Q 学习算法是求解信息不完全马尔可夫决策问题的一种强化学习方法.Q 学习中强化信号的设计是影响学习效果的重要因素.本文提出一种基于模糊规则的 Q 学习强化信号的设计方法,提高强化学习的性能.并将该方法应用于单交叉口信号灯最优控制中,根据交通流的变化自适应调整交叉口信号灯的相位切换时间和相位次序.通过 Paramics 微观交通仿真软件验证,说明在解决交通控制问题中,使用基于模糊规则的 Q 学习的学习效果优于传统 Q 学习. Q learning algorithm is an intensive learning method for solving incomplete Markov decision problems.Q-learning is an important factor that affects the learning effect.This paper presents a design method of Q-learning enhanced signal based on fuzzy rules To improve the performance of reinforcement learning.And this method is applied to the optimal control of single-intersection traffic lights.According to the changes of traffic flow, the phase switching time and phase order of traffic lights can be adaptively adjusted.According to Paramics microscopic traffic simulation software, In solving traffic control problems, learning using Q learning based on fuzzy rules is better than traditional Q learning.