To address the issue that vehicle behavior prediction based on in-vehicle monitoring cameras may infringe on the privacy of drivers and other road participants, this study develops a bus activity prediction model with vehicle motion and behavioral operation data as input. First, a naturalistic urban bus driving experiment was conducted, collecting vehicle motion and driver operation data via the CAN bus. Subsequently, segments corresponding to station entry, station exit, intersections, turning, and lane changing were selected. The phase space reconstruction algorithm was used to map the time-series data into a high-dimensional space, thereby generating an RGB image dataset. Finally, an E-bus Vehicle Behavior Prediction Model (E-VBPM) was established based on the ConvNeXt network. The results indicate that the developed E-VBPM achieved an accuracy of 84.62% in predicting driving activities, representing an improvement of approximately 6.8% over machine learning models that utilize time-series data inputs. These findings support the development of more intelligent on-board systems for electric buses, enabling better identification of vehicle operating modes and enhanced driver assistance.

A flexible management strategy for dedicated lanes of connected and automated vehicles (CAVs) is proposed to ensure CAVs priority while enhancing overall road traffic flow when the market penetration rate (MPR) of CAVs is relatively low. A mathematical model for the flexible management strategy is constructed under a multi-lane scenario, incorporating parameters such as CAV market penetration rate, queue intensity, driving style, and road traffic demand. Numerical analysis is conducted to evaluate traffic flow under different lane management strategies across multiple scenarios, comparing the proposed flexible strategy with a typical mandatory management strategy. The analysis also considers the coexistence of human-driven vehicles (HDVs) with varying driving styles in non-dedicated lanes. Numerical results indicate that when a one-way two-lane road is configured with the inner lane as a dedicated CAV lane, the flexible management strategy consistently outperforms the mandatory strategy in terms of road traffic flow when MPR is below 50%, particularly when MPR is below 30%. As traffic demand increases from 2000 veh/h to 5000 veh/h, the advantage of the flexible management strategy becomes more evident, suggesting its suitability under low MPR conditions. Additionally, the flexible management strategy supports a broader MPR range to achieve the target traffic demand, with an earlier starting point of the effective MPR interval. This difference becomes more pronounced as the driving style of CAVs becomes more aggressive. However, when MPR exceeds 50%, both strategies exhibit similar traffic flow trends under identical driving styles. For a one-way three-lane scenario with the inner lane designated as a dedicated lane, the analysis yields similar results to the two-lane scenario. Further extending the study to a four-lane scenario, both one and two dedicated lane configurations are examined, leading to four dedicated lane management schemes. Overall, the flexible management strategy remains superior to the mandatory strategy under low and medium penetration rates. Furthermore, simulations based on SUMO in a three-lane scenario validate the effectiveness of the proposed lane management strategies.

To enhance vehicle trajectory prediction accuracy in complex structured scenarios such as roundabouts, a deep learning framework, namely MST (MHA-SGC-TimeMixer), is proposed. The framework is built upon a macro-micro dual-encoder architecture. At the macro level, a Multi-Head Attention (MHA) mechanism is employed to capture the long-term guiding constraints between vehicles and the global road topology. At the micro level, a Simplified Graph Convolutional Network (SGC) first extracts instantaneous spatial relationships among vehicles. Subsequently, the TimeMixer mechanism is introduced to map the one-dimensional interaction sequence into multi-scale, multi-resolution 2D spatio-temporal images. By explicitly decoupling and hierarchically fusing periodic tactical behaviors and trending strategic intentions, a precise capture of deep interaction patterns is achieved. The information streams from both levels are integrated via a gated fusion network and then fed into a Gated Recurrent Unit (GRU) decoder to generate the final trajectory. Experiments on the public INTERACTION and RounD datasets demonstrate the framework's effectiveness. Within a 5-second prediction horizon, the proposed model achieves an Average Displacement Error (ADE) and a Final Displacement Error (FDE) of 1.19m and 1.85m on the INTERACTION dataset, and 1.16m and 1.80m on the RounD dataset, respectively, outperforming all baseline models. The results indicate that hierarchically modeling macro-level global constraints and micro-level spatio-temporal interactions, particularly through the decoupling analysis of interaction patterns, can significantly improve trajectory prediction performance in complex scenarios.

