As transportation systems continue to evolve, the demand for accurate prediction of traffic volume at highway entry and exit areas has become increasingly urgent. This study involves a two-year data collection from entry and exit points at four distinct locations along highways, revealing the distribution patterns of traffic volume across different time periods. By accounting for seasonal and long-term dependencies in time series data, both SARIMA and GRU models were developed. The results demonstrate that the SARIMA model achieved an R2 of 0.7762 and an RMSE of 8.6534, whereas the GRU model attained an R2 of 0.9533 and an RMSE of 3.9545. The GRU model outperforms the SARIMA model in traffic volume prediction tasks, particularly excelling in long-term forecasting. This research provides valuable insights for the future intelligent and precise management of highway traffic systems.
Traffic volume prediction for highway entry and exit areas based on machine learning
International Conference on Smart Transportation and City Engineering (STCE 2024) ; 2024 ; Chongqing, China
Proc. SPIE ; 13575
2025-04-28
Conference paper
Electronic Resource
English
Collaborative prediction method and system for entry and exit traffic flow of highway toll station
European Patent Office | 2025
|ENTRY/EXIT MANAGEMENT DEVICE, ENTRY/EXIT MANAGEMENT METHOD AND ENTRY/EXIT MANAGEMENT SYSTEM
European Patent Office | 2020
|