Abstract: Aiming at the optimization design of the three dimensional twisted profiles of the compound impeller in an aero centrifugal pump, the study used Bezier curves to control the circumferential angle distribution of the four axial streamlines based on the Matlab platform in order to achieve the parameterization design of the composite blade profiles. On this basis, 125 groups of optimized samples are simulated with batch processing simulation method. Taking the efficiency function as the target, the performance agent model of composite impeller type parameters was constructed based on the Gauss radial basis function. Finally, the particle swarm optimization (PSO) algorithm is used to search the optimal design parameters for global optimization. The results show that the modified complex correlation coefficient of the agent model is 0.9269 and the standard deviation is 0.1036, which can meet the performance requirements of the parameter prediction, and the hydraulic efficiency of the optimized pump is improved by 2.72%.

Key words: Fuel centrifugal pump;Particle swarm optimization;Composite impeller;Optimum design

摘要： 针对航空离心泵复合叶轮三元扭曲型线的优化设计问题，基于Matlab平台采用五点四次Bezier曲线控制四条轴面流线的空间圆周角度分布，实现复合叶型的参数化设计。在此基础上，应用批处理联合仿真方法对125组优化空间样本进行数值模拟，并以效率函数为目标，基于高斯型径向基函数构建复合叶轮型线参数的性能代理模型。最后采用粒子群算法对优化设计参数进行全局寻优搜索。结果表明，代理模型的修正复相关系数为0.9269，标准差为0.1036，能够满足参数预测的性能要求，优化后整泵水力效率提高2.72%。

关键词: 燃油离心泵;粒子群算法;复合叶轮;优化设计