With the rapid development of the road transport industry, the number of accidents involving operational vehicles has been continuously increasing, and the severity of these accidents has become a widespread concern. Especially, operational vehicles, due to their unique operational characteristics, face more complex traffic environments and risks. This study analyzes the impact of "unobserved heterogeneity" on the severity of operational vehicle accidents. Based on traffic accident data for operational buses and trucks in China, relevant variables are selected from four aspects: driver behavior, vehicle type, road characteristics, and environmental conditions, and a random parameters Logit model is constructed. By introducing random parameters, the model can effectively capture heterogeneity and uncertainty between individuals, improving its explanatory power and predictive performance. The SHAP method is further applied to analyze the direction, importance, and non-linear interactions between variables. The results show that, for operational buses, complex road shapes and vehicle types significantly increase the severity of accidents, especially the interaction effect between complex road conditions and improper operations, which notably raises the accident severity. For operational trucks, the interaction effect between hazardous material transport vehicles and complex road conditions is stronger, and speeding behavior significantly increases the probability of major accidents. The SHAP analysis quantifies the contribution of 10 multidimensional factors to accident severity, revealing that bus accidents are mainly influenced by road environment factors, while truck accidents are more significantly related to vehicle attributes. This further quantifies the differing impacts of human factors and environmental factors on the severity of accidents.

China has a large mileage base of low-grade highways, and accidents caused by brake failure of heavy-duty trucks occur frequently. To conduct an in-depth study on the braking behavior of truck drivers on continuous downhill sections of low-grade highways, this research, based on the field-measured data of six-axle articulated trains on low-grade highways during long downhill driving, identified the driver’s braking timing on downhill sections with complex horizontal and vertical alignments, analyzed and quantified the characteristics of truck drivers’ braking behavior on continuous downhill sections of low-grade highways. Furthermore, a joint simulation model of driver behavior (integrating Matlab/Simulink) was established in the TruckSim simulation software, and the reliability of the model was verified through comparison.The results show that:On horizontal curve sections, affected by the curve, drivers exhibit relatively high-intensity braking behavior at the end of straight sections and the start of curve sections;On straight downhill sections with a gradient of 2%~9%, the drivers’ braking magnitude is positively correlated with the gradient;When curves are classified into sharp curves, medium curves, and gentle curves based on radius, the average braking magnitude before entering sharp curves reaches 220 mm, while there is no significant difference in the average brake pedal stroke between medium curves and gentle curves, with their average braking magnitudes reaching 176 mm and 174 mm, respectively.Compared with the brake drum temperature rise prediction models in previous studies that did not consider horizontal alignment factors, the simulation model established in this research has higher prediction accuracy for brake drum temperature. It can provide a theoretical basis for strengthening traffic safety guarantees on low-grade highways.

Diesel monorail crane is one of the main equipment used for the transportation of equipment, personnel and materials in coal mine roadway. It has the advantages of large transportation capacity, strong adaptability of roadway, long continuous transportation distance, etc., which can greatly improve the underground transportation efficiency of coal mines, and reasonable braking control strategy is the fundamental guarantee of safe and efficient operation for monorail cranes. Therefore, a braking control strategy using visual perception is proposed to improve the braking safety and stability for monorail cranes. Firstly, a detection and recognition scheme of pedestrian obstacles in front of monorail cranes is constructed by binocular cameras, and the obstacle recognition and real-time distance measurement are realized with the help of visual detection and stereo matching algorithm. Secondly, with the real time obtained pedestrian obstacle distance information, the current running speed information and the load condition of monorail cranes, the overall braking control strategy for monorail cranes under different working conditions is designed with consideration of the braking stopping distance and braking deceleration requirements. Then, taking the uncertain disturbance during the running process into account, a real-time tracking control strategy of pump-controlled-motor system for monorail cranes is constructed using the disturbance observer and robust reverse step control principle. Finally, experimental verifications of the braking control strategy using visual perception for monorail cranes under different pedestrian obstacle distances, running speeds and simulated loads are carried out on the basis of a simulated monorail crane braking control test rig. The experimental results indicate that the real time distance measurement error of pedestrian obstacles is controlled within 4.4%, and the proposed braking control strategy using visual perception can effectively improve the safety and stability of monorail crane braking process.

