Reinforcement Learning Models for Charge and Discharge Control in Hybrid Railway Vehicles

Abstract

Hybrid railway vehicles have been introduced to save energy in non-electrified lines. However, their high degree of freedom makes it difficult to establish theories that will allow the development of optimal charge and discharge control methods. Nevertheless, it is possible to achieve optimal charge and discharge control by using reinforcement learning. Therefore, in this paper, reinforcement learning models are proposed for charge and discharge control in hybrid railway vehicles.