The invention provides a traffic prediction method based on deep learning. The method comprises the following steps: obtaining historical representation of traffic state information and spatio-temporal information representing a first number of historical time steps and future representation of spatio-temporal information representing a second number of future time steps; processing the historical representation by using a first BERT model to obtain a first state code; adding the first state code and the future representation to obtain a prediction representation; and processing the prediction representation by using a second BERT model to obtain a predicted traffic state. According to the method, the hidden space-time dependency in the traffic data can be effectively captured, and the accuracy of long-term prediction is improved.
本发明提供一种基于深度学习的交通预测方法,包括:获取表征第一数量个历史时间步的交通状态信息和时空信息的历史表征以及表征第二数量个未来时间步的时空信息的未来表征;应用第一BERT模型对所述历史表征进行处理,获取第一状态编码;将所述第一状态编码与所述未来表征相加,得到预测表征;应用第二BERT模型对所述预测表征进行处理,获取预测的交通状态。通过以上方式,本发明能够有效捕获交通数据中隐藏的时空依赖性,提高长期预测的准确率。
Traffic prediction method based on deep learning
一种基于深度学习的交通预测方法
2023-07-07
Patent
Electronic Resource
Chinese
Traffic flow prediction method based on hybrid deep learning
European Patent Office | 2023
|Traffic speed prediction using deep learning method
IEEE | 2016
|Urban traffic flow prediction method based on deep learning
European Patent Office | 2025
|