英文摘要 |
To explore the effects of freeway traffic accident duration on freeway traffic, the study combined both freeway accident and duration data sets from the Taiwan national highway police and freeway administration sources. We selected 12,894 accident records for year 2019, and established Generalized Log-Gamma (GLG) survival models with constant parameters, and a Generalized Log-Gamma survival model with probability distribution parameterization (P-GLG) to analyze freeway traffic accident duration. Unlike similar GLG survival models, the advantage of the P-GLG model is its location parameter, scale parameter and shape parameter of underlying distribution, which is formulated with covariates. The results show the superiority of the P-GLG model to other GLG models. The variables considered in the established models indicate high averages of traffic accident duration for severe crashes and those involving large freight vehicles. The situation is worse for crashes located at a side slope, rollover crashes and fire/smoke traffic accident involved. The referred risk factors have high accident duration averages and cause high variance, which are difficult to determine the accident clearance duration. Additionally, the estimates of the accident duration for year 2019 point out the averages of A1 level crash is at least 42 minutes, the A2 level crash is 30 minutes, and the A3 level is within 15 minutes. The average of involving heavy freight vehicles is more than 20 minutes. The study contributes to determine the national highway traffic accident duration, and provide precautions for national highway traffic freeway accident. |