Developing a Model for the Risk of the Rail Vehicles Collision Using Bayesian Network

Document Type : Research Paper


1 Department of Civil Engineering, Payam e Noor University(PNU), P.O.BOX 19395-4697. Tehran, Iran

2 Iran university of science and Technology

3 Department of Statistics, Faculty of Mathematical Sciences and Computer, Allameh Tabataba'i University, Tehran, Iran.

4 PhD candidate for Highway and Transportation Engineering, Department of Civil Engineering, Payame Noor University, Tehran, Iran.


Risk analysis of the rail vehicles collision as one of the most important accidents with the rate of nearly 4 percent of the total accidents has always been of great interest especially for the railway safety decision makers. The reason might be revealed while considering the chain of causes and the scenarios of consequences. The superiority of modeling by Bayesian networks method for analyzing the risk of such accidents is that not only the conditional probability of each cause (hazard) as a variable is assessed but also with having any new evidence, the model can be updated subsequently and the consequences can be monitored. The methodology consists of two major categories of qualitative and quantitative parts while the concluded model is a mixture of the two methods. Main causes of a collision are grouped as environmental, signaling and human factors. Resulted model gives the highest probability to the shunting limit signal as a subdivision of the human errors. Evaluating the consequences resulted in the severity of “second degree” based on the Iranian railway accidents severity categorization guideline.


- Butler, P. A. (2001). Train collision system, Google Patents.
- Li, C., W Li and J Ning. (2018). Calculation of Ship Collision Risk Index Based on Adaptive Fuzzy Neural Network. 2018 3rd International Conference on Modelling, Simulation and Applied Mathematics (MSAM 2018), Atlantis Press.
- Li, S., B Cai, J Liu and J Wang. (2018). "Collision risk analysis based train collision early warning strategy." Accident Analysis & Prevention 112: 94-104.
- Lin, C.-Y. (2019). Probabilistic risk assessment of railroad train adjacent track accidents, University of Illinois at Urbana-Champaign.
- Liu, X. (2016). "Analysis of collision risk for freight trains in the United States." Transportation Research Record 2546(1): 121-128.
- Mehranfar, H., A Tadayon and M Sadeghi. (2019). Major Factor Affecting Freight Train Collision Severity At Road-Rail Level Crossing Using Multinomial Logit Model. 2019 5th International Conference on Transportation Information and Safety (ICTIS), IEEE.
- Nielsen, T. D. and F. V. Jensen (2009). Bayesian networks and decision graphs, Springer Science & Business Media.
- Peng, Z., Y Lu, A Miller and C Johnson. (2016). "Risk assessment of railway transportation systems using timed fault trees." Quality and Reliability Engineering International 32(1): 181-194.
- Turla, T.,X Liu and Z Zhang. (2019). "Analysis of freight train collision risk in the United States." Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit 233(8): 817-830.
- Visintin, C., N Golding and R Van Der Ree. (2018). "Managing the timing and speed of vehicles reduces wildlife-transport collision risk." Transportation research part D: transport and environment 59: 86-95.
- Xu, Q. and K. Xu (2018). "Risk assessment of rail haulage accidents in inclined tunnels with Bayesian network and bow-tie model." Curr. Sci 12: 2530-2538.
- Yao, S., H Zhu, Z Li and P Xu. (2019). "A study on the frontal oblique collision-induced derailment mechanism in subway vehicles." Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit: 0954409719852478.
- Zheng, Z., P Lu, D Pan. (2019). "Predicting Highway–Rail Grade Crossing Collision Risk by Neural Network Systems." Journal of Transportation Engineering, Part A: Systems 145(8): 04019033.
- Zhu, X., XZ Lu, QL Cheng and Y Li. (2019). "Simulation of the running attitude of a train after derailment." International Journal of Crashworthiness: 1-7.