关键词: 短时交通流量预测, 时空特性, 卷积神经网络, 长短时记忆, 注意力机制

Abstract: Accurate traffic flow prediction is very important in helping traffic management departments to take effective traffic control and guidance measures and help travelers reasonably plan routes. Aiming at the problem that the traditional deep learning model does not consider the temporal and spatial characteristics of traffic data, the STCAL (Spatial-Temporal Convolutional Attention-LSTM network) based on attention mechanism was established under the theoretical framework of Convolutional Neural Network (CNN) and Long and Short-Term Memory unit (LSTM) and combined with the temporal and spatial characteristics of urban traffic flow. Firstly, the fine-grained grid method was used to construct the spatio-temporal matrix of traffic flow; secondly, CNN model was used as a spatial component to extract the spatial characteristics of urban traffic flow in different periods; finally, the LSTM model based on attention mechanism was used as a dynamic time component to capture the temporal characteristics and trend variability of traffic flow, and realize the prediction of traffic flow. The experimental results show that compared with GRU and ST-ResNet, the RMSE index of the model STCAL is reduced by 17.15% and 7.37%, the MAE index is reduced by 22.75% and 9.14%, and the R2 index is increased by 11.27% and 2.37% respectively. At the same time, it is found that the prediction effect of the model on weekdays with high regularity is higher than that on weekends, and the prediction effect of morning peak on weekdays is the best, which can provide a basis for short-term urban regional traffic flow change monitoring.

Key words: short-term traffic flow prediction, spatial-temporal feature, Convolutional Neural Network (CNN), Long Short-Term Memory(LSTM), attention mechanism