The invention discloses a dynamic graph convolution long-term traffic flow prediction method based on time perception. The method comprises the following steps: firstly, capturing traffic flow by using an attention-driven dynamic adaptive graph convolution network to obtain potential spatial dependence; then, time perception is combined with a multi-view time-dependent encoder to extract time correlation of a periodic mode, and the time perception capability of the model is enhanced through a time perception matrix; and finally, a time-aware dynamic graph convolutional network model for long-term traffic flow prediction is established. According to the method, the perception capability of time information is enhanced, the prediction result is more accurate, more powerful and more accurate reference can be provided for management of traffic decision makers, and the traffic jam phenomenon is greatly relieved.
本发明公开了一种基于时间感知的动态图卷积长期交通流预测方法,包括首先利用注意力驱动的动态自适应图卷积网络来捕交通流获潜在的空间依赖;接着利用时间感知联合多视图时间依赖编码器,来提取周期模式的时间相关性,还通过时间感知矩阵增强了模型的时间感知能力。最终建立一种面向长期交通流预测的时间感知动态图卷积网络模型。该方法增强了对于时间信息的感知能力,预测结果更加精确,能够为交通决策者的管理提供更有力和更准确的参考,大大缓解了交通拥堵现象。
Dynamic graph convolution long-term traffic flow prediction method based on time perception
基于时间感知的动态图卷积长期交通流预测方法
2024-10-08
Patent
Electronic Resource
Chinese
Traffic Flow Prediction Based on Dynamic Time Slot Graph Convolution
Transportation Research Record | 2025
|Traffic prediction method based on dynamic graph convolution
European Patent Office | 2023
|Long-time-sequence traffic flow prediction method based on graph convolution-Informer model
European Patent Office | 2021
|Dynamic graph convolution traffic speed prediction method
European Patent Office | 2020
|Traffic flow prediction method based on dynamic sparse graph convolution GRU
European Patent Office | 2025
|