Identifying critical links for delay spread mitigation in high-speed rail networks

Train delays are a common issue in high-speed rail (HSR) networks, significantly affecting overall operational efficiency. To evaluate how delays evolve among stations, researchers often model delay propagation as a network spreading process. A fundamental question then arises: which links in HSR networks are critical for controlling delay spread? In this paper, we address this question by investigating critical link identification methods. Specifically, we model delay spread as a Susceptible–Infected–Recovered process with heterogeneous spread rates, and develop a simple calibration algorithm that iteratively optimizes the spread rates to minimize the discrepancy between simulated and observed delay distributions. We argue that the optimized spread rate can be interpreted as a novel link centrality metric in the context of delay spreading. Through extensive experiments, we demonstrate that this metric generally outperforms purely topology-based metrics, such as betweenness and degree product, and its performance can be further enhanced when combined with traditional metrics. Our work provides new insights into the targeted control of delay propagation in HSR networks.