The invention discloses a variable lane and traffic signal cooperative control method based on deep reinforcement learning, and the method comprises the steps: collecting an intersection state observation value which comprises vehicle data, signal lamp data and variable lane data, carrying out the preprocessing, and inputting the preprocessed data to a neural network; continuously training and updating the neural network based on reinforcement learning until the model converges; and performing optimal control based on the trained neural network, and outputting an optimal control strategy. Coupling control over the variable lane and the traffic signal is achieved, real-time optimal control can be conducted according to the real-time state of the intersection area, manual operation is not needed, switching of the variable lane and control over the traffic signal are adjusted in a self-adaptive mode completely according to the traffic flow data, secondary parking of vehicles is not caused, and the time-space resource utilization efficiency of the signal control intersection is optimized.
本发明公开了一种基于深度强化学习的可变车道及交通信号协同控制方法,包括采集交叉口状态观测值,包含车辆数据、信号灯数据、可变车道数据,并进行预处理后输入至神经网络;神经网络基于强化学习不断训练更新,直至模型收敛;基于训练好的神经网络进行最优控制,输出最优控制策略。本发明实现了可变车道和交通信号的耦合控制,并且可以根据交叉口区域的实时状态,进行实时的最优控制,不需要人工作业,可变车道的切换和交通信号的控制完全根据车流数据进行自适应调节,并且没有造成车辆二次停车,优化了信号控制交叉口的时空间资源利用效率。
Variable lane and traffic signal cooperative control method based on deep reinforcement learning
基于深度强化学习的可变车道及交通信号协同控制方法
2020-11-10
Patent
Elektronische Ressource
Chinesisch
Europäisches Patentamt | 2024
|Traffic signal control method based on deep reinforcement learning
Europäisches Patentamt | 2021
|Deep Reinforcement Learning-based Traffic Signal Control
IEEE | 2023
|