Search

The majority of Train Accidents (TA) in Indonesia from 2015 to 2020 were caused by infrastructure factors, namely railway tracks, bridges, and signals. To mitigate these TAs, infrastructure maintenance is required, prioritizing locations with a high risk of TA. The prioritization of these locations can be accomplished through a model that estimates TA based on infrastructure risk factors. Locations with the highest TA estimates will be prioritized for infrastructure maintenance. This model illustrates the associative relationship between KKA as the dependent variable and exposure (train frequency and track length) and infrastructure risk factors (railway tracks, bridges, and signals) as independent variables. The model is constructed using the Generalized Linear Model (GLM), considering Poisson Regression (PR), Negative Binomial (NB), Zero Inflated Poisson (ZIP), and Zero Inflated Negative Binomial (ZINB) models. TA data from the Operational Areas (OA) of Jakarta, Bandung, and Cirebon during 2015-2020 were used to build the model, with the model entity being the segment between two train stations. The selection of the best model is based on tests of dispersion value, goodness-of-fit test, and Vuong test. Modeling results indicate that the NB model is the most suitable for illustrating the associative relationship between TA and infrastructure factors in the Indonesian Railway. The variables are train frequency (trains/day), track length (km), train speed (km/h), length of curves with a radius of 500 m to ≤ 1000 m (km), number of vulnerable areas (points), length of electricity network (km), and track type (single or double). [ABSTRACT FROM AUTHOR]