车用燃料电池空压机叶轮多工况气动优化设计

Aiming at variable working conditions and the problem of excessive parasitic power of fuel-cell system caused by over high pressure ratio of fuel-cell centrifugal compressor, a multi-objective and multi-condition aerodynamic optimal design method with constraints for fuel-cell air compressor impeller is proposed based on parametric design, Latin hypercube sampling, radial basis function neural network and multi-objective grey wolf optimization algorithm. The numerical programs are developed independently to realize the design process of aerodynamic optimization. Taking the impellers of a two-stage fuel-cell centrifugal air compressor as the optimization objects, Latin hypercube sampling is used to obtain the sample space of the key design variables of the impeller. Based on the internal code of flow-field analysis, the corresponding aerodynamic performance target parameters of the impeller samples are calculated. On this basis, the neural network program is used to establish the flow-field analysis surrogate model. Taking the efficiency of design point and the efficiency and pressure ratio of non-design point as objectives, and the design pressure ratio as constraint, the multi-objective grey wolf algorithm program is used to carry out the multi-condition and multi-objective aerodynamic optimization of impellers. The results show that the Latin hypercube sampling achieves the uniform distribution of sample points in the design variable space, and the surrogate model established by neural network can accurately describe the mapping relationship between the design variables and the performance targets, and the maximum error between the performance targets obtained by the surrogate model and flow-field calculation is less than 1%. The optimal efficiency and the corresponding impeller profile under the constraint of design pressure ratio are obtained by optimization calculation. After optimization, the low velocity region and entropy increment in the impeller flow field at the design and non-design operating points both reduced, and the isentropic efficiencies of the two impellers at the design operating point were increased by 2.2% and 2%, respectively, and the efficiencies of the two impellers at the non-design operating point were increased by 2.9% and 2.2%, respectively.