Applications of Innovative Accident Analysis Methods in Railways: A Review

  • Author(s) / Creator(s)
  • Accident analysis methods are used to determine the factors and their interrelationships that contributed to an accident. Various methods, including Fault Tree Analysis (FTA), Human Factors Analysis and Classification System (HFACS), and Systems Theoretic Accident Model and Processes (STAMP), were developed to model accident causation. However, the classical methods show weaknesses in accident modeling in sociotechnical systems that have complex dependencies of system components, and uncertainties in system behavior. To address the limitations, newer methods such as Bayesian networks (BNs), Petri nets (PNs), text mining (TM), and machine learning (ML) were used alone or with other techniques to model accidents. This article presents a review of publications in six databases (ScienceDirect, Scopus, Web of Science, SpringerLink, Google scholar, and IEEE Xplore) of these accident analysis methods for the railway industry. The publications are categorized into network-based and artificial intelligence (AI)-based accident analysis methods, and additional categories, such as the type of algorithms and techniques, data sources, and tools applied. The findings show that Bayesian networks and text mining are the most widely used network-based and AI-based methods for analyzing railway accidents.

  • Date created
  • Subjects / Keywords
  • Type of Item
  • DOI
  • License
    Attribution-NonCommercial 4.0 International