The inverted recirculating planetary roller screw mechanism is a novel transmission screw that enables integrated nut–motor design, offering simplified fabrication, high load capacity, and long service life. The nut engages with multiple annular-grooved rollers, which further mesh with the screw threads to transmit power. This paper first proposes an innovative structural design of the inverted recirculating planetary roller screw mechanism, derives the dimensional range of the screw's unthreaded section, and establishes the parameter matching equations for its components, followed by virtual modeling. Secondly, the calculation formulas for the motion parameters of the inverted recirculating planetary roller screw mechanism are derived. A dynamic simulation model was developed using Adams software, and the results were analyzed. The relative error between the simulation and theoretical results ranged from 0.2% to 1.12%, validating both the simulation approach adopted in this study and the feasibility of the proposed mechanism. Finally, the variation patterns of the thread contact force and the collision force between the rollers and the boss slope in the inverted recirculating planetary roller screw mechanism were quantitatively analyzed from two perspectives: operating condition parameters and the slope angle of the boss. The axial contact force in the threaded section is minimally affected by the rotational speed and the boss slope angle, but is significantly influenced by the external load; the average axial contact force increases markedly with increasing external load. As the nut’s rotational speed and the screw load increase, the collision force between the rollers and the cam ring boss slope also increases. However, as the boss slope angle increases, the range of collision force decreases significantly. This study provides a valuable reference for optimizing the design of the inverted recirculating planetary roller screw mechanism.

To explore the influence of the non-uniformity of quenching temperature on large-sized thin-walled multi-chamber aluminum alloys, this study aims to investigate the influence patterns of process parameters and develop an optimization strategy. A numerical model of the quenching process was established based on the Workbench software platform to analyze the effects of quenching methods, the strength of the cooling zone, the nozzle spacing, and operating speed affect the temperature field during the quenching process. Subsequently, a response surface method was employed to perform multi-objective optimization of key process parameters. The research results show that the use of stepped quenching can improve the uniformity of the temperature field during the quenching of profiles and ensure the critical cooling rate in the sensitive area. As the cooling intensity of the strong cooling zone increases, the cooling rate of the profile increases, while the uniformity of the temperature field decreases. As the longitudinal nozzle spacing in the strong cooling zone increases, the temperature difference of the profile in the strong cooling zone decreases, and at the same time, the cooling rate of the profile also slows down. As the running speed of the profile increases, the cooling rate of the profile shows a trend of first rising and then falling. The response surface optimization method was adopted to explore the influence laws of the cooling intensity coefficient of the strong cooling zone, the running speed of the profile and the longitudinal nozzle spacing of the strong cooling zone on the cooling rate of the profile, and the optimal process parameters of the profile were obtained. The optimized scheme successfully reduced the quenching temperature difference and eliminated temperature recovery by employing an alternating cooling mode of mist and high-intensity jet cooling in the intensive cooling zone.

