Research on the Stability Evaluation Method of Train Schedule Based on Delay Propagation Model

The setting of redundant time in the train schedule can improve the robustness of the railway operation system, reduce the delay caused by random disturbance, and make the existing delays dissipate more quickly. This paper aims to establish a data-based delay propagation model to analyze the stability of each node in the current train schedule, and use the result to auxiliary train schedule optimization decision. In the research, an integrated delay propagation model based on maximum algebra and ensemble learning is proposed, which describes the train delay propagation characteristics in combination with the train schedule and actual operation conditions. The maximum algebra block predicts the delay propagation based on the redundancy of the train schedule. The ensemble learning block uses the delay probability data obtained from actual operating records to predicts the joint delay time. The accuracy of the model prediction is verified on the actual operation data. Different levels of interference are designed as test case to analyze the stability of train schedule based on the model output.