燃料电池离心压缩机叶轮构型的参数优化及性能分析

In order to improve the aerodynamic performance of the centrifugal compressor to meet the demand of the fuel cell system, this paper takes a fuel cell centrifugal compressor impeller as the study object, selects the key configuration parameters such as the impeller inlet inclination angle, number of blades, radial flow path and other profile control points and blade mounting angle distribution control points as optimization variables, then uses the numerical calculation method to simulate and optimize the aerodynamic performance of the centrifugal impeller. Based on the Latin hypercube sampling and the BP neural network, this paper mapped the relationship between the impeller configuration parameters and the aerodynamic performance, then uses the maximum isentropic efficiency of the impeller as the optimization target, and takes the total pressure ratio and power as constraints. The genetic algorithm was used to optimize the key configuration parameters of the impeller, and the aerodynamic performance and internal flow characteristics of the impeller before and after optimization were compared and analyzed. The results show that the isentropic efficiency of the optimized impeller increases by 3.90%, its total pressure ratio is 1.742, and its power is 8.53 kW under the design conditions, which satisfies the design requirements. At the same time, the range of stable operating conditions of the impeller becomes wider, and the isentropic efficiency of the impeller is also improved in the whole range of operating conditions. The flow field analysis shows that the high entropy area on the cover side of the impeller is reduced after optimization, the velocity distribution in the pressure surface and outlet section of the blade is more uniform, the area of the low-speed gas mass is reduced in multiple directions, and the flow separation in the impeller channel is suppressed. The above verifies the effectiveness of the multi-parameter optimization design method of the centrifugal impeller in this paper.