Bulletin Board
Highlights
Supply-demand coordination optimization for energy self-sufficient highway intelligent and connected infrastructure, MA Xiaolei, et al
2026-03-09 TopSpatio-temporal differential memory attention model for traffic flow prediction, CAO Chunjie, et al.
2026-03-09 Top“Four chains””collaborative empowerment: a strategy research forhigh-quality development of the low-altitude economy throughmulti-path integration, LI Jun.
more..Airport delay propagation analysis based on causal complex network modeling
HUANG Haifeng;WANG Zhan;TIAN Yong;ZHANG Naizhong;LYU Yue;[Background] The increasing demand for air transport and complexity of route networks have given rise to flight delays, which tend to propagate across airport networks through flight connectivity. This can result in widespread operational disruptions. [Objective] This study aimed to reveal the propagation paths and underlying mechanisms of delay spread within airport networks,thereby providing theoretical support and decision-making guidance for flight scheduling optimization and delay mitigation. [Data] The analysis was based on FME segment-level operational data provided by the civil aviation administration of China(CAAC) for 2024. [Method] A hierarchical message-passing graph neural network(HMPGNN) was employed to identify the causal relationships of delay propagation between airports and to construct a dynamic delay propagation network.Route and traffic volume networks were also developed to compare topological structures using complex network theory, and a random forest model was used to identify the key factors influencing delay propagation. [Result] Small and medium airports exhibited higher causal intensity and stronger delay-propagation capabilities. Among all network indicators, the flight volume degree exerted the most significant influence and was the dominant factor driving delay propagation. [Application] The findings provide decision support for identifying critical airports in delay propagation and optimizing flight scheduling strategies, thereby enhancing operational resilience and improving the overall efficiency of the route network.
Robust runway scheduling under uncertain taxi-departure and flight-arrival times
LI Jian;ZHANG Junfeng;ZHOU Renhao;ZHU Ling;[Background] Efficient runway scheduling is crucial for reducing delays and improving operational efficiency. However, existing deterministic scheduling models cannot effectively capture uncertainties in real-world operations, and previous studies on uncertainty largely overlooked the integrated operation of arrivals and departures, thereby failing to fully adapt to practical operational requirements. [Objective] To address the integrated runway-scheduling problem while considering uncertainty, a novel two-stage stochastic optimization approach is proposed to achieve robust runwayassignment and-scheduling schemes for arrivals and departures. [Method] Given the uncertainties of taxi-departure and flight-arrival times, the first stage determines tactical runway-assignment schemes,and the second stage optimizes aircraft-takeoff and-landing times. The sample average approximation method is adopted for model reformulation, and the logic-based benders decomposition(LBBD)algorithm is employed to realize an efficient solution approach for the model. [Data] Validation and analysis are conducted based on the actual operational data of the Guangzhou Baiyun Airport on December 3, 2023. [Result] Compared with direct optimization using a commercial solver, the LBBD algorithm improves the average computational efficiency by approximately 33 times while obtaining high-quality solutions. In contrast to three benchmark runway-assignment strategies, the optimized assignment strategy significantly reduces flight delays. Robustness analysis confirms the significance of incorporating uncertainty into runway-assignment decisions, whereas sensitivity analysis provides practical operational insights. [Application] This optimization framework and solution method can be applied to the optimization of arrival-and departure-scheduling at busy airports, thereby providing scientifically grounded tactical decision support for airport-operation managers.
Optimization of maritime ship scheduling with heterogeneous fleet and draft limits
HU Pinghua;WU Lubin;YUAN Meng;CAO Changxin;YANG Kai;YUN Xinyu;ZHANG Zhenzhen;[Background] Rapid growth in global maritime trade has resulted in larger and increasingly heterogeneous shipping fleets. Ship scheduling encounters multiple real-world constraints such as time windows and draft limits, substantially increasing scheduling complexity. These constraints represent critical bottlenecks to improving the operational efficiency and economic performance of shipping companies. [Objective] This study investigated a maritime pickup and delivery ship-scheduling problem that considered heterogeneous fleets and draft limits. The objective was to maximize total profit while satisfying constraints, including time windows, vessel capacity, and port draft limits.[Method] This problem was formulated as the Heterogeneous Fleet Pickup and Delivery Problem with Time Windows and Draft Limits. A mixed-integer linear programming model was first developed, along with a hybrid metaheuristic algorithm that integrates a large neighborhood search for solution perturbation and a variable neighborhood descent for local search. Tailored destroy-and-repair and local search operators were introduced to improve the performance of the algorithm. [Data]Based on real-world data from COSCO SHIPPING Specialized Carriers Co., Ltd., two benchmark instance sets of varying scale were constructed. [Result] In a large-scale realistic scenario involving200 cargoes and 25 vessels, the proposed algorithm achieved an average profit improvement of 76.41% over CPLEX, with a maximum improvement of 941.32% in a single instance, demonstrating excellent performance.
