Safe Deep Learning-Based Offset-Free MPC for Green Hydrogen Injected PEMFC Systems

Safe and stable operation of a proton exchange membrane fuel cell (PEMFC) system requires stringent control of oxygen excess ratio (OER). However, the OER regulation in PEMFC is challenging due to frequent fluctuations of current, various modeling nonlinearities, constrained manipulated variable, and real-time requirements. Offset-free model predictive control (MPC) provides a useful means for controlling systems with disturbances and constraints, but suffers from the heavy computational burden of repeatedly solving an optimization problem in real time. Such computational issue precludes the possibility of meeting the real-time requirements of PEMFC. In this paper, a PEMFC cathode gas supply model is firstly established. Next, we develop a safe deep learning-based offset-free MPC algorithm. Based on the nominal offset-free MPC, the proposed MPC not only reserves the ability of disturbance rejection, but also leverages deep neural networks for approximating the explicit solution to the MPC problem to greatly reduce online computational time. Furthermore, a gauge map is used to guarantee the satisfaction of safe constraints regarding compressor voltage. The simulation results show that the proposed MPC is an order of magnitude faster than the nominal offset-free MPC.