Agent Modeling of Infrastructure Vulnerability with Temporal Centrality Patterns in Railway Transport Networks

Abstract

The resilience of railway transport networks is of paramount importance to modern socio-economic stability. Traditional methods for evaluating infrastructure vulnerability have largely relied on static topological metrics, which fail to capture the dynamic, time-varying nature of transportation systems. This paper presents a novel approach for predicting infrastructure vulnerability by integrating temporal centrality patterns with agent-based modeling. By conceptualizing the railway network as a dynamic temporal graph, the proposed framework captures the real-time operational states and the shifting importance of network components. Agent-based modeling is employed to simulate the heterogeneous behaviors of individual trains, dispatch centers, and passengers, providing a high-fidelity representation of system dynamics under varying stress conditions. The simulation outputs continuous temporal centrality metrics, specifically temporal betweenness and temporal closeness, which serve as predictive indicators of impending structural and operational bottlenecks. Through extensive simulation of a generalized regional railway network, this study demonstrates that temporal centrality patterns exhibit recognizable precursor anomalies prior to systemic failures. The predictive methodology significantly outperforms conventional static network analysis in identifying latent vulnerabilities. The findings offer critical insights for infrastructure managers, enabling proactive maintenance scheduling, optimized routing during disruptions, and enhanced overall network resilience.