Computer Engineering and Applications ›› 2016, Vol. 52 ›› Issue (13): 265-270.

Previous Articles    

Multi-Agent traffic signal control based on adaptive genetic algorithm

CAO Jie1,2, ZHANG Ling1   

  1. 1.College of Electrical and Information Engineering, Lanzhou University of Technology, Lanzhou 730050, China
    2.Gansu Manufacturing Information Engineering Research Center, Lanzhou 730050, China
  • Online:2016-07-01 Published:2016-07-15

自适应遗传算法的Multi-Agent交通信号优化控制

曹  洁1,2,张  玲1   

  1. 1.兰州理工大学 电气工程与信息工程学院,兰州 730050
    2.甘肃省制造业信息化工程研究中心,兰州 730050

Abstract: The control model of multi-agent distributed road traffic signal is presented based on the analyses of the regional traffic signal control and the characteristics of multi-agent technology. Firstly, in order to conquer the shortcomings of traditional genetic algorithm premature convergence, this paper brings the adaptive genetic algorithm into the intersection subarea agent, which improves the global optimization ability; secondly, according to the change of traffic flow, using the subarea agent substituting traditional intersection agent can optimize green ratio [λ], thus shortening the average delay time D of intersection. The experimental results show that subarea agent instead of intersection agent, the control effect is similar, and hardware resources are saved. In the subarea agent, the signal control under the adaptive genetic algorithm is introduced which can quickly find the best timing plan, and make the average delay time shortest. Finally, the simulation experiment shows that the combination of adaptive genetic algorithm and intersection subarea agent has better performance in intersection signal control, and proves the feasibility of subarea agent substituting intersection agent.

Key words: multi-agent, adaptive genetic algorithm, intersection subarea agent, average delay time

摘要: 在区域交通多智能体信号控制系统中,由于传统遗传算法早熟收敛,全局搜索能力不强,无法快速找到最佳配时方案,同时没有考虑相邻交叉口的关联性,针对这种情况,提出交叉口子区Agent代替传统的交叉口Agent,在交叉口子区Agent中引入自适应遗传算法,算法根据交通流量的变化对绿信比[λ]进行优化,使交叉口平均延误时间[D]最短。实验结果表明交叉口子区Agent代替交叉口Agent后,控制效果相似,节省了硬件资源,在交叉口子区Agent中引入自适应遗传算法下的信号控制能迅速找到最佳配时方案,使平均延误时间最短。仿真实验表明,将基于自适应遗传算法的交叉口区域控制应用到交叉口信号控制中有更好的性能,证明了用交叉口区域智能体替代交叉口智能体的可行性。

关键词: 多智能体, 自适应遗传算法, 交叉口子区Agent, 平均延误时间