基于孪生支持向量回归的多级离心泵外特性曲线预测及设计方法

doi:10.3901/JME.2024.14.364

机械工程学报 ›› 2024, Vol. 60 ›› Issue (14): 364-377.doi: 10.3901/JME.2024.14.364

扫码分享

基于孪生支持向量回归的多级离心泵外特性曲线预测及设计方法

童哲铭^1,2, 马鑫航^1,2

1. 浙江大学流体动力基础件与机电系统全国重点实验室杭州 310063;
2. 浙江大学机械工程学院杭州 310063

收稿日期:2023-08-03 修回日期:2024-04-01 出版日期:2024-07-20 发布日期:2024-08-29
作者简介:童哲铭,男,1988年出生,研究员,博士研究生导师。主要研究方向为旋转机械设计理论及方法。E-mail:tzm@zju.edu.cn
基金资助:
国家自然科学基金资助项目(52075481，52375274)。

Prediction and Design of the External Characteristic Curve of Multistage Centrifugal Pump Based on Twin Support Vector Regression

TONG Zheming^1,2, MA Xinhang^1,2

1. State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310063;
2. School of Mechanical Engineering, Zhejiang University, Hangzhou 310063

Received:2023-08-03 Revised:2024-04-01 Online:2024-07-20 Published:2024-08-29

摘要/Abstract

摘要： 准确快速地获取多级离心泵的外特性曲线，对设计过程中泵性能指标的判断具有重要的意义。现有基于计算流体力学(Computational fluid dynamics, CFD)数值模拟方法依赖设计人员经验、计算耗时以及偏工况点精度差，提出一种基于孪生支持向量回归(Twin support vector regression, TSVR)的离心泵外特性曲线预测方法。针对多级离心泵流道复杂的特点，筛选出10个结构参数和流量作为模型输入。通过TSVR方法建立输入变量与外特性值之间的复杂非线性关系，对多级离心泵的外特性曲线进行预测。与试验结果相比，所提方法可以较好地预测性能指标，对扬程、效率和轴功预测的平均相对误差分别为1.71%、3.56%、3.36%。此外，与BP神经网络相比，TSVR对扬程和效率预测的方均根误差显著降低。同时，在整个工况范围内，TSVR模型对比传统采用RNG湍流模型的CFD方法有更高的精度以及计算效率。对比表明，所提方法可提高多级离心泵性能曲线预测的速度和精度，为实现多级泵的快速预测及高效性能设计提供重要的支撑。

关键词: 离心泵, 支持向量回归, 数值模拟, 多工况, 性能预测

Abstract: Obtaining the external characteristic curve of a multistage centrifugal pump accurately and quickly is of great significance to the judgment of the pump performance index in the design process. The existing methods based on computational fluid dynanics(CFD) numerical simulation rely on the experience of designers, calculation time, and low accuracy of off-design points. Therefore, a prediction method of centrifugal pump external characteristic curve based on twin support vector regression(TSVR) is proposed. According to the complex characteristics of the multistage centrifugal pump flow channel, 10 structural parameters and flow are selected as the model input. The complex nonlinear relationship between structural parameters and external characteristic values is established by the TSVR method, and the external characteristic curve of the multistage centrifugal pump is predicted. Compared with the test results, the proposed method can better predict the performance indicators, and the average relative errors of head, efficiency, and axle work are 1.71%, 3.56%, and 3.36% respectively. In addition, compared with BP neural network, the root-mean-square error of TSVR for head and efficiency prediction is significantly reduced. At the same time, the prediction speed and accuracy of the TSVR model are far better than that of traditional CFD numerical calculation in the whole operating range. The comparison results show that the proposed method can improve the speed and accuracy of performance curve prediction of multistage centrifugal pumps, and can provide technical support for the fast prediction and efficient design of multistage pumps.

Key words: centrifugal pump, support vector regression, numerical simulation, multiple working conditions, performance prediction

中图分类号:

TH311

童哲铭, 马鑫航. 基于孪生支持向量回归的多级离心泵外特性曲线预测及设计方法[J]. 机械工程学报, 2024, 60(14): 364-377.

TONG Zheming, MA Xinhang. Prediction and Design of the External Characteristic Curve of Multistage Centrifugal Pump Based on Twin Support Vector Regression[J]. Journal of Mechanical Engineering, 2024, 60(14): 364-377.

