本发明提供一种多路口交通信号灯控制方法、装置、电子设备及存储介质,采用的本地共享模型在训练时并不需要其他路口的训练样本,而是以联邦强化学习算法,借助于初始本地私有模型以及中央处理器智能体中与初始本地共享模型结构相同的初始全局模型进行集中‑分布式协同训练得到,可以保证各路口的时序交通状态信息观测样本的隐私性,即使不共享的情况下也可以得到准确的本地共享模型,避免了数据孤岛问题的出现。而且,由于采用基于联邦强化学习算法的集中‑分布式协同训练的方式,可以避免出现现有技术中对强化学习智能体训练时探索空间会呈现指数级增长的问题出现,可以实现最优化全局道路网络的交通状况。

    The invention provides a multi-intersection traffic signal lamp control method and device, electronic equipment and a storage medium, and an adopted local sharing model does not need training samples of other intersections during training, but adopts a federal reinforcement learning algorithm. By means of an initial local private model and an initial global model with the same structure as an initial local shared model in a central processing unit intelligent body, the method is obtained through centralized-distributed cooperative training, so that the privacy of time sequence traffic state information observation samples of all intersections can be ensured; the accurate local sharing model can be obtained even under the condition that sharing is not carried out, and the problem of data islands is avoided. Moreover, due to the adoption of a centralized-distributed cooperative training mode based on a federal reinforcement learning algorithm, the problem of exponential growth of an exploration space during reinforcement learning agent training in the prior art can be avoided, and the traffic condition of the global road network can be optimized.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    多路口交通信号灯控制方法、装置、电子设备及存储介质


    Erscheinungsdatum :

    2024-05-14


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen