<p>This study investigates the passengers to urban rail transit within comprehensive transportation hub, from both inter-city modes and urban transport. A model that leverages continuous short-term passenger flow prediction results for passenger flow simulation and management policy optimization is proposed to rationally allocate facility resources on both weekday peak and off-peak hours, aiming to overcome the limitations of traditional passenger flow simulations, which typically focus on peak hours or specific periods. The model consists of three components, namely short-term passenger flow prediction, AnyLogic modeling and simulation, and multi-dimensional index evaluation and optimization. Firstly, multiple linear regression model, long short-term memory networks, and the model of catchment area based on lattice distribution are selected, combined with the field passenger flow surveys, to capture the passenger flow via high-speed rail, buses, walking, taxies, and shared bikes during peak periods. Secondly, a simulation software package, AnyLogic, is introduced to establish a model for the passenger interchanges within the hub. Indicators such as transfer time, area passenger flow density, and space utilization rate are obtained to analyze the transfer bottlenecks. Finally, the optimization strategies for the comprehensive transportation hubs are proposed by taking Xi’an North Railway Station, China as an empirical example. The results demonstrate that the average passenger transfer time is reduced by 48.75%, and the passenger flow density in the bottleneck area decreases by 15.16% after the optimization. Findings of this study may provide technical support and guidance for further improvement of multi-model transportation in urban comprehensive hubs.</p>