华南理工大学学报(自然科学版) ›› 2012, Vol. 40 ›› Issue (7): 41-45,56.

• 交通与运输工程 • 上一篇    下一篇

基于Ordered Probit 模型的交通事故受伤人数预测

宗芳 许洪国 张慧永   

  1. 吉林大学 交通学院,吉林 长春 130022
  • 收稿日期:2011-09-20 修回日期:2011-11-24 出版日期:2012-07-25 发布日期:2012-06-01
  • 通信作者: 许洪国(1955-) ,男,博士,教授,主要从事车辆安全与交通事故工程研究. E-mail: xhg335@163.com E-mail:zongfang@ jlu.edu.cn
  • 作者简介:宗芳(1979-) ,女,博士,副教授,主要从事运输系统规划与管理研究.
  • 基金资助:

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

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

Zong Fang  Xu Hong-guo  Zhang Hui-yong   

  1. College of Traffic,Jilin University,Changchun 130022,Jilin,China
  • Received:2011-09-20 Revised:2011-11-24 Online:2012-07-25 Published:2012-06-01
  • Contact: 许洪国(1955-) ,男,博士,教授,主要从事车辆安全与交通事故工程研究. E-mail: xhg335@163.com E-mail:zongfang@ jlu.edu.cn
  • About author:宗芳(1979-) ,女,博士,副教授,主要从事运输系统规划与管理研究.
  • Supported by:

    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.

关键词: 交通事故, Ordered Probit 模型, 受伤人数, 态势预测, 快速响应

Abstract:

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.

Key words: traffic accident, Ordered Probit model, number of injuries, severity forecast, quick response