High-speed Train Operation Adjustment Strategy Based on Deep Deterministic Policy Gradient

High-speed train operation process becomes complex and changeable with the increasing train fleet size and railway network scale, and it is easy to be affected by perturbance. Current researches in train operation focused on trajectory optimization or accurate tracking operation, lacking online adjustment ability to resist perturbance. This paper proposes an energy-saving, rapid and real-time operation adjustment strategy to reduce delays under perturbance circumstances. First, we formulate the train operation process and definite the concept of elasticity. Then, we quantify the elasticity and design an elastic adjustment strategy to control train operation. Furthermore, the deep deterministic policy gradient algorithm (DDPG) is applied to improve the adaptability in different operation environments. Finally, comparative experiments based on the section between Wuhan and Changsha show that the operation adjustment strategy trained by DDPG has a better performance in energy-saving, dynamic adjustment and delay recovery.