交通运输系统工程与信息 ›› 2018, Vol. 18 ›› Issue (5): 136-142.

• 系统工程理论与方法 • 上一篇    下一篇

基于极限学习机的公交行程时间预测方法

宋现敏*,刘明鑫,马林,夏英集   

  1. 吉林大学 交通学院,长春 130022
  • 收稿日期:2018-06-12 修回日期:2018-08-14 出版日期:2018-10-25 发布日期:2018-10-26
  • 作者简介:宋现敏(1978-),女,山东菏泽人,副教授,博士.
  • 基金资助:

    国家自然科学基金/National Natural Science Foundation of China(51278220);吉林省自然科学基金/ Jilin Natural Science Foundation(20180101063JC).

Bus Travel Time Prediction Based on Extreme Learning Machine

SONG Xian-min, LIU Ming-xin, MA Lin, XIA Ying-ji   

  1. College of Transportation, Jilin University, Changchun 130022, China
  • Received:2018-06-12 Revised:2018-08-14 Online:2018-10-25 Published:2018-10-26

摘要:

以公交车GPS数据为基础,建立了基于极限学习机方法的公交站点间行程时间预测模型.依据GPS数据在站点附近的特征表现,定义了公交车到站临界点,并分析了临界点处车辆的5种运行状态;提出了公交车到站时刻估算方法,进而得到公交车行程时间数据;通过分析公交车行程时间数据内在特征,确定了极限学习机模型关键参数及其纬度;最后,以长春市88路公交车GPS数据为基础进行了方法验证.结果表明,所用ELM方法预测误差约为11%,并与应用广泛的BP神经网络、RBF神经网络、SVM进行对比分析,发现ELM方法在满足精度前提下拥有更快训练速度与预测可靠性.

关键词: 智能交通, GPS数据, 极限学习机, 公交行程时间预测

Abstract:

A travel time prediction model of Extreme Learning Machine is established based on the bus GPS data. According to the characteristics of GPS data near the station, the critical point of bus arrival is defined. Through analyzing 5 running states of the vehicle at critical point, the estimation method of bus arrival time is proposed, and then get the bus travel time data. By analyzing the travel time data features of the bus, the key parameters and its number are determined. Finally, the GPS data of the 88 bus in Changchun city are taken to verification. The results show that the prediction error of the ELM method is about 11%. Compared with BP neural network, RBF neural network and SVM which are widely used, the ELM method has faster training speed and prediction reliability under the premise of satisfying the accuracy.

Key words: intelligent transportation, GPS data, Extreme Learning Machine, bus travel time prediction

中图分类号: