The invention discloses an expressway road cooperative control system and method based on deep reinforcement learning, wherein the system comprises a traffic information interaction module, a trafficcontrol module, a deep learning network training module, and a plurality of variable speed limit and ramp control units. The traffic status of ta road is obtained through the information interaction module, and then the traffic status is transmitted to the traffic control module. And the control strategy is continuously optimized through the training module, and the stability of the training process is ensured by adopting a deep reinforcement learning algorithm with an actor-critic architecture. All traffic control units in the system can be controlled at the same time, the problems of trafficstatus space explosion and the like cannot be caused, it can be guaranteed that vehicles pass through bottleneck road sections at a high speed, and passing of surrounding road vehicles cannot be affected by queuing and the like.
本发明公开了一种基于深度强化学习的高速公路道路协同控制系统及方法,系统包括交通信息交互模块、交通控制模块、深度学习网络训练模块以及若干可变限速和匝道控制单元,通过信息交互模块获取道路的交通状态,再传递给交通控制模块。后者通过训练模块不断优化控制策略,并采用具有actor‑critic架构的深度强化学习算法保证训练过程的稳定性。本发明能同时控制系统中的所有交通控制单元,且不会造成交通状态空间爆炸等问题,能保障车辆以较高速度通过瓶颈路段,且不会因为排队等问题影响周边道路车辆的通行。
Expressway road cooperative control system and method based on deep reinforcement learning
基于深度强化学习的高速公路道路协同控制系统及方法
2021-01-29
Patent
Electronic Resource
Chinese
Expressway vehicle-road cooperative control system based on V2X
European Patent Office | 2021
|Vehicle-road cooperative control method for expressway entrance ramp
European Patent Office | 2023
|Expressway generalized vehicle-road cooperative traffic management and control system
European Patent Office | 2021
|European Patent Office | 2022
|