基于PSO-LSSVR代理模型的石化多级离心泵叶轮优化设计  

Optimization design of petrochemical multistage centrifugal pump impeller based on PSO-LSSVR proxy model

作　　者：李良 李晓俊[1] 杨顺银 陈小光 朱祖超[1] LI Liang;LI Xiaojun;YANG Shunyin;CHEN Xiaoguang;ZHU Zuchao(Key Laboratory of Fluid Transmission Technology of Zhejiang Province,Zhejiang Sci-Tech University,Hangzhou 310018,China;Ebara Great Pumps Co.,Ltd.,Wenzhou 325000,China)

机构地区：[1]浙江理工大学流体传输系统技术国家地方联合实验室,杭州310018 [2]嘉利特荏原泵业有限公司,浙江温州325000

出　　处：《流体机械》2023年第9期42-50,共9页Fluid Machinery

基　　金：国家自然科学基金项目(U21A20126);浙江省自然科学基金项目(LR20E090001);浙江省重点研发计划项目(2020C01027,2020C01091)。

摘　　要：针对加氢进料多级泵的高功耗和低能效的情况,有必要对该泵进行结构优化设计并减少能量损失,引入机器学习,建立了支持向量回归(SVM)代理模型,并采用粒子群优化算法(PSO)解决最小二乘支持向量回归(LSSVR)代理模型的超参数确定问题。以HDB型多级离心泵为研究对象,对首级双吸叶轮与次级单吸叶轮的叶片数、进出口角、包角以及出口宽度同时进行了优化,并对优化结果进行了响应分析;用数学性能指标评估了代理模型的精度,通过数值模拟对比了优化前后模型泵的性能。结果表明,PSO-LSSVR代理模型精度高,可以用来预测离心泵的外特性并优化,优化后多级泵的扬程和效率分别提高了3.99%和2.91%,优化后的内部流场更加稳定,更有利于流体的能量转换。In view of the high power consumption and low energy efficiency of the hydrogenation feed multistage pump,it is necessary to optimize the structure of the pump and reduce its energy loss.The machine learning was introduced to establish the support vector regression(SVM)proxy model,and particle swarm optimization(PSO)was used to solve the problem of determining the hyperparameters of the least squares support vector regression(LSSVR)proxy model.Taking an HDB multistage centrifugal pump as the research object,the number of blades,inlet and outlet angles,wrap angles and outlet widths of the first stage double suction impeller and the second stage single suction impeller were optimized simultaneously,and the response analysis of the optimization results was carried out.At the same time,the accuracy of the proxy model was evaluated by mathematical performance indicators,and the performance of the model pump before and after optimization was compared by numerical simulation.The results show that the PSO-LSSVR proxy model has high accuracy and can be used to predict and optimize the external characteristics of centrifugal pumps;After optimization,the head and efficiency of the multistage pump increased by 3.99%and 2.91%respectively;The optimized internal flow field is more stable and more conducive to the energy conversion of the fluid.

关 键 词：多级离心泵 支持向量机 粒子群算法 叶片优化 数值仿真 

分 类 号：TH3[机械工程—机械制造及自动化] TK73[交通运输工程—轮机工程]