2026 Volume 30 Issue 1 Pages 95-102
This study presents a fundamental investigation aimed at optimizing track maintenance planning in JR East. Although maintenance plans are based primarily on inspection data, they also reflect local track conditions, on‑site environments, and the skill levels of maintenance personnel, making quantitative evaluation difficult. This paper focuses on developing risk assessment indicators for track condition, which form the basis of maintenance planning. Previous research showed that applying survival analysis to track geometry data and infrastructure specifications could help visualize the empirical knowledge maintained by engineers. In this study, we additionally incorporated track repair records and applied dynamic survival analysis, which accounts for time‑varying covariates in predicting survival probabilities. We then evaluated how accurately the model could predict survival rates over a two‑ to three‑year horizon. This paper summarizes the results of these analyses.
