Abstract High-speed rail (HSR) stations are critical interfaces between national transport networks and urban systems, shaping both regional accessibility and local development. Yet, previous studies seldom offer a nationally comparable assessment of HSR station-area functional profiles, defined here as the combined performance and balance of transportation-related node functions and surrounding urban place functions within station areas. It remains unclear how railway-level attributes and city-level contexts jointly influence these profiles. This study proposes an integrated evaluation framework grounded in the node-place approach to quantify, classify, and explain the functional characteristics of HSR station areas. The framework combines a set of multidimensional indicators capturing transportation connectivity (node) and urban development attributes (place) through a weighted multicriteria evaluation to derive overall functional performance and incorporates an interpretable machine learning model to uncover nonlinear effects and interaction mechanisms among key determinants. A case study is conducted using 90 major HSR stations in China within the 15-min walking zone. Results show that the station-area function evaluation index is low on average but highly heterogeneous across space. Stations in eastern and northeastern China more often exhibit balanced node-place profiles, whereas stations in western regions more frequently display functional deficiencies. A common structural pattern is that node-related functions exceed place-related functions, indicating that network accessibility is not consistently matched by surrounding urban development. Railway-level attributes explain more variance than city-level factors, with station built time emerging as the most influential determinant. Its interactions with distance to the city center, station building area, and urban population shape distinct station-area functional profiles within threshold ranges. These findings provide an evidence base for differentiated station-area planning and targeted investments, supporting the transformation of HSR stations into integrated, multifunctional urban nodes.
Abstract Railways, as critical infrastructure, represent a key domain for digital and intelligent transformation. This transformation places new demands on railway emergency capability, including dynamic perception, intelligent decision-making, and coordinated response. However, traditional railway emergency capability evaluation systems have difficulty effectively quantifying these new requirements, resulting in certain limitations in adapting to the context of digital and intelligent empowerment. Based on existing railway emergency capability evaluation indicators, this study incorporates digital and intelligent factors to optimize the evaluation framework and develops a comprehensive analytical framework for railway emergency capability under digital and intelligent empowerment. A second-order confirmatory factor analysis (CFA) model is introduced to construct and validate the hierarchical measurement structure of railway emergency capability and to identify the relative contributions of different capability dimensions to the overall construct. In addition, the evidential reasoning (ER) algorithm is employed to integrate evaluation information and achieve the quantitative evaluation of railway emergency capability. Finally, China Railway Nanchang Group Co., Ltd. is selected as a case study to empirically assess railway emergency capability under digitally intelligent enabling conditions. The results indicate that the overall emergency capability of the Nanchang Group remains at a moderate level. The belief degree associated with the “Medium” grade is the highest, at 49.85%, followed by the “Good” grade at 35.00%, suggesting a moderately above-average performance that is broadly consistent with current development conditions. By integrating a second-order CFA and ER, the proposed approach enables more objective weight determination and a more rational classification of capability levels, and may provide both theoretical support and a practical reference for the evaluation and improvement of railway emergency capability.
Abstract Ride-hailing services using autonomous electric vehicles (AEVs) at transportation hubs face strong demand fluctuations and limited charging resources. During peak periods, pricing decisions directly affect passenger demand, while charging decisions determine vehicle availability. Treating these two decisions independently often results in inefficient operations under highly dynamic conditions. Therefore, a coordinated decision-making approach is required. This study proposes a joint pricing and charging decision framework for AEV ride-hailing systems at transportation hubs based on deep reinforcement learning (DRL). A simulation environment is developed to represent passenger arrivals, AEV state-of-charge dynamics, and charging station constraints. A proximal policy optimization (PPO)–based model, named JPCPPO, is designed to jointly determine pricing levels and charging thresholds at each decision stage using aggregated system state information. Through continuous interaction with the environment, the model learns adaptive strategies for demand regulation and fleet energy management under time-varying conditions. Simulation results under different environment scales show that the proposed JPCPPO model consistently outperforms fixed pricing strategies, heuristic methods, and conventional actor-critic baselines in terms of overall reward and service performance. The additional experiments indicate that the computational cost of model training remains stable as the environment scale increases, demonstrating the applicability of the proposed approach to large-scale transportation hub operations.
Abstract To enhance the safety and adaptability of intelligent connected vehicle (ICV) in complex environments, this study proposes a hierarchical trajectory planning and tracking framework centered on a personalized driving risk field (DRF). The core innovation is quantifiable driver behavior factor integrated into the DRF, enabling it to dynamically modulate risk perception based on inferred driving styles. This mechanism translates behavioral differences into a continuous risk landscape. Within this framework, kinematically feasible trajectories are generated using quintic and quartic polynomials and are optimized by a multiobjective cost function that balances DRF-based safety against comfort and efficiency. A decoupled controller combining lateral linear quadratic regulator and longitudinal dual proportional integral derivative ensures precise tracking. Extensive simulations, including sensitivity analysis and validation in mixed dynamic-static and unsignalized intersection scenarios, demonstrate the framework’s effectiveness. The main findings are: (1) The personalized DRF generates significantly differentiated risk perception, with aggressive vehicles induce a peak field intensity approximately 2.6 times higher than conservative ones. (2) This perception drives intelligent decision-making. For high-risk interactions, the ego vehicle employs earlier and stronger braking to shorten high-risk exposure yet achieves a higher passing speed, demonstrating an optimization strategy beyond safety rules. (3) The controller reliably executes these differentiated commands, with root mean square errors for trajectory, speed, and heading tracking errors remaining below 0.038 m, 0.15 m / s , and 0.0042 rad, respectively. The robust closed-loop performance from decision-making to execution has been demonstrated. The framework enables ICVs to execute context-aware, personalized strategies, enhancing interactive safety and intelligence.
Abstract Traditional traffic state evaluation systems for multilane highways predominantly focus on one-dimensional longitudinal kinematics, thereby systematically neglecting the impact of lateral spatial distribution and lane utilization imbalance. To address this limitation, this study proposes a multidimensional traffic system disorder evaluation method that integrates macroscopic flow regimes, microscopic driving behaviors, and lane distribution characteristics through a hybrid criteria importance through intercriteria correlation (CRITIC)–Entropy–technique for order of preference by similarity to ideal solution (TOPSIS) model. Utilizing empirical high-precision radar-video trajectory data, convergent validity analysis quantitatively demonstrates that the proposed index maintains a moderate but highly significant correlation ranging from 0.57 to 0.65 at p < 0.001 across temporal, longitudinal, lateral, and kinetic dimensions with four mainstream surrogate safety measures including time exposed time-to-collision, proportion of stopping distance, lateral conflict frequency, and crash potential index. Furthermore, unsupervised clustering identified a four-stage evolutionary pattern characterizing the flow transition from stable states to turbulent regimes. Our findings reveal that while high-density synchronized flow maintains temporary stability through driver adaptation, frequent lane-changing maneuvers trigger systemic oscillations. Severe disorder eventually degrades into a physical deadlock where extreme saturation structurally restricts lateral maneuverability, inducing asynchronous interlane stop-and-go oscillations and a complete loss of system self-organization. The developed evaluation approach provides a theoretical foundation for proactive risk monitoring and dynamic management strategies in highway traffic operations.
