption,and average AoI.