基于Nested Logit与Random Parameters Logit模型的摩托车事故伤害估计与对比  

Comparative Study on Motorcycle Crash Injury Severity Estimation Based on Nested Logit and Random Parameters Logit Models

作　　者：温惠英[1] 汤左淦 WEN Huiying;TANG Zuogan(School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510640,Guandong,China)

机构地区：[1]华南理工大学土木与交通学院,广东广州510640

出　　处：《华南理工大学学报（自然科学版）》2018年第11期83-91,共9页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(51578247;51378222)~~

摘　　要：为了给我国的摩托车事故伤害分析提供指导与依据,通过抽取美国印第安纳州2013—2015年的1947起摩托车单车事故,分别建立Nested Logit与Random Parameters Logit模型,分析摩托车事故伤害程度的影响因素,模型参数分别采用全信息最大似然估计法与蒙特卡洛模拟方法进行估计.两个模型的估计结果均表明:女性、年龄、使用头盔、酒驾、甩出车外、超速、冲出道路、载人、车龄>10年、路面潮湿、曲线坡度、交叉口、限速>80 km/h、4月份、7月份、夜间无灯光、郊区、事故碰撞物(防护栏、树、墙、路缘、电线杆、涵洞)等与摩托车事故伤害程度显著相关.通过对比Nested Logit与Random Parameters Logit模型的AIC与BIC准则值,发现Random Parameters Logit模型对事故数据的拟合优度更高,能够得到更好的参数估计结果.In order to provide reference and guidance for the analysis of crash injury severity in domestic motorcycle crashes, Nested Logit and Random Parameters Logit models are developed according to the motorcycle crashes data with a total of 1947 single-vehicle motorcycle crashes occurring from 2013 to 2015 in Indian in America, and the model parameters are estimated by using the full information maximum likelihood(FIML) method and the Monte Carlo method, respectively. The estimation results obtained by the two models show that female, age, helmet use, alcohol use, ejection, speeding, run off of road, passenger on vehicle, vehicle age over 10 years old, wet pavement, horizontal curve with slope, intersection, speed limits over 80 km/h, months of April and July, nights without street lights, rural area and fixed collision objections(guardrail, tree, wall, curb, pole and culvert) are signi-ficantly related to motorcycle crash injury severity. By comparing the AIC and BIC values of Nested Logit and Random Parameters Logit models, it is found that the overall fit of Random Parameters Logit model is better than that of Nested Logit model.

关 键 词：交通安全 摩托车事故 事故伤害程度 RANDOM PARAMETERS LOGIT模型 Nested LOGIT模型 

分 类 号：U491.1[交通运输工程—交通运输规划与管理]