中文摘要：

为明晰公路隧道交通事故严重程度的影响因素,从134起事故统计数据中选出13个事故严重程度的潜在影响因素,分别采用有序Logit模型和广义有序Logit模型,建立交通事故严重程度预测模型。分析这13个因素对交通事故严重程度的影响程度,并对比2个模型的预测效果。结果表明：事故发生日期、发生时间、是否超速、天气和大型车比例5个自变量与事故严重程度显著相关。从广义有序Logit模型来看,事故发生时间不满足比例优势假设;广义有序Logit模型可以放宽部分自变量的比例优势假设,能给出更好的预测结果。

英文摘要：

To clarify the influence factors on traffic accident severity in highway tunnels,13 potential factors influencing traffic accident severity were selected from consideration of data on 134 cases of traffic accidents occurred in four highway tunnels. An ordered Logit model and a generalized ordered Logit model were respectively used to build models for predicting traffic accident severity. The influence degrees of different factors on traffic accident severity were analyzed. Prediction effects of the both models were also analyzed and compared. The results show that date of accident,time of accident,speeding or not,weather and percentage of large vehicles are significantly related to the accident severity,that the time of accident violates the proportional odds assumption in the generalized ordered Logit model,and that the generalized ordered Logit model can relax the proportional odds assumption for some independent variables,and can make a corrector prediction.