Assessing operational impacts of large-scale disruptions in urban rail transit: An improved multilayer interdependent network cascading failure model by data calibration

Large-scale disruptions in urban rail transit can severely disrupt operations, causing significant economic losses, adverse social impacts, and even casualties. Existing impact analysis methods often depend on assumptions or are constrained by small data, which impede accurate inference of the outcomes of large-scale disruptions. In this study, different systems within the urban rail transit network are modeled as a multilayer interdependent network. Based on historical failure data, the impact of major emergency events is inferred using an improved node-edge joint cascading failure model. The results show that, after calibrating model parameters with historical failure data, the proposed model can accurately estimate the consequences of large-scale disruptions, while other benchmark models often fail to produce accurate results. We also find a negative correlation between train capacity and the severity of event impact. These findings provide valuable insights for emergency management in urban rail transit. Additionally, the inputs and outputs of the proposed model can enhance historical accident data, offering a new tool for inferring accident impacts via machine learning methods.