Travel Time Forecasting with Combination of Spatial-Temporal and Time Shifting Correlation in CNN-LSTM Neural Network

摘要

The problem of short-term travel time estimation has been intensively investigated recently. However, accurate travel time predicting is still a challenge due to dynamic changes of the traffic and the difficulty of extracting urban traffic data features. In this paper, we mainly focus on time shifting feature of urban roads, which represents the impact of the upstream sections that will be conveyed to the downstream sections after a certain period of time △t. Firstly, we obtain the spatial relationships of the traffic time with Kullback-Leibler divergence (KL-divergence) and urban road networks. Then a Convolutional Neural Network (CNN) module is adopted to extract the spatial-temporal and time shifting information of the target road. Finally, a novel deep architecture combined CNN and Long-short Term Memory Recurrent Neural Network (LSTM) is utilized to predict the short-term travel time. The experimental result on the real data set shows that the proposed model is more effective than other existing approaches.