Digital Twin based Train Delay Prediction System: Design and Realization

This paper proposes a digital twin-based train delay prediction system for a high-speed railway network to realize real-time train delay prediction. Specifically, a deep learning model is proposed for delay prediction, whose input is the current state of the system, while output is the predicted delays. Then a system following the digital twin perspective is designed to observe the train operation state and transmit the state information to the delay prediction unit to fulfill the delay prediction process. The architecture of the digital twin system is designed, and the actual infrastructure data and operation data of the Beijing Railway Bureau are adopted to construct a prototype system. The function of the system is verified with numerous experiments.