参考文献

[1] LIU H，TONG Z，SHANG B. et al. Cavitation diagnostics based on self-tuning VMD for fluid machinery with low-SNR conditions[J]. Chinese Journal of Mechanical Engineering，2023，36，102，1-15.
[2] 欧顺冰，刘小兵，曾永忠. 基于CFX的中比转速离心泵的性能特性预测[J]. 水力发电，2012，38(6)：62-65. OU Shunbing，LIU Xiaobing，ZENG Yongzhong. CFX-Based performance and characteristics forecast of medium specific speed centrifugal pump[J]. Water Power，2012，38(6)：62-65.
[3] WANG B，GUAN H，YE Z. Numerical study of a fuel centrifugal pump with variable impeller width for aero-engines[J]. International Journal of Turbo and Jet Engines，2015，32(4)：341-350.
[4] BOZORGASAREH H，KHALESI J，JAFARI M，et al. Performance improvement of mixed-flow centrifugal pumps with new impeller shrouds：Numerical and experimental investigations[J]. Renewable Energy，Elsevier Ltd，2021，163：635-648.
[5] 谈明高，刘厚林，袁寿其. 离心泵水力损失的计算[J].江苏大学学报(自然科学版)，2007(5)：405-408. TAN Minggao，LIU Hou-lin，YUAN Shou-qi. Calculation of hydraulic loss in centrifugal pumps[J]. Journal of Jiangsu University(Natural Science Edition))，2007(5)：405-408.
[6] 杨军虎，姬亚亚，林彬，等. 单级单吸离心泵多工况的能量性能预测[J]. 兰州理工大学学报，2020，46(2)：48-54. YANG Junhu，JI Yaya，LIN Bin，et al. Energy performance prediction of single-stage single-suction centrifugal pump under multiple operating conditions [J]. Journal of Lanzhou University of Technology，2020，46(2)：48-54.
[7] FRIEDRICH G J. Centrifugal pumps[M]. Berlin：Springer，2008.
[8] DING S F，QI B J，Tan H Y. An overview on theory and algorithm of support vector machines[J]. Dianzi Keji Daxue Xuebao/Journal of the University of Electronic Science and Technology of China，2011，40(1)：2-10.
[9] HENGNIAN Q I. Support vector machines and application research overview[J]. Computer Engineering，2004，30(10)：6-9.
[10] 张学工. 关于统计学习理论与支持向量机[J]. 自动化学报，2000(1)：36-46. ZHANG Xuegong. Introduction to Statistical learning theory and support vector machinces[J]. Acta Automatica Sinica，2000(1)：36-46.
[11] DENG H，LIU Y，LI P，et al. Whole flow field performance prediction by impeller parameters of centrifugal pumps using support vector regression[J]. Advances in Engineering Software，Elsevier Ltd，2017，114：258-267.
[12] 罗会灿，周佩剑，吴登昊，等. 基于遗传-支持向量回归的离心泵性能曲线预测[J]. 水力发电学报，2022，41(3)：83-91. LUO Huican，ZHOU Peijian，WU Denghao，et al. Prediction of centrifugal pump performance curves based on genetic-support vector regression[J]. Journal of Hydroelectric Engineering，2022，41(3)：83-91.
[13] PENG X. TSVR：An efficient Twin Support Vector Machine for regression[J]. Neural Networks，Elsevier Ltd，2010，23(3)：365-372.
[14] HUANG H，WEI X，ZHOU Y. An overview on twin support vector regression[J]. Neurocomputing，Elsevier B.V.，2022，490：80-92.
[15] ZHANG X，YUAN S，ZHANG J，et al. Effects of spiral volute type on performance of high specific speed centrifugal pump[J]. Paiguan Jixie Gongcheng Xuebao/Journal of Drainage and Irrigation Machinery Engineering，2017，35(1)：25-31.
[16] CAO W，ZHANG Y，YAO L. Influence of outlet blade angel on performance of centrifugal pump and correction of slip factor[J]. Paiguan Jixie Gongcheng Xuebao/Journal of Drainage and Irrigation Machinery Engineering，2017，35(9)：755-760，773.
[17] PEI J，YUAN S，YUAN J. Dynamic stress analysis of sewage centrifugal pump impeller based on two-way coupling method[J]. Chinese Journal of Mechanical Engineering (English Edition)，2014，27(2)：369-375.
[18] HUANG R F，LUO X W，JI B，et al. Multi-objective optimization of a mixed-flow pump impeller using modified NSGA-II algorithm[J]. Science China Technological Sciences，2015，58(12)：2122-2130.
[19] CHEN R W，ZENG Y Z，LIU X B. Research on effects of blade thickness on the low-specific-speed centrifugal pump performance[J]. Machinery，Materials Science and Energy Engineering，2015(207-214)： 51279172.
[20] 谈明高，刘厚林，王勇，等. 叶轮外径对离心泵内流影响的CFD分析[J]. 排灌机械，2009，27(5)：314-318. TAN Minggao，LIU Houlin，WANG Yong，et al. CFD analysis on effects of impeller outlet diameter on flow field in centrifugal pump[J]. Drainage and Irrigation Machinery，2009，27(5)：314-318.
[21] 刘向东，骆斌，陈兆乾. 支持向量机最优模型选择的研究[J]. 计算机研究与发展，2005(4)：576-581. LIU Xiangdong，LUO Bin，CHEN Zhaoqian. Optimal model selection for support vector machines[J]. Journal of Computer Research and Development，2005(4)：576-581.
[22] 朱树先，张仁杰. 支持向量机核函数选择的研究[J].科学技术与工程，2008(16)：4513-4517. ZHU Shuxian，ZHANG Renjie. Research for selection of kernel functions used in support vector machine[J]. Science Technology and Engineering，2008(16)：4513-4517.

