Short-term traffic flow prediction method based on deep learning and multi-layer spatial-temporal feature map

The invention provides a short-term traffic flow prediction method based on deep learning and a multi-layer spatial-temporal feature map. The method comprises the steps of obtaining historical trafficdata of a road segment needing traffic flow prediction; constructing a plurality of multi-layer space-time feature maps with the dimensions of M * N * C from all the traffic data according to a spatial position relationship, a time sequence and a data category; and finally, training the deep convolutional neural network by taking the multi-layer spatial-temporal feature map as a sample to obtaina short-term traffic flow prediction model, and performing traffic flow prediction. According to the method, the potential association and spatial-temporal correlation of historical traffic data are fully mined, and the accuracy and reliability of short-term traffic flow prediction are better improved by utilizing the redundancy of the data. 本发明提出了一种基于深度学习和多层时空特征图的短时交通流预测方法，获取需要进行交通流预测的路段的历史交通数据；将所有交通数据按空间位置关系、时间先后顺序、数据类别构