Energy sources durability energy management for fuel cell hybrid electric bus based on deep reinforcement learning considering future terrain information

Abstract

Environmental working conditions have a great impact on the power demand and fuel consumption of vehicles, and the rational use of road terrain information plays an important role in the improvement of vehicle economy and energy sources durability. In this paper, a fuel cell hybrid electric bus (FCHEB) energy management strategy (EMS) based on deep deterministic policy gradient (DDPG) algorithm taking into account future terrain information is proposed based on cyber-physical system (CPS) to reduce the economic cost of FCHEB. First, this paper implements the information exchange between the vehicle system and the network layer through CPS to obtain the environmental working conditions of the vehicle operation. Second, future terrain information is introduced into the framework for the first time with hydrogen consumption, battery degradation and fuel cell durability constraints to rationally allocate the power of the FCHEB. The results show that the proposed strategy improves the power battery durability by 7.39% and reduces the total operating cost by 5.76% compared to the EMS that ignores the future terrain information.