The invention discloses a digital twinborn traffic flow prediction system based on graph federal learning, which is characterized in that a sensor twinborn model and a sensor data processing model are arranged for each sensor in a traffic network and are deployed on a base station closest to the sensor; a sensor global twinborn model construction module and a global traffic flow prediction model are deployed in a server, then the models are trained by adopting a federated learning method, and when traffic flow prediction needs to be performed on a traffic network, the trained sensor data processing model processes data acquired by sensors, and the traffic flow prediction is performed on the traffic network. And inputting the obtained probability vector and the feature matrix into a global twin model construction module and a global traffic flow prediction model to obtain predicted traffic flow. According to the method, the influence of the dynamic nature of the traffic data and the time periodicity is comprehensively considered, and the image federation learning is further fused, so that the accuracy of traffic flow prediction is finally improved under the condition of ensuring the privacy of the user.
本发明公开了一种基于图联邦学习的数字孪生交通流预测系统,为交通网络中每个传感器设置一个传感器孪生模型和传感器数据处理模型,并部署在与传感器距离最近的基站上,在服务器部署传感器全局孪生模型构建模块和全局交通流预测模型,然后采用联邦学习的方法训练以上模型,当需要对交通网络进行交通流量预测时,训练好的传感器数据处理模型对传感器采集数据进行处理,将得到的概率向量和特征矩阵输入全局孪生模型构建模块和全局交通流预测模型得到预测交通流量。本发明综合考虑交通数据的动态性和时间周期性的影响,进一步融合图联邦学习,最终达到在保证用户隐私的情况下提高交通流量预测的准确性。
Digital twin traffic flow prediction system based on graph federated learning
基于图联邦学习的数字孪生交通流预测系统
2024-06-14
Patent
Electronic Resource
Chinese
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 |
Traffic flow prediction method based on federated learning and graph neural network
European Patent Office | 2024
|Cluster type federated learning-based urban traffic flow prediction method
European Patent Office | 2024
|