This paper introduces an online optimal controller for train operation that considers dynamic train information. The optimal control problem is solved using a shrinking horizon model predictive framework, enabling real-time integration of operation information, such as temporary speed restrictions and real-time train interactions. The original nonconvex optimization problem is transformed into a much easier solution in each prediction range, while the transformed convex optimization problem does not lose the optimality of its solution. A complete train speed trajectory can be generated online by continuously resolving the optimal control problem at each sampling step. If the optimization problem becomes infeasible due to prediction errors, infeasible journey time, and unreasonable braking requirements, the objective function or constraint can be adjusted to ensure the proper functioning of the controller. The proposed method’s effectiveness and robustness are validated through numerical simulations using real-world railway data.
Li et al. (Fri,) studied this question.