关键词: 高速公路, 事故严重程度, 影响因素分析, 短时交通流, 随机参数Logit模型

Abstract:

The severity of freeway traffic accidents is influenced by a multitude of factors, among which short-term traffic flow characteristics prior to an accident play a particularly critical role. To systematically analyze the impact of short-term traffic flow conditions on accident severity, this study draws on historical traffic accident data, ETC gantry passage data, and meteorological data from three expressways in Guangdong Province—Nan Erhuan Expressway, Jingguang Expressway, and Western Coastal Expressway—for the years 2021–2022. A random parameter Logit model incorporating mean heterogeneity is constructed to explore the heterogeneous characteristics of accident - related factors. This research identifies 29 potential variables from four perspectives: road characteristics, environmental characteristics, traffic flow characteristics, and accident characteristics. Then, it models accident severity using a standard multinomial Logit model, a random parameter Logit model, and a random parameter Logit model with mean heterogeneity. By comparing the goodness of fit of the models using the pseudo R-squared, Akaike Information Criterion, and Bayesian Information Criterion, the results demonstrate that the random parameter Logit model with mean heterogeneity outperforms the others, capturing the heterogeneous characteristics of accident - related factors more accurately. Further evaluation of the impact of different factors on accident severity based on the average elasticity coefficients of the variables shows that, at a 99% confidence level, 22 parameter variables related to road, environmental, accident, and traffic flow characteristics significantly affect accident severity. Specifically, certain factors such as a six-lane design and increased visibility reduce accident severity, while others, including road rescue processing time, average speed of large vehicles, proportion of large vehicles, and speed difference between large and small vehicles, intensify accident severity when increased. The findings of this study provide a scientific basis for freeway accident prevention and management.

Key words: freeway, injury severity, factors analysis, short-term traffic flow, random parameters Logit model