Abstract
The primary objective of this study is to predict the short-term metro passenger flow using the proposed hybrid spatiotemporal deep learning neural network (HSTDL-net). The metro passenger flow data is collected from line 2 of Nanjing metro system to illustrate the study procedure. A hybrid spatiotemporal deep learning model is developed to predict both inbound and outbound passenger flows for every 10 minutes. The results suggest that the proposed HSTDL-net achieves better prediction performance on suburban stations than on urban stations, as well as generating the best prediction accuracy on transfer stations in terms of the lowest MAPE value. Moreover, a comparative analysis is conducted to compare the performance of proposed HSTDL-net with other typical methods, such as ARIMA, MLP, CNN, LSTM, and GBRT. The results indicate that, for both inbound and outbound passenger flow predictions, the HSTDL-net outperforms all the compared models on three types of stations. The results suggest that the proposed hybrid spatiotemporal deep learning neural network can more effectively and fully discover both spatial and temporal hidden correlations between stations for short-term metro passenger flow prediction. The results of this study could provide insightful suggestions for metro system authorities to adjust the operation plans and enhance the service quality of the entire metro system.
Highlights
A comparative analysis is conducted to compare the performance of proposed HSTDL-net with other typical methods, such as ARIMA, multilayer perceptron (MLP), CNN, long short-term memory neural network (LSTM), and gradient boosting regression tree (GBRT). e results indicate that, for both inbound and outbound passenger flow predictions, the HSTDL-net outperforms all the compared models on three types of stations. e results suggest that the proposed hybrid spatiotemporal deep learning neural network can more effectively and fully discover both spatial and temporal hidden correlations between stations for short-term metro passenger flow prediction. e results of this study could provide insightful suggestions for metro system authorities to adjust the operation plans and enhance the service quality of the entire metro system
For inbound passenger flow prediction, the mean absolute error (MAE) values of HSTDL-net on terminal stations have decreased by 7.17%, 33.92%, 29.57%, 43.42%, and 63.31% for each
49.5 compared model (GBRT, CNN, LSTM, MLP, and ARIMA), respectively. e MAE values of HSTDL-net on transfer stations have decreased by 41.43%, 43.87%, 41.40%, 47.27%, and 55.01% for each compared model, respectively. e MAE values of HSTDL-net on regular stations have decreased by 24.32%, 39.23%, 39.15%, 52.34%, and 66.26% for each compared model, respectively. e results of comparative analysis indicate that the proposed HSTDL-net can more effectively and fully discover both spatial and temporal hidden correlations between stations for short-term metro passenger flow prediction
Summary
Cantarella e primary objective of this study is to predict the short-term metro passenger flow using the proposed hybrid spatiotemporal deep learning neural network (HSTDL-net). E results suggest that the proposed hybrid spatiotemporal deep learning neural network can more effectively and fully discover both spatial and temporal hidden correlations between stations for short-term metro passenger flow prediction. Most previous studies have considered the short-term metro passenger flow prediction only as a typical time-series problem and failed to incorporate the hidden spatial correlations between stations to enhance the prediction accuracy [2, 7, 8]. Considering the spatiotemporal nature, it is essential to integrate both spatial and temporal characteristics into the short-term metro prediction models, which has great potential for improving the prediction performance in practical applications [10]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
