A Rescheduling Approach for Freight Railway considering Equity and Efficiency by an Integrated Genetic Algorithm

Abstract

Since unexpected event occurrences are inevitable, an efficient and effective rescheduling approach is critical in freight railway day-to-day operations. Represented by the minimal-delay objectives, the most commonly used efficiency-oriented approach ignores the role of train priority and poses equity problems in rescheduling. For equity, different train priority also reflects the preference in deciding train orders; i.e., the high-priority train is likely to be favorable. However, a conflict may be laid between reducing delays and emphasizing train priority. Hence, it is critical to decide the criteria in train order for freight railway, especially with heterogeneous priority in a competitive resource. To make a tradeoff between equity and efficiency, this paper makes train priority evolutionary and proposed the dynamic train priority considering delay time and static priority. We formulate an optimization model based on the rescheduling strategies such as retime, reorder, and retrack in a complex railway network containing single-track, double-track, and quadruple-track sections. An integrated two-dimension genetic algorithm (ITGA) approach is developed to reobtain an optimized timetable of sufficient quality while meeting the time requirements for real-time rescheduling. In the experiment, the effectiveness of the ITGA approach was employed in a simple case and a real-world case in the Netherlands freight railway. The result shows that there is a synergy between delay time and train priority, where the threshold to upgrade the evolutionary train priority plays an important role. The proposed approach is compared with the benchmark solution first-in-first-out (FIFO) approach in a real-world case to verify the performance and efficiency. Our work extends the rescheduling approach considering both equity and efficiency and provides auxiliary operation support for the dispatcher’s operation rescheduling.