An Automatic Driving Control Approach for Heavy-Haul Trains Based on PER-DDQN

Résumé fourni par la source

The automatic train operation (ATO) system for heavy-haul trains (HHT) is crucial to ensure efficient and stable train operation. In this study, a HHT control method using preferential empirical replay (PER) and double-deep Q-network (DDQN) algorithms is proposed. The model takes into account the characteristics of the HHT as well as constraints such as speed, coupling forces and inflation time. Subsequently, the control process is described as a Markov decision process, while a control method based on the PER-DDQN algorithm is proposed. A 20,000-ton HHT is used as the control object in simulation experiments conducted at different gradients and radii of curvature. The experimental results show that the proposed control method can operate efficiently and safely.