考虑时间不稳定性的货车事故严重程度分析

Severity analysis of freight car accidents considering time instability

为提高货车交通安全,以英国STATS数据库中2014—2019年29213条事故数据为研究对象,将事故严重程度作为因变量,将人、车、路和环境等事故特征作为自变量,构建随机参数有序Logit模型,并采用对数似然比,检验测试6年内事故数据的时间稳定性。结果表明:事故数据存在严重的时间不稳定性,应将其划分为2014—2016、2017、2018和2019年4个年份分组分别建模;60岁以上小汽车驾驶员、货车转弯、货车变道等8个变量在4个模型中均显著影响事故严重程度,其他变量仅在特定年份事故模型中有显著影响;此外,60岁以上小汽车驾驶员、货车变道等9个变量在特定年份事故模型中对事故严重程度具有异质性影响。

In order to improve truck traffic safety,a total of 29213 accident data from 2014 to 2019 in the British STATS database were taken as the research object,the accident severity was taken as the dependent variable,and the crash characteristics related to people,vehicle,roadway and environment were considered as independent variables.The random parameter ordered logit model was constructed.Meanwhile,the log-likelihood ratio tests were used to test the time stability of the crash data within years period.The results show that there is serious time instability in the accident data,which should be divided into four groups:2014-2016,2017,2018 and 2019.this study finds 8 variables exhibit significant impacts on accident severity in all four crash models,including car drivers over 60 years old,truck turning,truck lane changing and so on.Other variables are only statistically significant in year-specific crash models.In addition,9 variables exhibit heterogeneous effects on accident severity in a specific year accident models,including car drivers over 60 years old,truck lane changing,etc.