Many-objective optimization of fuel cell flow channel with triangular baffles based on many-population many-objective grey wolf optimizer

Synergistic optimization of flow channel structures and operating parameters is effective for mitigating concentration loss and enhancing net power in proton exchange membrane fuel cells (PEMFCs). However, existing optimization studies often address these aspects in isolation. In this study, a PEMFC flow channel simulation model based on experimental data and the multi-physical field characteristics of PEMFC mass transfer was constructed. On this basis, optimal Latin hypercube sampling (OLHS) was adopted to obtain comprehensive samples, and a high-precision data-driven surrogate model was established by combining the eXtreme Gradient Boosting (XGBoost) algorithm. Furthermore, a many-population many-objective grey wolf optimizer (MPMOGWO) was proposed to achieve global optimal matching between triangular baffle structural parameters and operating parameters. The findings indicate that the validated simulation model accurately captures the multi-physical behaviors with a maximum error of 1.76%. Supported by OLHS uniform sampling, the high-fidelity surrogate model (R2 > 0.98) ensures excellent predictive performance. The MPMOGWO-optimized scheme (N = 7, H = 0.6, Z = 4, T = 85, P = 160, S = 3.5) yields an average performance improvement of 19.78% under various operating conditions. This design induces intense secondary flows that significantly enhance reactant convective transport to the gas diffusion layer (GDL), while the local flow acceleration generates a robust convective sweeping effect for efficient water vapor removal. These findings provide critical data support for theoretical research and engineering optimization of triangular baffle flow channels in PEMFCs.