When the planetary gear transmission system operates under extreme conditions such as poor lubrication, it is prone to tooth surface pitting faults, which seriously affects the transmission accuracy and service life. In order to ensure the reliable operation of the system, it is necessary to deeply explore the dynamic characteristics of the system under pitting failure. In this paper, a nonlinear dynamic model of planetary gear transmission system coupled with multiple excitation factors is established by considering the time-varying meshing stiffness, time-varying friction, transmission error and backlash under the condition of pitting fault. The fourth-order Runge-Kutta numerical integration method is used to solve the vibration differential equation of the system. The vibration characteristics of the system under different rotational speeds and pitting faults are analyzed by time domain diagram, phase plane diagram, spectrum diagram, wavelet time-frequency diagram, three-dimensional spectrum diagram and bifurcation diagram. The results show that the system exhibits rich nonlinear dynamic behaviors at different rotational speeds, including periodic motion, multi-periodic motion and chaotic motion. The pitting fault of the gear tooth causes a sudden change in the vibration displacement of the system and reduces the stability of the system by changing the time-varying meshing stiffness. With the increase of the rotational speed, the amplitude of the system vibration caused by the pitting fault also increases. The accuracy of the model and calculation method is verified by building a dynamic characteristic test bench of planetary gear transmission system. This paper reveals the intrinsic relationship between pitting fault and system dynamic characteristics, and provides an important reference for fault diagnosis and condition monitoring of planetary gear transmission system.

Addressing the inherent trade-off among structural complexity, grasping force, and compliance in conventional multi-fingered hands, this paper proposes a modular design methodology that synergizes the transmission advantages of linkages and tendons. The finger modules integrate rigid linkage mechanisms with tendon-spring compliant systems to achieve coupled proximal/distal phalanx motion, delivering enhanced grasping force and compliance while maintaining low systemic complexity. Through analysis of force-displacement coupling relationships, kinematic models correlating fingertip contact forces with compliant joint rotations are established, quantitatively characterizing grasp compliance. The thumb employs a unidirectional tendon-driven mechanism, where mapping relationships between actuation angles and phalange rotations combined with staggered-stiffness torsional springs ensure grasping adaptability while minimizing volumetric footprint and enabling precise fingertip control. Modular architecture significantly streamlines manufacturing, installation, and maintenance processes, reducing overall costs. Experimental results demonstrate high compliance and adaptability: 11 precision grasps and 5 power grasps classified under the Feix Taxonomy are achieved, with single-finger lifting capacity reaching 98 N at conventional dimensions. The implemented hand has been successfully integrated into GAC R&D Center's GOMATE Humanoid Robot System, providing high-adaptability grasping solutions for humanoid service robotics.

As the core component of the precision positioning platform, the control accuracy of the macro micro composite driver directly affects the performance of the platform. To improve the control accuracy of macro micro composite drivers, a dual loop composite control strategy with a master-slave structure is proposed. In the design of the macro motion control loop, a dynamic model is established based on the principle of electromagnetic force drive. The PSO-LQR controller tuned by particle swarm optimization algorithm (PSO) and the traditional LQR controller are used to construct the controller. Through comparative analysis of numerical simulation and experimental testing, it is verified that the PSO-LQR controller has higher displacement positioning accuracy in the outer loop control. In the design of micro motion control loop, a dynamic model was established based on linear pressure magnetic equation, and a non singular terminal sliding mode controller was designed as the inner loop by integrating exponential convergence law and hyperbolic tangent function, effectively suppressing the system chattering phenomenon. To verify the effectiveness of the proposed control strategy, a macro micro composite positioning experimental platform was built using a laser interferometer, and its positioning performance was tested. The experimental results show that under the same conditions, compared with the traditional LQR controller, the PSO-LQR controller has superiority in step positioning and displacement tracking; The micro motion part can achieve a step size of 1 μ m without generating displacement overshoot. When the positioning target is 15 μ m, the displacement error under the control of the non singular terminal sliding mode controller is about 0.24 μ m, effectively suppressing system chattering. In terms of macro micro composite positioning, a laser interferometer was used for closed-loop calibration of the grating displacement sensor, and repeated experiments were conducted. The final positioning accuracy was 370 nm.

