The centrifugal air compressor is the core component of the cathode air supply circuit in fuel cells, and its aerodynamic performance directly determines fuel cell performance and operational efficiency. To address low computational efficiency, long time consumption, and high computing power costs of CFD simulation for air compressors, and to improve compressor aerodynamic performance, this study constructed and experimentally calibrated the compressor’s numerical model and Stacking model. The NSGA-II optimization algorithm was used for multi-objective optimization of impeller structural parameters, with comparative analyses of structural parameters, aerodynamic performance, and flow field characteristics. Results show the proposed RF-XGB-CB Stacking model has high accuracy and generalization ability, and the NSGA-II algorithm achieves effective optimization. Compared with the baseline model, the optimized model maintains stable, high isentropic efficiency; its compression effect is significantly improved by 8%, and isentropic efficiency is increased by 2%. In terms of flow field characteristics, the optimized model has more uniform blade force, enhanced blade pressurization performance, lower entropy generation, and reduced flow and friction losses. This study provides important theoretical support for the multi-objective optimization design of centrifugal air compressor structural parameters.
Yin et al. (Fri,) studied this question.