The invention relates to the field of traffic management, and discloses a deep reinforcement learning traffic signal decision system and method based on an attention mechanism. Comprising a traffic global controller, and a traffic signal lamp, a traffic information indication screen, a traffic information acquisition module, a data storage management module and a deep reinforcement learning module which are connected with the traffic global controller, the deep reinforcement learning module is used for obtaining a traffic flow prediction model and a traffic guidance decision model corresponding to the road network, and correcting the traffic flow prediction model and the traffic guidance decision model in time by using follow-up road traffic flow information; and the traffic global controller is used for receiving and forwarding the road vehicle target information and the road traffic flow information from the traffic information acquisition module, and generating traffic light signal control data and traffic information indication information according to the traffic flow prediction model and the decision model. The traffic efficiency of road network vehicles is greatly improved.
本发明涉及交通管理领域,公开了一种基于注意力机制的深度强化学习交通信号决策系统及方法,包括交通全域控制器以及与所述交通全域控制器连接的交通信号灯、交通信息指示屏幕、交通信息采集模块、数据存储管理模块以及深度强化学习模块,其中,所述深度强化学习模块用于获得对应该路网的交通流预测模型和交通指挥决策模型,并利用后续道路车流量信息对所述交通流预测模型和交通指挥决策模型进行及时修正;所述交通全域控制器用于接受并转发来自于交通信息采集模块的道路车辆目标信息、道路车流量信息,以及根据所述交通流预测模型和决策模型生成交通灯信号控制数据和交通信息指示信息。本发明大大地提升了路网车辆的通行效率。
Deep reinforcement learning traffic signal decision system and method based on attention mechanism
基于注意力机制的深度强化学习交通信号决策系统及方法
2024-03-26
Patent
Electronic Resource
Chinese
Deep reinforcement learning traffic signal control method based on attention mechanism
European Patent Office | 2024
|Adaptive traffic signal control method based on reinforcement learning and self-attention mechanism
European Patent Office | 2024
|Deep Reinforcement Learning-based Traffic Signal Control
IEEE | 2023
|Traffic signal control method based on deep reinforcement learning
European Patent Office | 2021
|