Abstract: With the increasing data volume of train on-board system, real-time performance has become the most critical factor to ensure the safety of train operation. Considering that standard Ethernet cannot meet the real-time requirement of existing train communication network (TCN), the time-sensitive network (TSN) technology for TCN is introduced. To solve the time-delay problem, an adaptive switch queue selection mechanism for traffic scheduling is proposed. Firstly, the topology model of TCN based on TSN and the traffic model are described. Then, the K shortest path routing algorithm based on load balancing provides the optimal routing for the scheduling process. Finally, the adaptive switch queue selection mechanism is introduced to solve the aggregation flow conflict problem effectively, queue resources are properly allocated, and the gate control list (GCL) of each frame in the queue is obtained. Experimental results show that compared with the traditional constraint model, the schedulability of the model with an adaptive switch queue selection mechanism increases by 33.0%, and the maximum end-to-end delay and network jitter decrease by 19.1% and 18.6% on average respectively. It can provide theoretical support and application reference for the real-time performance optimization of TCN based on TSN.