Abstract: To address the high complexity in matching and optimization design of multi-stage turbocharging systems for hydrogen engines， a coupled performance simulation model for two-stage turbocharged hydrogen engines based on turbocharger geometric configuration was constructed. Performance prediction models for turbines and compressors considering turbomachine thermodynamic processes and quasi-dimensional flow were established， and coupled with the hydrogen engine prediction model using GT-Power and Simulink， enabling coupled performance simulation and prediction of the hydrogen engine and turbocharging system. Based on the simulation method， a back-propagation（BP） neural network surrogate model between key geometric parameters of the two-stage turbocharging system and hydrogen engine thermal efficiency was developed， and sensitivity analysis was used to quantify the influence weights of key geometric parameters of the turbocharging system on engine performance. Finally， based on the reduced-order model combined with a multi-objective optimization algorithm， coordinated optimization and matching design of key geometric parameters for the two-stage turbocharging system were achieved. Coordinated optimization results indicate that the impeller trim ratio of the low-pressure stage compressor was increased from 50.96 to 59.50 to expand the stable operating range and the high-pressure stage turbine throat area was enlarged by 10% to improve the exhaust gas energy recirculation rate， positioning the compressor operating lines precisely within the high-efficiency zones. For the four operating conditions， the effective thermal efficiency showed a relative improvement of 0.88%~2.16%， the hydrogen consumption rate was reduced by 0.63~1.60 g/（kW·h） and the absolute value of pumping mean effective pressure（PMEP） markedly decreased. Meanwhile， the maximum in-cylinder combustion pressure was reduced by 1.12%~3.26%. This coordinated optimization method achieves synergistic improvements in multi-condition fuel economy and operational stability.