Flexible Land-Use Control Planning for Bus Terminals at Mega Railway Hubs: An AI-Enabled Approach

Abstract

In recent years, bus terminals serving mega railway hubs in China’s major cities and megacities have faced a prominent contradiction between declining passenger demand and redundant land provision. This study examines four benchmark railway hubs in China and abroad and shows that three land-use optimization strategies-distributed layout, functional sharing, and three-dimensional integration-can substantially reduce terminal land requirements while maintaining essential feeder-transport functions, thereby providing an alternative to the conventional approach of sizing land solely from peak passenger demand. On this basis, a flexible land-use control model is developed by dividing terminal land into rigid service space and flexible adjustment space. The former is sized against off-peak passenger demand to meet basic operational needs, while the latter accommodates peak-period reserve capacity, mixed-use functions, and long-term development reserves. The study also explores an implementation pathway supported by deep learning and video-image recognition technologies and establishes an operational mechanism that integrates AI-enabled multi-source data acquisition, passenger-flow prediction, dynamic spatial regulation, and closed-loop feedback. This framework shifts bus-terminal land-use control from a static and rigid model toward dynamic and adaptive spatial governance. The findings provide a conceptual and methodological reference for intensive land use and adaptive planning of bus terminals at mega railway hubs.