作　　者： ; ; (林永杰）;

机构地区： 长安大学公路学院

出　　处： 《中国安全科学学报》 2016年第3期86-90,共5页

摘　　要： 为了定量分析高速公路交通事故的主要影响因素,以国内3条高速公路227起交通事故为基础,根据人、车、路和环境交通4要素,提出运用有序Logit和多项Logit模型分别建立事故伤害程度预测模型。通过显著性水平为5%的假设检验,从15个备选变量中分别确定2类Logit模型的输入自变量。显著性分析表明:道路线形、路面通行条件、事故发生时间和能见度4个自变量与交通事故严重程度均显著相关。事故伤害程度预测模型对比结果显示:有序Logit模型仅对有财产损失的事故估计准确率为72.3%,多项Logit模型对死亡事故估计的准确率高达95%,总体上多项式模型的准确率比有序模型高20%。 227 field cases of traffic accidents on three freeways in China are selected to quantitatively analyze factors affecting traffic accidents. Based on 4 components of transportation system that are human, vehicles, roads and environment, the ordered Logit model and the multinomial Logit model are presented to model traffic accident severity estimation, respectively. The model inputs are decided from15 independent variables in terms of statistical hypothesis testing of 5% significance level. The statistical results show that accident severity is significantly related to road alignment, road surface condition,accident occurrence time, and visibility. In tests that used field data, the results show that the estimation accuracy of the ordered Logit model for property damage only accident occurrence is approximate 72.3%,and one of the multinomial Logit model for fatal accident occurrence is up to 95%. In addition, the average accuracy of the multinomial Logit model is 20% better than that of other one.

关 键 词： 高速公路 交通事故 严重程度 有序 模型 多项 模型

领　　域： [环境科学与工程]