基于Ordered Probit模型的交通事故受伤人数预测  

Forecast of Injury Number Due to Traffic Accident Based on Ordered Probit Model

作　　者：宗芳[1] 许洪国[1] 张慧永[1] 

机构地区：[1]吉林大学交通学院,吉林长春130022

出　　处：《华南理工大学学报（自然科学版）》2012年第7期41-45,56,共6页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家"863"计划项目(2009AA11Z201);国家自然科学基金资助项目(50908099;50878095)

摘　　要：建立了交通事故受伤人数预测的Ordered Probit模型,应用极大似然方法进行了模型标定.应用模型分析了受伤人数的影响因素,计算了各影响因素的边际贡献,并进行了受伤人数的预测.结果表明,对受伤人数影响较大的因素有路表是否干燥、是否有大中型车辆、是否在交叉口.所建模型在进行事故受伤人数预测的同时,也可用于分析各因素对受伤人数的影响方向和影响程度.文中结果可为交通管理部门迅速准确地判断事故态势并做出快速响应提供决策支持.In this paper, first, a forecast model of injury number due to traffic accident is established based on the Ordered Probit method and is calibrated via the maximum likelihood estimation. Then, the factors affecting the number of injuries are analyzed and the marginal contribution of each factor to the number of injuries is calculated. Moreover, a forecast of injury number is carried out with the proposed model. The results indicate that the major factors impacting the number of injuries are whether the surface of the road is dry, a bus or truck involved or not, and whether it happened at an intersection. The proposed model can be used to forecast the number of injuries in an accident and analyze the influence orientation and severity of each factor on the independent variable. The study provides decision-making supports for traffic management departments in terms of accident severity determination and fast accident response.

关 键 词：交通事故 ORDERED PROBIT模型 受伤人数 态势预测 快速响应 

分 类 号：U491.1[交通运输工程—交通运输规划与管理]