Optimization of Energy-Saving Operation Strategy for Virtual Coupling based on DDPG

Virtual coupling has emerged as a prominent research focus in the railway industry in recent years. At present, the existing research on synchronous operation of virtually coupled trains focuses on synchronization but neglects energy-saving. In pursuit of minimizing energy consumption, while ensuring synchronous operation, a virtual coupling speed curve optimization method based on deep deterministic policy gradient (DDPG) is proposed. Firstly, the virtually coupled trains are considered to build an optimization model aiming at energy-saving, synchronous parking, synchronous departure and on-time operation. Secondly, considering the characteristics of high state dimension and difficult solution of optimization model, a speed curve solving algorithm based on DDPG is designed. The optimal speed curve is considered a benchmark or reference for virtually coupled trains. Finally, the results of numerical experimental using the data of Beijing Metro Line demonstrate that the proposed method can reduce the traction energy consumption by 5.8% and the arrival time difference by 50%. In comparison to with genetic algorithms (GA), the optimization results are more energy-efficient and synchronous.