A data-driven framework for multi-criteria risk prioritization in metro track geometry: Dynamic segmentation and causality-aware FMEA

Resource limitations and short maintenance windows in metro systems necessitate risk-based prioritization of track geometry maintenance to ensure safety and ride quality. While existing studies largely focus on ballasted tracks, comprehensive prioritization frameworks for metro slab track systems remain limited. This study proposes a data-driven and causality-aware Failure Mode and Effects Analysis (FMEA) framework that explicitly accounts for interdependencies among risk factors through integrated multi-criteria decision-making (MCDM). Track geometry data collected by the EM-50 inspection vehicle on the Tehran Metro network are segmented into homogeneous sections using Cumulative Difference Analysis (CDA). Severity, occurrence, and detection indices are quantified based on tolerance-based deviations, statistically derived severity thresholds, and alignment-dependent detection characteristics, respectively. Interdependencies among these risk factors are modeled using the DEMATEL method, with the direct-relation matrix constructed through expert judgment, multiple linear regression, and mutual information analysis. Risk prioritization is performed using TOPSIS, EDAS, MARCOS, and COCOSO, and aggregated via the Borda count. Sensitivity analysis based on Spearman’s rho and Kendall’s tau indicates strong rank consistency across methods (ρ, τ > 0.9), confirming result robustness. The framework effectively identifies high-risk segments and supports proactive, targeted maintenance planning for metro slab track systems.