The invention discloses a short-term traffic flow prediction method based on deep learning and belongs to the field of traffic prediction. The method comprises steps of firstly, extracting spatial features of traffic flow by using a convolutional neural network; secondly, extracting time features by using a gating circulation unit introducing an attention mechanism, and calculating importance of traffic flow features at different moments through the attention mechanism so that the model pays more attention to the features with high importance; extracting periodic features by using the periodicfeatures of the traffic flow data; and lastly, fusing all the features for prediction. The method is advantaged in that the defect that a prediction method in the prior art cannot fully utilize the spatial and temporal features of the traffic flow data is solved, prediction precision of the traffic flow is improved, and a problem of short-term traffic flow prediction can be better solved.
一种基于深度学习的短时交通流预测方法属于交通预测领域。本发明首先使用卷积神经网络提取交通流的空间特征;然后使用引入注意力机制的门控循环单元提取时间特征,通过注意力机制计算不同时刻交通流特征的重要性,使模型更关注重要性大的特征;接着利用交通流数据的周期特性提取周期特征;最后融合所有特征进行预测。该方法解决了现有预测方法无法充分利用交通流数据时空特征的缺点,提高了交通流的预测精度,可以更好地解决短时交通流预测问题。
Short-term traffic flow prediction method based on deep learning
一种基于深度学习的短时交通流预测方法
2020-05-29
Patent
Electronic Resource
Chinese
Deep Learning for Short-Term Traffic Flow Prediction
Online Contents | 2016
|Deep learning for short-term traffic flow prediction
Elsevier | 2017
|Short-term traffic flow prediction method and device using deep learning
European Patent Office | 2023
|