黑龙江科技大学学报2025，Vol.35Issue(5)：855-861,7.DOI:10.3969/j.issn.2095-7262.2025.05.025
多智能体深度强化学习的区域交通信号协同控制方法
Cooperative control method for regional traffic signal based on multi-agent deep reinforcement learning
摘要
Abstract
This paper aims to address the proliferation of spatial dimensions of the control model caused by the increase in the number of agents in the regional traffic signal control system.The study consists of adopting the regional division strategy;establishing the regional multi-agent cooperative frame-work by constructing joint state,joint reward and spatial discount factors;combining the introduction of recurrent neural network and the improved DQN algorithm of BN algorithm,identifying a multi-agent deep reinforcement learning MADGB algorithm for regional traffic signal control.The simulation results show that compared with the current fixed timing strategy,the MADGB algorithm can improve the traffic efficiency of regional road network by 64.40%,and reduce fuel consumption by 38.39%and carbon di-oxide emission by 38.41%.关键词
交通信号控制/深度强化学习/多智能体/区域划分Key words
traffic signal control/deep reinforcement learning/multi-agent/regional division分类
信息技术与安全科学引用本文复制引用
时颖,曹青辰,陈义平..多智能体深度强化学习的区域交通信号协同控制方法[J].黑龙江科技大学学报,2025,35(5):855-861,7.基金项目
黑龙江省省属高校基本科研业务费项目(2023-KYYWF-0531) （2023-KYYWF-0531）
长安大学中央高校基本科研业务费专项资金资助项目(300102343516) （300102343516）
