Keywords
artificial intelligence
causal relationships
system analysis
technical causes
automation
How to Cite
Abstract
The article examines the features of forensic railway transport examination as a tool for establishing the mechanism of railway accidents and determining their direct technical causes in the context of the development of modern technologies, particularly artificial intelligence algorithms. The research problem lies in the complexity of establishing causal relationships in railway accidents caused by the complex interaction of technical, operational, and organizational factors. The purpose
of the study is to provide a comprehensive analysis of the specifics of causality determination in forensic railway transport examination, to identify the role of modern analytical technologies in this process, and to substantiate the limits of their application. The methodological basis of the study is based on general scientific and special research methods, including the system-structural method, comparative analysis, and methods of analysis and generalization of expert practice. This approach made it possible to consider the mechanism of a railway accident as a dynamic system of interaction between technical, operational, and organizational factors and to determine their role in the formation of the causal chain of the event. The scientific novelty of the study lies in substantiating a systemic approach to establishing causal relationships in railway accidents, as well as in defining the role of artificial intelligence algorithms as an auxiliary analytical tool. The conclusions emphasize that full automation of determining the direct technical cause of a railway accident is currently impossible, and the most promising approach is a hybrid “expert–artificial intelligence” model.
References
Islam M. et al. (2022). Data-driven modeling of rail damage using machine learning techniques. Proceedings of the 40th International Conference on Nondestructive Evaluation. URL: https://pubs.acs.org/doi/abs/10.1021/acsanm.2c03564(accessed: 10.01.2026) [in English].
Keško A., Puška H. (2024). Artificial intelligence in predictive maintenance and fault detection on railway tracks. Advanced Technologies, Systems and Applications: Conference Proceedings, 128–134. URL: https://www.atct.ba/conference-proceedings/2024/18-artif icial-intelligence-in-predictive-maintenance-and-fault-detection-on-railway-tracks.pdf (accessed: 11.01.2026) [in English].
Liao Y., Han L., Wang H., Zhang H. (2022). Prediction models for railway track geometry degradation using machine learning: a review. Sensors, 22(19), Article 7275. DOI: https://doi.org/10.3390/s22197275 [in English].
Ahmed M., Palese J.W., Zarembski A. (2023). Predicting Track Geometry Using Machine-Learning Methods: research report. Newark: University of Delaware. 75 p. URL: https://rosap.ntl.bts.gov/view/dot/80897/dot_80897_DS1.pdf(accessed: 20.01.2026) [in English].
Hume P.C., Gräbe H., Markou G. (2024). Using machine learning techniques as track geometry predictors for railway track. Civil Engineering, 32(9), 27–32 [in English].
Batih A.V. (2024). Perspektyvy u obiednanni ekspertnoho ta shtuchnoho intelektu pry vykonanni zaliznychno-transportnykh ekspertyz [Prospects for combining expert and artificial intelligence in conducting railway transport examinations]. Sudovo-ekspertna diialnist: problemy, stratehii ta innovatsii: materialy mizhnarodnoi naukovo-praktychnoi konferentsii, Lviv, Chisinau, Kyiv, Odesa, November 28, 2024, 74–76. URL: https://hdl.handle.net/11300/31202 (accessed: 01.03.2026) [in Ukrainian].
Batih A.V. (2024). Zastosuvannia metodiv mashynnoho navchannia v zadachakh sudovoi zaliznychno-transportnoi ekspertyzy [Application of machine learning methods in the tasks of forensic railway transport examination]. Vplyv innovatsii na rozvytok sudovoi ekspertyzy: vid tradytsiinykh metodiv do tsyfrovoi transformatsii: materialy Vseukrainskoi naukovo-praktychnoi konferentsii, Lviv, April 26, 2024. Lviv: LNDISE Miu Ukrainy. 219 p. [in Ukrainian].