交通运输系统工程与信息 ›› 2009, Vol. 9 ›› Issue (3): 135-139 .

• 系统工程理论与方法 • 上一篇    下一篇

基于动态规划的城市交通流预测与分配

李一龙*   

  1. 湖南铁路科技职业技术学院,湖南 株洲 412000
  • 收稿日期:2008-01-04 修回日期:2009-04-07 出版日期:2009-06-25 发布日期:2009-06-25
  • 通讯作者: 李一龙
  • 作者简介:李一龙(1962-),男,湖南湘乡市人,副教授

Urban Traffic Flow Prediction and Assignment
with Dynamic Programming

LI Yi-long   

  1. Hunan Railway College of Science and Technology, Zhuzhou, Hunan 412000, China
  • Received:2008-01-04 Revised:2009-04-07 Online:2009-06-25 Published:2009-06-25
  • Contact: LI Yi-long

摘要: 城市交通系统是一个复杂的大系统,针对交通流的不确定性,本文基于动态规划思想提出了交通流预测与分配的方法。考虑路段容量对交通分配的影响,建立了路网流量预测和分配模型。为了保持并充分利用传统网络模型的性质和特征,引入惩罚函数,将容量约束条件转换到目标函数中,使模型符合传统均衡网络流结构。将凸规划法作为一个子过程植入惩罚函数,得到模型的求解算法。控制策略上采取预测控制、反馈校正和滚动优化的方式。最后,通过算例分析,进一步阐述模型和算法的应用,验证算法的有效性。为交通流预测和分配提供一定的参考。

关键词: 动态规划, 交通量预测, 交通流分配, 凸规划法

Abstract: In view of the uncertainty of the complex urban traffic system, this paper presents a traffic flow prediction and assignment method with dynamic programming. The model is developed by considering the capacity limitation in each section of the road network. In this model, the newly-added restraint condition is transferred by introducing punishment function, and the fine characters of traditional assignment model are maintained. Then, the algorithm is documented, in which Frank-Wolfe algorithm is combined with the punishment function. The control strategy consists of prediction control, feedback correction, and rolling optimization. Finally, a numerical example is used to illustrate the availability of the proposed method which provides reference for traffic flow prediction and assignment.

Key words: dynamic programming, traffic flow prediction, traffic flow assignment, Frank-Wolfe algorithm

中图分类号: