Study on Predicting Station-To-Door Delivery Time in Railway Logistics Based on Dynamic Data

To address the lack of effective time-efficiency evaluation in the “station-to-door” segment of modern railway logistics, this study proposes a travel time prediction method based on dynamic traffic data. Using the AMap (Gaode) API, real-time traffic information on medium-duty truck routes is collected across multiple time intervals to capture temporal variations. A directed graph model is constructed by segmenting typical delivery paths and assigning time-dependent weights to each arc based on observed travel times. The Dijkstra algorithm is then applied to estimate the minimum travel time during specific time windows. Empirical validation is conducted using representative station-to-door routes, demonstrating the model’s ability to produce accurate lower-bound time estimates for last-mile delivery. This enables dispatchers to anticipate potential delays and plan resources more effectively. The proposed method provides a quantitative and scalable framework for railway logistics operators to improve the reliability and responsiveness of their delivery services, and lays a foundation for integrating real-time traffic forecasting into intelligent dispatch systems.