Yongsheng Qian

Restoring high-speed railway services after severe disruptions requires continuous adaptation as recovery conditions evolve. While traditional delay-focused rescheduling models strive for mathematical optimality, they frequently overlook the practical burden of repeated instruction revisions and concentrated dispatching issuance. To bridge this gap between theoretical planning and operational implementability, we develop a practice-oriented spatiotemporal network model that incorporates uncertain recovery states. This model captures the probabilistic transitions from complete blockage to normal operation while explicitly retaining previously issued instructions and their execution states. By establishing a defined service-loss bound, the approach sequentially optimizes both the total instruction count and peak issuance. We propose a safety-verified model predictive control framework combining masked trajectory ranking and a rolling horizon architecture (Safe-M³PC-RH). An offline-trained model efficiently screens candidate paths to reduce the online search space, a mixed-integer linear program determines joint train decisions, and an independent module verifies safety constraints before execution. Evaluated using operational data from the Beijing-Shanghai high-speed railway, the framework demonstrates that explicit instruction optimization significantly reduces dispatching interventions. Without deteriorating service loss, average reductions in instruction count and peak issuance reach 8.55% and 9.41%, respectively. Introducing a marginal 1% tolerance on expected service loss further increases these reductions to 18.97% and 23.52% compared to service-priority solutions. These improvements are primarily achieved by replacing frequent train-order changes with stable holding and stopping adjustments. Ultimately, the model quantifies the critical trade-off between service performance and dispatching workload, offering a practical framework for transport scheduling problems where human-centric plan consistency is essential.

With the rapid development of high-speed rail and passenger-dedicated lines, the load in the throat area of railway passenger stations is increasing. Asymmetric characteristics, including turnout resource competition and dynamic assignment of arrival–departure tracks, are notably pronounced in the throat area; nevertheless, there is a lack of adequate research addressing these issues. This paper took the Lagrange coordinate system as the foundation and adopted the multi-dimensional cellular automaton and classical NaSch model. Treating trains as a one-dimensional continuous medium, we proposed a cellular automaton model for train operations in the throat area of a railway passenger terminal. The improved cellular automaton inherited the core vehicle update rules of the NaSch model, and its stochastic rules were adapted and simplified to match the practical operating features of railway station throat areas. Considering stop time, departure interval, and mixed-traffic ratio as three influencing factors, the simulation analyzed the average delay time of three speed types of trains in the throat area and the respective impacts of these factors on the throat area through capacity. The results showed that this model can depict the running status of the train in the throat area realistically and can provide a theoretical basis for subsequent research and analysis.

Coordinating public service supply and demand is essential to urban social sustainability. This study examines ten suburban districts of Beijing at six observation points from 2015 to 2024 and develops an integrated framework encompassing resource endowment, demand pressure, supply–demand matching, and spatial equity. Resource endowment and demand pressure are measured using spatial accessibility and objective weighting methods, while relative matching coefficients, Lorenz curves, and Gini coefficients are employed to evaluate spatial mismatch and equity. The results reveal a shift toward a dynamic polycentric supply pattern, persistent relative undersupply in Tongzhou and Daxing, and dimension-specific deficits in several otherwise balanced districts. Overall spatial equity improved during the study period, although localized mismatches remained. Public service planning should therefore move from uniform expansion toward demand-responsive, place-specific, and service-specific optimization.