A risk assessment model for traffic crashes problem using fuzzy logic: a case study of Zonguldak, Turkey

ABSTRACT

In this Fuzzy Inference System based study, a model is developed which can present a real-time risk assessment model to the drivers. The model includes parameters car speed, accident frequency, weather conditions, tire tread depth and fatigue. In order to evaluate the performance of the model, the road between Zonguldak and Düzce is chosen, which has a high accident incidence and can cause anxiety on the drivers. The statistical results of the real and artificial data sets that were created to demonstrate the model performance were found as follows; Mean Square Error is 5.07551 for real and 6.43029 for artificial data set, Mean Absolute Error is 1.8 and 1.69864, Correlation Coefficient is 95.594% and 97.412%, and Determination Coefficient is 84.98% and 93.6% respectively. Investigation of the actual accidents reveals that the accident risk ratios in these accidents were between 60% and 80% according to the determined parameters.

KEYWORDS:

• Fuzzy Inference System (FIS)
• drivers’ decisions
• quantitative risk assessment
• decision support system
• road crashes
• road safety

Disclosure statement

All the authors declare that there are no conflicts of interests regarding the publication of this article.