基于XGBoost和PB－NSGA-Ⅲ的热水循环泵水力性能优化OA

In order to improve the hydraulic performance of the hot water circulation pump,the hy-draulic efficiency and head of the YGB type hot water circulation pump were taken as the optimization objectives.The Plackett－Burman experimental design method was used to screen the variables that sig-nificantly affect the optimization objectives,namely the impeller outlet width,blade wrap angle,and blade outlet setting angle.The accuracy of three machine learning prediction models was compared and an approximate model for impeller structural parameters and hydraulic performance indicators were es-tablished based on the XGBoost algorithm.Based on this approximate model,second-generation and third-generation genetic algorithms were used for optimization,and a mapping method was used to se-lect a compromise solution as the optimization model.After optimization,the hydraulic efficiency of the model pump is increased by 3.20 percentage points.By comparing the numerical calculation results of the internal flow field of the hot water circulation pump before and after optimization,it is found that the optimized impeller channel becomes longer and the diffusion degree decreases.The high dissipation area of turbulent kinetic energy at the impeller tail and working face is controlled,and the flow loss is reduced.Based on the theory of entropy production,the local entropy production distribution in the span-wise direction of the impeller is explored.After optimization,the high entropy production areas near the blade tail and midspan are significantly reduced.Under design conditions,the entropy production in the impeller and volute is reduced by 7.84%and 5.58%,respectively,compared with the original model.