A Model Predictive Control Approach for Virtual-Coupling Train Operations Using Mixed-Integer Linear Programming

Virtual-coupling (VC) train operation is seen as an emerging technology that can effectively increase the capacity of railway transportation. Based on advanced train-to-train communication, VC allows trains to travel minimal distances without physical connections. This paper proposes a model predictive control approach based on Mixed-Integer Linear Programming (MILP) for virtually coupled trains, the nonlinear characteristics, and space-depending variables such as route gradients can be modeled based on time in the MILP model, leading to more flexibility in modeling for complex engineering scenarios. Result shows the model predictive control approach based on MILP needs about 18% solving time compared with Nonlinear Model Predictive Control with the equivalent prediction horizon. Our simulation results also demonstrate the proposed method can control the leader train to follow the reference speed and the follower train to keep an acceptable interval during the journey, which means the proposed method is effective for virtual coupled trains.