A Multilayer Digital Twin Framework for Real‐Time Optimal Response to Collision and Obstacle‐Intrusion Risks in Railway Rolling Stock
Journal of Advanced Transportation · 2026-01 · 2026 (1)
DOI: 10.1155/atr/4691291 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Advanced Transportation, online erschienen am 14.11.2026. Verfasst von Hyung Suk Mun und 1 weiteren Autor:innen. Die Arbeit nennt 55 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Korea Agency for Infrastructure Technology Advancement, Ministry of Land, Infrastructure and Transport, Korea Railroad Research Institute.
Abstract (Originalsprache)
Train protection that responds to a track obstacle with a single threshold‐triggered emergency‐brake application, as in automatic train protection braking‐curve supervision, is robust but abrupt. This paper proposes a five‐layer digital twin framework (physical, sensing, virtual twin, cognitive decision and service layers) that adds a graded response ahead of an unchanged protection layer. An extended Kalman filter fuses radar, LiDAR, camera and odometry data into a twin state with explicit covariance; a composite risk index combining time to collision, stopping‐margin ratio and sensor confidence enables a model predictive control scheme that issues a jerk‐limited service‐brake demand subject to the protection envelope, a terminal stopping constraint and an emergency‐braking backup policy. The framework is evaluated in a cosimulation of an eight‐car urban electric multiple unit with brake build‐up delay, brake‐effectiveness dispersion and adhesion‐limited deceleration, over 1800 paired Monte Carlo trials in three scenario families at 40, 60 and 80 km/h. Collision outcomes were governed by detection range and available deceleration, not by the controller: The baseline and the proposed system each avoided contact in 58.7% of trials, against a kinematic ceiling of 59.0%. Where the detection range permitted a graded response (600 trials, all on dry rail under clear‐weather sensing), the proposed system stopped the train without the emergency brake in 99.0% of cases, reducing mean peak deceleration from 1.52 to 0.99 m·s −2 and mean peak jerk from 2.72 to 0.34 m·s −3 ; the jerk reduction depends on the assumed emergency‐brake build‐up. An ablation showed that the risk index gated the optimisation without altering any outcome. The decision loop never exceeded the 50‐ms control period on general‐purpose computers (99.9th percentile at most 2.8 ms). The digital twin therefore does not extend the physical limits of collision avoidance; its benefit is an earlier, smoother response inside the protection envelope.
Förderung
- Korea Agency for Infrastructure Technology Advancement
- Ministry of Land, Infrastructure and Transport
- Korea Railroad Research Institute
Zitieren
APA
Mun, H. S., & Park, C. W. (2026). A Multilayer Digital Twin Framework for Real‐Time Optimal Response to Collision and Obstacle‐Intrusion Risks in Railway Rolling Stock. Journal of Advanced Transportation, 2026(1). https://doi.org/10.1155/atr/4691291
Vancouver
Mun HS, Park CW. A Multilayer Digital Twin Framework for Real‐Time Optimal Response to Collision and Obstacle‐Intrusion Risks in Railway Rolling Stock. Journal of Advanced Transportation. 2026;2026(1). doi:10.1155/atr/4691291
BibTeX
@article{Mun2026,
author = {Mun, Hyung Suk and Park, Chan Woo},
title = {A Multilayer Digital Twin Framework for Real‐Time Optimal Response to Collision and Obstacle‐Intrusion Risks in Railway Rolling Stock},
journal = {Journal of Advanced Transportation},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/atr/4691291}
}
EUVIMED