Development of Traffic Accident Prediction Model Based on Traffic Node and Link Using XGBoost

Kim, Un-Sik; Kim, Young-Gyu; Ko, Joong Hoon

doi:10.11627/jksie.2022.45.2.020

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In January 2022, black ice led to an accident on the road towards Soheul-eup, Pocheon, on the Guri-Pocheon highway in Gyeonggi Province, in which approximately 40 vehicles were involved in a multi-car collision. This resulted in one fatality, three serious injuries, and 25 casualties [ 6 ]. In November 2021, black ice resulted in seven injuries on roads such as the Yeongdong highway in the Gangwon Province [ 7 ].…”

Section: Introductionmentioning

confidence: 99%

Predicting black ice-related accidents with probabilistic modeling using GIS-based Monte Carlo simulation

Hong,

Yun

2024

PLoS ONE

View full text Add to dashboard Cite

Black ice, a phenomenon that occurs abruptly owing to freezing rain, is difficult for drivers to identify because it mirrors the color of the road. Effectively managing the occurrence of unforeseen accidents caused by black ice requires predicting their probability using spatial, weather, and traffic factors and formulating appropriate countermeasures. Among these factors, weather and traffic exhibit the highest levels of uncertainty. To address these uncertainties, a study was conducted using a Monte Carlo simulation based on random values to predict the probability of black ice accidents at individual road points and analyze their trigger factors. We numerically modeled black ice accidents and visualized the simulation results in a geographical information system (GIS) by employing a sensitivity analysis, another feature of Monte Carlo simulations, to analyze the factors that trigger black ice accidents. The Monte Carlo simulation allowed us to map black ice accident occurrences at each road point on the GIS. The average black ice accident probability was found to be 0.0058, with a standard deviation of 0.001. Sensitivity analysis using Monte Carlo simulations identified wind speed, air temperature, and angle as significant triggers of black ice accidents, with sensitivities of 0.354, 0.270, and 0.203, respectively. We predicted the probability of black ice accidents per road section and analyzed the primary triggers of black ice accidents. The scientific contribution of this study lies in the development of a method beyond simple road temperature predictions for evaluating the risk of black ice occurrences and subsequent accidents. By employing Monte Carlo simulations, the probability of black ice accidents can be predicted more accurately through decoupling meteorological and traffic factors over time. The results can serve as a reference for government agencies, including road traffic authorities, to identify accident-prone spots and devise strategies focused on the primary triggers of black ice accidents.

show abstract

Section: Introductionmentioning

confidence: 99%