Proton exchange membrane (PEM) electrolyzer have attracted increasing attention from the industrial and researchers in recent years due to its excellent hydrogen production performance. Developing accurate models to predict their performance is crucial for promoting and accelerating the design and optimization of electrolysis systems. This work developed a Koopman model predictive control (MPC) method incorporating fuzzy compensation for regulating the anode and cathode pressures in a PEM electrolyzer. A PEM electrolyzer is then built to study pressure control and provide experimental data for the identification of the Koopman linear predictor. The identified linear predictors are used to design the Koopman MPC. In addition

the developed fuzzy compensator can effectively solve the Koopman MPC model mismatch problem. The effectiveness of the proposed method is verified through the hydrogen production process in PEM simulation.