引入MARKOV过程预测的强化学习下的城市交叉口自适应交通信号配时决策

Urban Intersection Adaptive Traffic Signal Timing Decision with Markov Process Model Using Reinforcement Learning

针对城市交叉口交通流具有的动态性、不确定性等特性,采用Markov过程对进入交叉口各入口的交通流占有率进行预估,通过构造Q-强化学习方法对道路交叉口交通信号配时决策进行优化。以交叉口车均延误为性能指标,与传统的定时配时决策方法进行对比分析,验证了此强化学习配时决策方法的有效性,其对于动态交通流环境具有一定的适应能力。最后从系统集成等方面就提高城市交叉口交通信号配时决策效率提出一些对策。

Due to dynamic and uncertainty of the intersection' s traffic flow,Markov process was used to estimate traffic flow occupancy of every entrance of a intersection. Then,the Q-learning method which is one of the reinforcement learning(RL) tools was established in order to optimize the traffic signal timing decision. Simulation results show that the method surpass traditional fixed signal timing algorithm causing less average delay time. Therefore,the proposed RL algorithm could be a useful tool to adjust the proper traffic signal timing decision in practice. Finally,based on the above analysis,the suggestions on improving the efficiency of traffic signal timing decision were put forward.