Thermal Management of Water-Cooled PEM Fuel Cell System with DDPG-FLC Strategy

The performance of polymer electrolyte membrane (PEM) fuel cell system is impacted directly by the load and thermal condition in practical road missions. Motivated by this, this paper proposes a hierarchical strategy combining the deep deterministic policy gradient (DDPG) method and fuzzy logic control (FLC), to satisfy the operating requirements of PEM fuel cell system under dynamic load conditions. Foremost, a hydro-electro-thermal coupled system model is established with the consideration of time varying characteristics. On this premise, the DDPG-based deep reinforcement learning method is exploited to allocate smartly the power demand of hybrid propulsion system. Following this endeavor, the FLC-based strategy integrating the transient feedback control is employed to confine the fuel cell system within the expected temperature boundary. Results suggest that the proposed strategy is efficient regarding the power allocation and thermal regulation.