Journal of Transportation Systems Engineering and Information Technology ›› 2018, Vol. 18 ›› Issue (4): 83-87.

• Intelligent Transportation System and Information Technology • Previous Articles     Next Articles

A Support Vector Machine Approach on Real-time Hazardous Traffic State Detection

YOU Jin-ming, FANG Shou-en, TANG Tang, ZHANG Lan-fang   

  1. Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University, Shanghai 201804, China
  • Received:2018-03-28 Revised:2018-05-09 Online:2018-08-25 Published:2018-08-27

不良交通流状态实时监测支持向量机模型算法研究

游锦明,方守恩*,唐棠,张兰芳   

  1. 同济大学 道路与交通工程教育部重点实验室,上海 201804
  • 作者简介:游锦明(1992-),男,江西进贤人,博士生.
  • 基金资助:

    国家重点研发计划/National Key R & D Program of China(2016YFC0802701).

Abstract:

The real-time detection of traffic flow safety is the prerequisite for active traffic safety management. The paper sets the traffic states prior to the crashes as the criterion of hazardous traffic states. Parameters are extracted based on the lane-level traffic data and 9 parameters are selected with the principle component analysis. A support vector machine approach with radial basis kernel function is utilized to detect the real-time safety of the traffic flow. Grid search method is employed to select the optimized penalty parameter C and the kernel function parameter γ. Results indicate that the support vectors machine classifier could successfully classify 79.55% of the hazardous traffic states prior to the crashes. The method could be utilized to detect the hazardous traffic states on freeway in real time efficiently.

Key words: traffic engineering, hazardous traffic state, real-time detection, support vector machine

摘要:

交通流安全实时预警是交通主动安全防控的重要前提.采用实际事故发生前的交通流状态作为不良交通流状态判别标准,通过对车道级交通流数据进行参数提取,结合主成分分析法进行参数降维后得到9个主要参数.建立以径向基为核函数的交通流安全实时预警支持向量机模型,采用网格遍历法确定最优的支持向量机模型的惩罚参数C和核函数参数γ,最终构建的支持向量机模型能够成功地识别79.55%事故对应的不良交通流状态,能够有效地对高速公路上的不良交通流状态进行实时监测预警.

关键词: 交通工程, 不良交通流状态, 实时监测, 支持向量机

CLC Number: