交通运输系统工程与信息

交通运输系统工程与信息 ›› 2019, Vol. 19 ›› Issue (5): 59-65.

• 智能交通系统与信息技术 • 上一篇    下一篇

基于突变强度的交通事件自动检测算法

李红伟1，姜桂艳*2, 3, 4, 5，李素兰6，朱宏伟7   

  1. 1. 河海大学土木与交通学院，南京 210024；2. 宁波大学海运学院宁波港航物流服务体系协同创新中心，浙江宁波 315211；3. 新疆理工学院，新疆阿克苏843100；4. 国家道路交通管理工程技术研究中心宁波大学分中心，浙江宁波 315211；5. 现代城市交通技术江苏高校协同创新中心，南京 210000；6. 武汉理工大学交通学院，武汉 430063；7.武汉市交通科学研究所，武汉 430015
  • 收稿日期:2019-04-22 修回日期:2019-05-22 出版日期:2019-10-25 发布日期:2019-10-25
  • 作者简介:李红伟（1982-），女，吉林蛟河人，讲师，博士.
  • 基金资助:

    国家自然科学基金/National Natural Science Foundation of China(71501061)；江苏省自然科学基金/Natural Science Foundation of Jiangsu Province, China(BK20150821)；湖北省交通运输科技项目/Hubei Transportation Science and Technology Project(2017-538-3-3).

An Automatic Incident Detection Algorithm Based on Mutation Strength

LI Hong-wei1, JIANG Gui-yan2, 3, 4, 5, LI Su-lan6, ZHU Hong-wei7   

  1. 1. College of Civil and Transportation Engineering, Hohai University, Nanjing 210024, China; 2. Faculty of Maritime and Transportation, Ningbo University, Ningbo 315211, Zhejiang, China; 3. Xinjiang Institute of Technology, Akesu 843100, Xinjiang, China; 4. National Traffic Management Engineering & Technology Research Centre Ningbo University Sub-centre, Ningbo 315211, Zhejiang, China; 5. Jiangsu Province Collaborative Innovation Center for Modern Urban Traffic Technologies, Nanjing 210000, China; 6. School of Transportation,Wuhan University of Technology,Wuhan 430063, China; 7.Wuhan Transportation Science Research Institute,Wuhan 430015, China
  • Received:2019-04-22 Revised:2019-05-22 Online:2019-10-25 Published:2019-10-25

PDF

摘要：

为设计1 种检测率高的快速路交通事件自动检测(Automatic Incident Detection， AID)算法，基于突变强度理论，分析交通事件下流量、速度、占有率突变强度在纵向时间序列的变化特征，得出事件时段，交通参数突变强度值较大. 本文以三参数突变强度乘积为事件评价指数设计了1 种快速路AID 算法. 新算法与3 种AID 算法对比得出：新算法检测率高 (100.00%)，误检率低(5.75%)；与横向时间序列相比，纵向时间序列数据稳定性更好；参数数量的增加可提高检测率. 新算法适用于各种流量，在低峰检测率为100.00%，误检率为0，检测效果最佳；高峰时段保持100.00%高检测率，误检率为5.66%，误检事件多发生在上下班早晚高峰和午休3个交通流量变化较大的时段.

关键词: 智能交通, 交通事件自动检测算法, 纵向时间序列, 突变强度, 快速路

Abstract:

In order to design an automatic incident detection(AID) algorithm for expressway traffic incidents with high detection rate, this paper analyses the change characteristics of traffic flow, speed and occupancy parameters in the longitudinal time series when traffic accidents occur based on the mutation strength theory, drawing the following conclusions: the values of mutation strength of traffic parameters in incident time are larger. An AID algorithm for expressway based on the product of mutation strength of three parameters as an incident evaluation index was established. By comparing the new algorithm with three algorithms based on the measured data of expressway traffic incidents, it was verified that the detection rate of the new algorithm is high, which is 100.00%, the false detection rate is 5.75%. At the same time, it is concluded that the stability of the longitudinal time series data is better than that of the transverse time series; the increase in the number of introduced parameters can improve the detection rate and reduce the false detection rate. The new algorithm is suitable for all kinds of traffic flow; in the low peak, the detection rate is 100.00% and the false detection rate is 0, the detection effect is the best; the detection rate in the peak period remains 100.00%, the false detection rate is 5.66% which is within the acceptable range; most incidents with wrong detection occurred in the morning peak, evening peak and noon break when traffic flow changes greatly.

Key words: intelligent transportation, automatic incident detection algorithm, longitudinal time series, mutation strength, expressway

中图分类号: