Accurate prediction of traffic flow is the basis for realizing intelligent transportation systems. It is challenging to achieve accurate prediction of highway traffic flow because of the characteristics of highway traffic state evolution such as temporal non-linearity and spatial heterogeneity. In this paper, a Conv-LSTM freeway traffic flow prediction model based on an attention mechanism is proposed, which can automatically extract the inherent features of historical traffic flow data. First, the convolution and Long Short-Term Memory model are combined to form a Conv-LSTM module based on the attention mechanism, which could take out the time-space features of the traffic flow data. The attentional mechanisms are designed to identify the importance of different flow series. In addition, the Bi-LSTM module is used to analyze the historical traffic flow data to capture the trend of traffic flow in the forward and backward directions to extract the daily and weekly traffic flow cycle features. Finally, the results show a better prediction performance realized by the proposed integrated model compared to other available approaches.
Research on freeway traffic flow prediction method based on Att-Conv-LSTM model
Seventh International Conference on Traffic Engineering and Transportation System (ICTETS 2023) ; 2023 ; Dalian, China
Proc. SPIE ; 13064
2024-02-20
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Instant traffic flow prediction method based on Conv-LSTM
Europäisches Patentamt | 2023
|Conv-Bi-LSTM model-based ship heaving motion prediction system and prediction method
Europäisches Patentamt | 2023
|A Deep Learning Model with Conv-LSTM Networks for Subway Passenger Congestion Delay Prediction
DOAJ | 2021
|