The invention provides a traffic flow prediction method based on a dynamic space-time similar pyramid network. The traffic flow prediction method can effectively support traffic management departments to carry out traffic management and decision making. The method comprises the following steps of: extracting dynamic long-term characteristics of traffic flow by using improved multi-head space-time self-attention, capturing global space information and space similarity by using a space pyramid module, extracting time dependence of different granularities by using a time pyramid module, transmitting output of a space-time block to a prediction layer through residual connection, and predicting the traffic flow by using the prediction layer. And finally obtaining a traffic flow prediction result. Based on the result obtained by the method, the traffic flow information of the urban road can be efficiently and accurately predicted, abnormal congestion or abnormal dredging in the urban road can be accurately found, and help is provided for a traffic management department to formulate traffic scheduling and control strategies.
本发明提出一种基于动态时空相似金字塔网络的交通流预测方法,能够有效支持交管部门进行交通管理以及决策。其中包括使用改进的多头时空自注意力提取交通流的动态长期特征、使用空间金字塔模块捕获全局空间信息和空间相似性、使用时间金字塔模块提取不同粒度的时间依赖、将时空块的输出通过残差连接传递到预测层,最终得到交通流预测结果。基于本方法得到的结果能够高效准确地预测城市道路的交通流量信息,准确发现城市道路中的异常拥堵或异常疏通,为交通管理部门制定交通调度和控制策略提供帮助。
Traffic flow prediction method based on dynamic space-time similar pyramid network
一种基于动态时空相似金字塔网络的交通流预测方法
2024-06-21
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
European Patent Office | 2024
|Space-time adaptive dynamic graph convolutional network traffic flow prediction method
European Patent Office | 2024
|Traffic flow prediction method based on dynamic space-time correlation
European Patent Office | 2021
|Traffic flow prediction method based on dynamic multi-graph space-time synchronization network
European Patent Office | 2024
|Traffic flow prediction method based on dual dynamic space-time automatic encoder network
European Patent Office | 2024
|