In transportation systems, a vast volume of traffic data is generated on a daily basis. The contributing factors for this traffic include expanding urban population, aging infrastructure, uncoordinated traffic timings, etc. Since traffic congestion costs valuable time and fuel every day, it needs to be monitored every day to avoid accidents and to enable the proper flow of traffic. This study aims to predict traffic flow using advanced forecasting techniques such as Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and XGBoost models. Both the LSTM (Long-Short-Term Memory) and GRU (Gated Recurrent Unit) networks are used to predict the vehicle traffic flow and their prediction errors over different road junctions are compared to know which network works better. Experiments demonstrate that the proposed GRU model performs slightly superior to the LSTM model. The evaluation also shows that XGBoost-based methods perform the best in short-term and long-term traffic flow prediction.
Predicting Traffic Flow with Deep Learning
Lect. Notes in Networks, Syst.
International conference on soft computing for problem-solving ; 2023 ; Roorkee, India August 10, 2023 - August 12, 2023
Proceedings of the 12th International Conference on Soft Computing for Problem Solving ; Kapitel : 37 ; 527-541
01.07.2024
15 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Europäisches Patentamt | 2024
|