基于Q-learning算法的单点信号控制研究  

Research on Single Point Signal Control Based on Q-learning Algorithm

作　　者：王祉祈 赵顗 马健霄[1] 吴林 WANG Zhi-qi;ZHAO Yi;MA Jian-xiao;WU Lin(College of Automobile and Traffic Engineering,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学汽车与交通工程学院,江苏南京210037

出　　处：《物流工程与管理》2021年第4期93-95,109,共4页Logistics Engineering and Management

摘　　要：为减少交通拥堵对于城市可持续发展的制约,优化交通控制,降低交通流延误。文中对单点两相位信号控制交叉口进行研究,运用强化学习中Q学习理论构建模型,通过VB配合仿真软件实现了单点交叉口的Q学习。并设计感应式信号配时方案进行仿真,将两种结果进行比较发现改进的信号配时方案确实能够减少排队长度,降低延误,在信号控制上能起到一定作用。In order to reduce the constraints of traffic congestion on urban sustainable development,optimize traffic control and reduce traffic flow delay.In this paper,the single point two-phase signal control intersection is studied,using the Q-learning theory in reinforcement learning to build the model,the Q-learning of single point intersection is realized by VB and simulation software.By comparing the two results,it is found that the improved signal timing scheme can reduce the queue length and delay,and can play a certain role in signal control.

关 键 词：单点交叉口 信号控制 强化学习 交通仿真 感应式信号控制 

分 类 号：U491.1[交通运输工程—交通运输规划与管理]