The invention discloses a traffic flow prediction method based on time window division modeling dynamic space-time correlation. The method comprises the following steps: inputting historical traffic flow data and traffic transfer flow data; reading and processing historical data; historical data embedding is standardized and generated by using time window expansion embedding; short-term local time features are extracted through a time attention module in the window; modeling a long-term global time pattern through a time attention module between windows; capturing dynamic spatial correlation through a spatial attention module between the windows; and projecting the embedded linearity to a multi-step prediction result through an output layer.
本发明公开了一种基于时间窗口划分建模动态时空关联的交通流量预测方法,包括以下步骤:输入历史的交通流量数据和交通转移流量数据;读取并处理历史数据;利用时间窗口展开嵌入来规范和生成历史数据嵌入;通过窗口内的时间注意力模块来提取短期的局部时间特征;通过窗口间的时间注意力模块来建模长期的全局时间模式;通过窗口间的空间注意力模块来捕捉动态的空间相关性;通过输出层将嵌入线性投影到多步预测结果。
Traffic flow prediction method based on time window division modeling dynamic space-time correlation
基于时间窗口划分建模动态时空关联的交通流量预测方法
2024-01-05
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Traffic flow prediction method based on dynamic space-time correlation
European Patent Office | 2021
|Time-space correlation traffic flow prediction method based on deep learning
European Patent Office | 2025
|Traffic flow prediction method based on dynamic space-time similar pyramid network
European Patent Office | 2024
|Traffic flow prediction method and system based on space-time correlation and error compensation
European Patent Office | 2023
|