基于LSTM预测交通流的DRL交通信号灯自动控制优化研究

Optimization of DRL traffic signal automatic control based on LSTM traffic flow prediction

随着城市交通流量的增加,高效的交通信号灯控制系统对于缓解交通拥堵具有重要意义。为此,对基于长短期记忆网络交通流预测与深度强化学习交通信号灯控制策略进行分析。研究首先构建了一个基于长短期记忆网络的交通流预测模型,然后引入思维进化算法进行优化。仿真结果显示,改进后的交通信号灯控制,使堵车次数从12次减少至6次,堵车时长从605 s缩短至334 s。结果表明,该方法可以有效提升城市交通管理的效率,减轻交通拥堵问题。

With the increase of urban traffic flow,an efficient traffic signal control system is of great significance to alleviate traffic congestion.Therefore,traffic flow prediction based on long short-term memory network and traffic signal control strategy based on deep reinforcement learning are analyzed.This paper firstly constructs a traffic flow prediction model based on long short-term memory network,and then introduces the thought evolution algorithm to optimize the traffic flow.The simulation results show that the improved traffic signal control reduces the number of traffic jams from 12 to 6,and the duration of traffic jams from 605s to 334s.The results show that this method can effectively improve the efficiency of urban traffic management and reduce traffic congestion.