干旱区生态环境知识资源中心

Traffic accidents prediction models may help for understanding accidents' causes and predict their occurrence under certain circumstances. This study aimed at developing an accident prediction model for a highway in Egypt using the fuzzy logic approach. Then, analyzing the sensitivity of different accident reduction policies to help decision makers to implement the proper improvements. The model utilized actual accidents data on Aswan western desert road with a survey data for road width, pavement conditions, average hourly traffic volume, speed, the number of access points to the highway and traffic signs conditions. The developed model predicts the approximate number of accidents that can happen per year and the performed sensitivity analysis suggests the proper steps and precautions decrease accidents. Overall results revealed that predictions using the proposed fuzzy logic system produce an accurate and stable traffic accident predictions with a goodness-of-fit determination coefficient of more than 87%. Consequently, a short term and long term accident reduction plans are recommended that can reduce accident numbers by 15.2% and 42.3% respectively.