A Two-Step Method for the Operation Rescheduling Problem at the Railway Marshalling Station

Abstract

Current research on the operation planning problem at the railway marshalling station are mainly dependent on the assumption of correct and unaltered input data. However, due to force majeure disturbances such as train delays, engine breakdown, and so on, the operation plan may easily fail. To increase the robustness of the operation plan, we propose a two-step method for solving the operation rescheduling problem at the marshalling station, the first step is to create a static model in order to maximise the efficiency of the operation plan, and the second step is to reschedule the static plans and unexecuted plan caused by unexpected disturbances. A lexicographic multi-objective model is built to reduce disturbance perturbation while maximising station benefits. An iteration optimisation approach with initial solutions is designed to solve the whole model iteratively, and each sub-model is solved by the heuristic backtracking algorithm with constraint propagation (CPr) mechanism called CPr-HBT. Some real-life cases from Shijiazhuang marshalling station of the China Railway show that the proposed two-step method increases the robustness of the operation plans effectively, rescheduling the failed operation plans takes 60 s at most, and CPr-HBT is more efficient than the other algorithms without CPr or heuristics.