基于遗传算法的离心泵叶轮参数化造型及优化设计

根据流面流动理论,通过坐标变换实现离心泵叶轮子午面及叶片结构的参数化造型.采用自适应策略等技术对遗传算法的遗传操作进行改进以提高算法搜索效率,并设计了多目标决策的适应值函数.采用CFD软件NUMECA对叶轮及压水室等主要过流部件内部流动进行三维定常计算,从而预测离心泵的水力性能.编写FORTRAN程序实现参数化、搜索算法及性能预测3部分的联合,开发出一种离心泵叶轮造型的自动优化方法.以离心泵的2个外特性参数效率和扬程作为优化目标,应用该方法对Dn1000型潜油泵及IS80-65-125型清水泵的叶轮结构进行多目标优化.优化后两泵的水力性能有了明显提高,在设计工况下扬程分别提高了0.21 m和1.70 m,效率分别提高了1.9%和2.3%.

Parametric modeling and optimization design of centrifugal pump impeller based on genetic algorithm

An automatic optimization method for centrifugal impeller shapes was established with FORTRAN program,in which the impeller meridian plane and blade profile were parameterized according to the theory of stream surface and coordinate transformation,genetic algorithm was improved via adaptation strategy to increase algorithm efficiency.CFD software NUMECA was used to simulate the 3D steady internal flow in impellers,diffusers,and other main hydraulic components of pumps to predict the hydraulic performance.Taking the pump performance parameters of efficiency and head as the optimization multi-objectives,the impeller structures of Dn1000 oil submerged pump and IS80-65-125 clean water pump were optimized.The performance of the optimized pumps is improved obviously.Under the design condition,the two pumps′ head increases by 0.21 m and 1.70 m,and the efficiency increases by 1.9% and 2.3%,respectively.