计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (27): 231-233.DOI: 10.3778/j.issn.1002-8331.2010.27.065

• 工程与应用 • 上一篇    下一篇

基于SA-FQL算法的区域交通控制方法

邓 军,刘智勇   

  1. 五邑大学 信息学院,广东 江门 529020
  • 收稿日期:2009-01-13 修回日期:2009-03-20 出版日期:2010-09-21 发布日期:2010-09-21
  • 通讯作者: 邓 军

Urban traffic control based on SA-FQL

DENG Jun,LIU Zhi-yong   

  1. School of Information,Wuyi University,Jiangmen,Guangdong 529020,China
  • Received:2009-01-13 Revised:2009-03-20 Online:2010-09-21 Published:2010-09-21
  • Contact: DENG Jun

摘要: 将模拟退火算法的Metropolis准则用于平衡模糊Q学习中探索和扩张之间的关系,提出基于Metropolis准则的模糊Q学习算法Simulated Annealing Fuzzy Q-learning(SA-FQL)。利用SA-FQL算法优化区域的公共周期,在给定周期的基础上再用SA-FQL算法优化区域中各干线相邻两路口的相位差,最后根据交通流量确定各路口的绿信比。TSIS仿真结果表明,相比基于Q学习和模糊Q学习的控制方法,该方法能显著提高学习速度和交通效率。

关键词: 区域交通控制, 模拟退火模糊Q学习算法, 模糊Q学习, Q学习, Metropolis准则

Abstract: For area traffic control,the Simulated Annealing Fuzzy Q-Learning(SA-FQL),a modified version of the popular Fuzzy Q-Learning,is used.Firstly,this paper optimizes the common cycle of the traffic network using SA-FQL,then optimizes offset for every trunk road using the same method based on the common cycle,lastly adjusts split for every intersection according its traffic volume.TSIS simulation results show that,compared with Fuzzy Q-Learning method and Q-Learning method,the method proposed in this paper can significantly accelerate learning and improve traffic efficiency.

Key words: area traffic control, Simulated Annealing Fuzzy Q-Learning(SA-FQL), fuzzy Q-learning, Q-learning, Metropolis criterion

中图分类号: