Real-Time Traffic Forecast System for the Accident-Prone Large-Scale Transportation Network in the Seoul Metropolitan Area

This study proposes a reliable and efficient real-time forecasting platform for use in an accident-prone large-scale transportation network. We showed the method could be applied to various roadway sections without any loss of performance or efficiency. Due to its robustness, efficiency, and versatility, the method could be implemented in the Seoul Metropolitan Area to provide traffic authorities and road users with future traffic information even under accident conditions. This is the major contribution of this research and contrasts with state-of-the-art techniques proposed by prior studies, which rely heavily on parameter tuning with large historical datasets and produce only site-specific forecasts with limited prediction horizons under recurrent traffic conditions. The proposed method makes no assumptions about the physical structure of the transportation network and can be applied to different roads under different traffic conditions without time constraints.