Research on Coordinated Control Strategy of Fuel Cell Anode Pressure and Flow Based on Deep Reinforcement Learning

In view of the control problem of hydrogen pressure and flow coupling in the anode subsystem of the Proton Exchange Membrane Fuel Cell (PEMFC) under dynamic load conditions, this paper proposes an intelligent control framework based on the deep deterministic policy gradient (DDPG) algorithm. Firstly, a dynamic model of the fuel cell anode system is established to transform the multivariable coupling control problem into a reinforcement learning state space; secondly, a multi-objective reward function integrating pressure tracking accuracy, hydrogen utilization efficiency and actuator smoothness is designed, and a dual-channel Actor-Critic network structure with time series feature extraction capability is constructed to achieve transient complex condition control. Simulation experiments show that compared with the traditional model predictive control (MPC), the DDPG controller reduces the anode pressure fluctuation amplitude by 40.6% and the hydrogen excess coefficient overshoot by 20.2% under step load scenarios.