为了减少车辆通过路口时的延误,采用Q-学习方法对智能体控制的单路口进行信号配时的优化,在模糊控制规则集的基础上,通过Q-学习来改进控制规则的组合,从而达到改善信号控制效果的目的.仿真实验的结果表明,基于Q-学习的信号控制方法优于定时控制、感应式控制和基于遗传算法的信号控制方法.研究说明,基于Q-学习的信号控制方法适合城市交通控制. In order to reduce the delay when the vehicle passes through the intersection, the signal timing of the single-port controlled by the agent is optimized by Q-learning method. Based on the fuzzy control rule set, the combination of control rules is improved by Q-learning, Improve the signal control effect.The simulation results show that the Q-learning based signal control method is superior to the timing control, inductive control and signal control method based on genetic algorithm.The research shows that the Q-learning based signal control method is suitable Urban traffic control.