平纵组合路段事故严重程度致因辨识模型

为解析平纵组合路段事故严重程度致因及影响机制，在系统选取影响事故严重程度潜在变量的基础上，利用有序Logit模型获取影响事故严重程度的显著自变量，并运用偏比例优势模型修正变量参数，从而构建平纵组合路段事故严重程度致因辨识的两阶段模型 (TSM)，以云南省元(谋)—双(柏)公路为例进行分析.结果表明：提出的TSM模型比有序概率模型拟合度更优，更适用于研究该问题；涉及车辆数、平曲线曲率及竖曲线曲率等7个变量对一般及以上事故具有正效应，驾驶人性别及接入口等3个变量对一般及以上事故具有负效应；就影响程度来看，接入口最大(边际效应为14.466%)，驾驶人性别次之(10.581%)，竖曲线长度最小(0.114%).

Severity Factors Analysis Model of Traffic Accident on Combined Horizontal and Vertical Alignments

To analyze the severity factors and influence mechanism of traffic accidents on combined horizontal and vertical alignments, the potential variables influencing accident severity were selected first. And an ordered Logit model was used to obtain significant independent variables influencing accident severity. A partial proportional odds model was used to modify the parameters of variables. The two- stage model (TSM) was then proposed for severity factors analysis of accidents on combined horizontal and vertical alignments. A verification analysis was conducted by taking Yuanmou- Shuangbai secondary road in Yunnan province as a case area. The results show that the TSM has a higher fitting degree than ordered probability models and is more suitable for the study. In particular, seven factors including the number of vehicles, curvature of horizontal curves, curvature of vertical curves have positive influences on general and above accidents, while the other three factors including driver's gender and access segment have negative influences. In terms of influence degree, the impact of the access segment is the largest (marginal effect: 14.466% ), followed by the drivers' gender (10.581% ), and the length of vertical curves is the smallest.