Development of Train Operation Risk Assessment System Based on a Data-Driven Bayesian Network
- 26 August 2026
- journal article
- research article
- Published by SAGE Publications in Transportation Research Record: Journal of the Transportation Research Board
Abstract
Risks of severe train accidents vary along a railway corridor by track sections with different infrastructure, rolling stock, and operational characteristics because of the complicated nature of train driving environment. Therefore, this research develops a Train Operational Risk Assessment System (TORAS) that assists train drivers to recognize the level of risk for upcoming track sections to enhance their situational awareness. This research proposes a five-phase approach that integrates the Human Factors Analysis and Classification System (HFACS) with Bayesian networks (BN) to identify potential risk events through historical accident data and selected environmental and operational factors that may contribute to the occurrence of risk events. A BN constructed using historical Automatic Train Protection (ATP) data quantifies the relationships between various risk factors while calculating the probabilities of risk events. The results show that the new framework can predict track-section-specific train driving risks by using train operational and route information. TORAS can enhance train driver’s situational awareness by informing them of the risk levels of different track sections in advance, which enables proper caution and proactive preparation, improving the overall safety of train operations by ensuring that drivers are well-informed and able to respond effectively to varying risk conditions.Keywords
Funding Information
- National Science and Technology Council (113-2221-E-A49-037-MY2)
- Ministry of Education (Yushan Young Scholar Program)
This publication has 53 references indexed in Scilit:
- Evaluating train drivers’ performance to inform the development of automatic train operationApplied Ergonomics, 2025
- High-speed train drivers’ operation performance: Key factors, models, and management implicationsInternational Journal of Industrial Ergonomics, 2023
- Score-based Bayesian network structure learning algorithms for modeling radioisotope levels in nuclear power plant reactorsChemometrics and Intelligent Laboratory Systems, 2023
- Quantitative risk assessment of railway intrusions with text mining and fuzzy Rule-Based Bow-Tie modelAdvanced Engineering Informatics, 2022
- Dynamic Bayesian networks for temporal prediction of chemical radioisotope levels in nuclear power plant reactorsChemometrics and Intelligent Laboratory Systems, 2021
- The impact of personality on driving safety among Chinese high-speed railway driversAccident Analysis & Prevention, 2016
- Overview on Bayesian networks applications for dependability, risk analysis and maintenance areasEngineering Applications of Artificial Intelligence, 2012
- Supporting skill-, rule-, and knowledge-based behaviour through an ecological interface: An industry-scale applicationInternational Journal of Industrial Ergonomics, 2009
- Application of a human error framework to conduct train accident/incident investigationsAccident Analysis & Prevention, 2006
- Confidential Incident Reporting on the UK Railways: The ‘CIRAS’ SystemCognition, Technology & Work, 2000