Train-to-train communication constitutes the foundational architecture for China’s railway-dedicated 5G-R communication systems, and achieving time synchronization for train to train communication is crucial for train operation safety. To address the issue of poor time synchronization performance caused by non-stationary wireless channels and transmission delays in train to train communication, this paper proposes an enhanced time synchronization approach based on variational mode decomposition combined with a bidirectional long short-term memory network incorporating time attention mechanisms. First, a 5G-R train-to-train communication clock model is established by analyzing the delay errors in train-to-train communication. Then, the VMD model is employed to decompose the train-to-train communication time series into intrinsic mode functions of different frequencies, thus isolating noise elements and enhancing the signal-to-noise ratio. Next, noise-dominated components are identified by calculating energy values, and wavelet soft thresholding is applied to denoise these components, enhancing the quality of the train to train communication synchronization time series. Finally, a TA-BLSTM network is proposed, which integrates a time attention mechanism into a bidirectional LSTM framework. This network extracts long-term temporal features from the train to train time synchronization sequence using the bidirectional LSTM, while the time attention mechanism dynamically captures temporal dependencies, enabling high-precision prediction and dynamic compensation of time synchronization deviations in train to train communication, thus achieving accurate time synchronization. Simulation experiments demonstrate that the proposed method can effectively achieve train-to-train time synchronization in both relay and non-relay communication scenarios. Compared with other methods, the proposed approach significantly reduces synchronization offset errors and offers faster convergence speed and greater stability during the train to train time synchronization process.

In complex electromagnetic environments, tactical wireless communication links face severe jamming threats, which can easily lead to communication disruption and adversely affect the stability and reliability of mission execution. To enhance the anti-jamming capability of wireless communication systems in dynamic interference scenarios, this paper proposes an adaptive anti-jamming architecture that integrates Reconfigurable Intelligent Surfaces (RIS) with Deep Reinforcement Learning (DRL). The proposed architecture improves the robustness of the communication link and the intelligence of decision-making from two dimensions: enhancing the strength of useful signals and generating dynamic anti-jamming strategies. In terms of methodology, the system first leverages the beamforming capability of RIS to actively manipulate the wireless propagation environment, thereby improving the channel signal-to-noise ratio, effectively suppressing interference, and accelerating the convergence of learning strategies. Next, frequency selection and power control are modeled as a Markov Decision Process. A greedy action selection strategy incorporating historical value estimation is introduced, forming a reinforcement learning framework based on Double Deep Q-Networks (Double DQN) with prioritized experience replay. The RIS-enhanced signal improves the stability of the learning strategy and significantly shortens the training period. Simulation results demonstrate that under typical jamming scenarios—such as wideband frequency sweeping, random pulse jamming, and intelligent adversarial games—the proposed method achieves an average communication success rate improvement of over 15% compared to approaches that employ either DRL or RIS alone. These results validate the robustness and broad adaptability of the proposed architecture in highly dynamic environments.

Addressing the anchor selection problem for the Decentralized Networked Navigation System Based on DDS (DDS-DNNS), a node location graph and a node location estimation model were constructed. The anchor selection problem was formulated as a D-optimal metric for maximizing the Fisher information matrix under a fixed cardinality constraint. Based on this, an optimization algorithm for anchor selection in DDS-DNNS was designed, leveraging graph topology. This algorithm capitalizes on the connection between graph topology and the D-optimal metric of the Fisher information matrix, approximately transforming the maximization of the D-optimal metric of the Fisher information matrix into maximizing the logarithmic determinant value of the dimensionality-reduced weighted Laplacian matrix. Furthermore, based on the relevant properties of set functions and Cauchy's alternating theorem, it was proven that the approximate optimization model is a non-regular, non-monotonic, and non-negative submodular maximization problem. Consequently, an improved greedy algorithm was devised to solve the approximate optimization model, incorporating sparse Cholesky decomposition based on approximate minimum degree sorting, lazy evaluation, matrix dimension preservation and permutation vector reuse, and Cholesky decomposition result reuse. It was also demonstrated that the algorithm possesses approximate optimization performance guarantees and a computational complexity significantly lower than that of the classical random greedy algorithm. Finally, the effectiveness of the algorithm was verified through experiments.