东北大学学报（自然科学版）

东北大学学报(自然科学版) ›› 2011, Vol. 32 ›› Issue (3): 443-446.DOI: -

• 论著 • 上一篇    下一篇

基于模糊神经网络的短时公交到站时间预测

李大铭;赵新良;林永杰;邹难;   

  1. 东北大学工商管理学院;山东大学控制科学与工程学院;
  • 收稿日期:2013-06-19 修回日期:2013-06-19 发布日期:2013-04-04
  • 通讯作者: -
  • 作者简介:-
  • 基金资助:
    辽宁省教育厅人文社会科学基金资助项目(2009JD31)

Short-term bus arrival time prediction using a fuzzy neural network

Li, Da-Ming (1); Zhao, Xin-Liang (1); Lin, Yong-Jie (2); Zou, Nan (2)   

  1. (1) School of Business Administration, Northeastern University, Shenyang 110819, China; (2) School of Control Science and Engineering, Shandong University, Jinan 250061, China
  • Received:2013-06-19 Revised:2013-06-19 Published:2013-04-04
  • Contact: Li, D.-M.
  • About author:-
  • Supported by:
    -

RichHTML

0

PDF (PC)

631

摘要： 研究了中国公交运营特点,利用全球定位系统和电子票务收费系统收集的车辆实时信息,建立了路段和站点补偿模糊神经网络模型,分别预测车辆的路段行驶时间和站点停留时间.路段预测模型的输入是所有重合线路的站点行驶数据,改变了现有预测模型只采用单线路数据的不足.以济南市一条实际公交线路为例,利用VISSIM模拟专用道和非专用道两种道路结构并计算到站时间预测值,结果证明:提出的模型性能明显优于平均值法和卡尔曼滤波法,15min内预测累积误差小于10%,而在公交专用道上误差小于7%.

关键词: 公交到站时间, 短时预测, 模糊策略, 补偿模糊神经网络, 重合线路

Abstract: This paper studies bus operation in China and presents a short-term bus arrival time prediction model using fuzzy neural network (FNN) with real-time traffic information collected by a global positioning system and an electronic payment system. Road segment and bus stop based FNN are proposed to predict bus travel time on road segments and time spent at bus stops. As opposed to previous studies, the present approach considers traffic data collected at bus stops with overlapping multiple bus routes instead of at those with single routes. Taking a bus artery in Jinan city as an example, this paper conducted intensive numerical experiments with simulated data from the microscopic simulator VISSIM in which both exclusive and non-exclusive bus lanes were simulated. Cumulative errors of predicted travel time were under 10% in bus non-exclusive and less than 7% in bus exclusive lanes when the prediction time window was less than 15 minutes, indicating that the proposed approach significantly outperforms the mean value method and the Kalman filtering model.

中图分类号: