The invention provides a signal lamp and route cooperative control method based on offline reinforcement learning, and the method comprises the steps: obtaining data of different data sources, and carrying out the preprocessing of the data; designing the priority of the data source, and controlling the priority of the data source input graph neural network model according to the priority of the data source; setting an adaptive data fusion algorithm to fuse the data; performing graph neural network model training on the preprocessed data and the reward function by using an offline reinforcement learning algorithm; and detecting the traffic state and the output of the graph neural network model in real time, and switching to a predefined security policy when an abnormal or unsafe condition is detected. By means of the collaborative optimization mode, complex and dynamic traffic problems can be effectively solved, high safety and user experience are achieved, and the method is a traffic management solution with high innovativeness and practicability.
本发明提出了一种基于离线强化学习的信号灯及路由协同控制方法,方法包括:获取不同的数据源的数据,并对数据进行预处理;设计数据源优先等级,根据数据源的优先级对数据源输入图神经网络模型的优先级进行控制;设置自适应数据融合算法对数据进行融合;使用离线强化学习算法对预处理后的数据和奖励函数进行图神经网络模型训练;实时检测交通状态和图神经网络模型输出,并在检测到异常或不安全情况时切换到预定义的安全策略。本发明通过这种协同优化的方式,不仅能有效地解决复杂和动态的交通问题,而且具有很高的安全性和用户体验,是一种具有高度创新性和实用性的交通管理解决方案。
Signal lamp and route cooperative control method based on offline reinforcement learning
一种基于离线强化学习的信号灯及路由协同控制方法
2024-03-01
Patent
Elektronische Ressource
Chinesisch
Adaptive traffic signal control method based on CQL offline reinforcement learning
Europäisches Patentamt | 2023
|Europäisches Patentamt | 2024
|Traffic signal lamp control method based on migration reinforcement learning
Europäisches Patentamt | 2023
|Crossroad traffic signal lamp control method based on reinforcement learning
Europäisches Patentamt | 2020
|Traffic signal lamp control method and system based on reinforcement learning
Europäisches Patentamt | 2021
|