Research on Energy-saving Speed Curve of Heavy Haul Train Based on Reinforcement Learning

Total freight transportation is increasing year by year, and freight railway energy consumption is also increasing. This paper studies the energy-saving speed curve of heavy haul freight trains. Taking electric locomotive as the research object, this paper establishes a train dynamic model and the energy consumption method. An improved Q-learning algorithm is proposed to generate speed curve of the heavy haul trains, taking energy consumption and running time as the optimization indexes. In order to balance these two optimization indexes, the energy consumption target reward and the time target reward are updated in different ways. During the solution process, the action space can be cut according to the current state, and the updating formula is revised. Finally, the proposed algorithm is verified in a 20 km railway, and the results show that the speed curve calculated by the proposed algorithm is smoother and more energy-efficient compared with that of the traditional Q-learning algorithm.