基于孪生支持向量回归的多级离心泵外特性曲线预测及设计方法

Prediction and Design of the External Characteristic Curve of Multistage Centrifugal Pump Based on Twin Support Vector Regression

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	郑洋, 赵梓昊, 刘伟, 余政哲, 牛伟, 雷贻文, 孙荣禄. 高性能镁合金增材制造技术研究进展[J]. 机械工程学报, 2024, 60(7): 385-400.
[2]	胡龙, 刘红艳, 成慧梅, 陈维奇, 冯广杰, 叶延洪, 邓德安. 超高强耐磨钢NM500多层多道对接接头残余应力的研究[J]. 机械工程学报, 2024, 60(4): 335-344.
[3]	宋军, 唐倩, 罗智超, 冯琪翔, 聂云飞, 任治好. 马氏体时效钢激光选区熔化成形过程介观尺度数值模拟[J]. 机械工程学报, 2024, 60(3): 282-295.
[4]	黄浩, 单忠德, 张丽娇, 孙正, 郭子桐, 刘检华, 金鹏. 异形截面复合材料构件成形及力学性能预测方法研究[J]. 机械工程学报, 2024, 60(2): 107-118.
[5]	胡成亮, 苗宏量, 曾凡, 赵震, 汤敏俊, 汤晓峰. 热成形条件下软磁材料的磁感应强度预测模型[J]. 机械工程学报, 2024, 60(2): 132-139,149.
[6]	樊丁, 李德全, 侯英杰, 黄健康, 冯毅. GMAW潜弧焊电弧-熔池耦合行为数值分析[J]. 机械工程学报, 2024, 60(2): 159-167.
[7]	王刚, 谷诤巍, 于歌, 李欣. 热成形工艺条件下7075-H18铝合金板材塑性流动行为的本构建模[J]. 机械工程学报, 2024, 60(2): 188-196.
[8]	刘建琴, 胡桂菘, 乔金丽, 郭晓. 基于双路神经网络的TBM实时性能预测研究[J]. 机械工程学报, 2024, 60(16): 43-53.
[9]	戴志远, 李田, 张卫华, 张继业. 不同海拔高度环境下高速列车气动特性研究[J]. 机械工程学报, 2024, 60(16): 291-305.
[10]	陈佳佳, 刘松炎, 杨勇, 袁冬冬, 张立勇, 傅玉灿, 钱宁. 纳米流体热管砂轮成型磨削钛合金换热性能评价[J]. 机械工程学报, 2024, 60(15): 407-419.
[11]	傅德彬, 李超艳, 陈四春, 徐晓明. 密闭空间内射流冲击波传递特性[J]. 机械工程学报, 2024, 60(14): 338-346.
[12]	田文卿, 蔡超, 郭瑞鹏, 史玉升. 热等静压近净成形数值模拟研究现状与展望[J]. 机械工程学报, 2024, 60(1): 13-26.
[13]	于正洋, 钟斌, 张传伟, 赵升吨. 304不锈钢棒料连续旋弯低应力精密下料断裂预测[J]. 机械工程学报, 2024, 60(1): 190-197.
[14]	陈鑫, 王佳宁, 杨立飞, 张冠宸. 6061-T6铝合金薄板剪切试件设计及动态剪切力学特性分析[J]. 机械工程学报, 2023, 59(4): 62-70.
[15]	孙瑶, 蔡路, 秦登, 李田, 张继业. 车窗开闭状态对双层列车车厢内火灾烟气特性的影响[J]. 机械工程学报, 2023, 59(4): 232-240.