Task-driven multi-dimensional feature dynamic selection of ship trajectory
ZHANG Xinwei;LIU Wen;[Background] The accuracy improvement of ship trajectory classification and prediction tasks relies heavily on the selection of trajectory features. The flexible and dynamic selection of features suitable for tasks has become an important issue in the current field. [Objective] To enhance the adaptability of features and tasks and tackle limitations, such as the insufficient generalization ability caused by traditional methods relying on artificial rules, the high complexity caused by feature redundancy, and cross-task weight solidification. [Method] A task-driven dynamic feature-selection framework is proposed. First, a multidimensional feature system that decouples features from the attribute(motion/behavior/spatiotemporal/statistical features) and structural dimensions(temporal/nontemporal features) to form a complete set of features was constructed. Subsequently, a twostage feature selection mechanism was designed. First, the common feature subset was screened using a multicriteria filtering method, and then, task-specific fine screening was performed by combining the weighted fusion of the importance of the lightweight model. Finally, a dynamic feature network embedded with a learnable mask module was developed. The weights were initialized with the importance of the model features and adaptively updated with the task loss gradient. [Result] Verification based on the Baltic Sea ship trajectory dataset showed that the two-stage screening feature subset improved the baseline model accuracy by 2.62 % ～4.07 % in the classification task and reduced the haversine error by 0.11～0.74 km in the prediction task. The average accuracy of the proposed dynamic feature network in the classification task was 92.35 %, and the haversine error of the prediction task was reduced to 1.59 km. The average accuracy of the dynamic feature network proposed in the experiment on the newly added Chinese offshore water dataset was 96.1 % in the classification task, and the haversine error of the prediction task was reduced to 1.48 km. [Application] The framework provides a lightweight and high-precision feature optimization path for ship trajectory analysis that effectively promotes the intelligent development of maritime supervision.
Joint optimization of passenger-freight capacity allocation and pricing in metro systems with flexible marshalling
WANG Ruoshui;GUO Xin;SUN Huijun;[Background] Under fixed marshalling, freight capacity in metro mixed passenger-freight transport is constrained by passenger flow levels and total train capacity, making it difficult to accommodate the continuously growing demands of the urban logistics market. [Objective] This study introduces flexible train marshalling with dedicated freight cars in a multi-line metro network to enhance the profitability and freight service capability of the mixed passenger-freight metro system while maintaining passenger service levels. [Method] Considering the spatiotemporal heterogeneity and uncertainty of passenger and freight demand, passenger demand is characterized using data-driven stochastic scenarios, while freight demand is represented via a price-elastic demand function and hard time windows. A joint optimization model for capacity allocation and pricing is developed to maximize metro profit, which comprehensively determines passenger-freight capacity allocation,freight train configuration, freight pricing, and train-level order allocation. [Data] A case study is conducted based on passenger smart card data from Beijing Metro Lines 6 and 9 and simulated freight data. [Result] The results show that flexible train marshalling can overcome the freight capacity limitations imposed by fixed train marshalling and improve the adaptability of the mixed passenger-freight metro system to different logistics market demands. As the scale of logistics market demand increases, both metro profit and optimal pricing exhibit an increasing trend, and passenger flow characteristics exert a significant impact on the operation of the mixed passenger-freight system. [Application] These findings can provide a reference for the optimization of mixed passengerfreight operation organization, flexible marshalling configuration of freight trains, and capacity expansion decision-making in metro systems.
About Journal
Founded in 2003, Bimonthly
Competent Authority：Ministry of Education of PRC
Sponsor：Southwest Jiaotong University
Editor in Chief： LIU Xiaobo; HE Zhengbing
E-mail：jtt@swjtu.edu.cn
ISSN ：1672-4747
CN ：51-1652/U
Impact Factor: 3.245
Category Ranking: 7/169
Quartile: Q1
Indexed
CSTPCD
Scopus
RCCSE
JST
CSTJ
CNKI
COJ
Tracking the information about your manuscript
Communicate with the editorial office
Query manuscript payment status Peer ReviewCollecting, editing, reviewing and other affairs offices
Managing manuscripts
Managing author information and external review Expert Information Office WorkOnline Review
Online Communication with the Editorial Department