Prediction of Flow Fields in Centrifugal Pumps Based on the Proper Orthogonal Decomposition-Long Short-Term Memory Network Model

XIAO Ying; XIAO Xiangyu; DUAN Zhuang; SUN Zhongguo; XI Guang

doi:10.7652/xjtuxb202412012

您当前的位置：

首页 >

文章列表页 >

Prediction of Flow Fields in Centrifugal Pumps Based on the Proper Orthogonal Decomposition-Long Short-Term Memory Network Model

更新时间：2025-07-28

- Prediction of Flow Fields in Centrifugal Pumps Based on the Proper Orthogonal Decomposition-Long Short-Term Memory Network Model
- Vol. 58, Issue 12, Pages: 119-130(2024)
- 作者机构：
  
  西安交通大学能源与动力工程学院,西安,710049
- 作者简介：
- 基金信息：
- DOI：10.7652/xjtuxb202412012
  CLC： TP391.9
- Online First：10 December 2024，
  
  Published：2024
- 稿件说明：
移动端阅览
XIAO Ying, XIAO Xiangyu, DUAN Zhuang, et al. Prediction of Flow Fields in Centrifugal Pumps Based on the Proper Orthogonal Decomposition-Long Short-Term Memory Network Model[J]. 2024, 58(12): 119-130.
DOI：

XIAO Ying, XIAO Xiangyu, DUAN Zhuang, et al. Prediction of Flow Fields in Centrifugal Pumps Based on the Proper Orthogonal Decomposition-Long Short-Term Memory Network Model[J]. 2024, 58(12): 119-130. DOI： 10.7652/xjtuxb202412012.

摘要

针对流体机械领域中瞬态流场预测的复杂性以及现有本征正交分解-径向基函数(POD-RBF)模型在时间依赖性预测方面的局限性

引入本征正交分解-长短期记忆网络(POD-LSTM)模型

以提升预测的准确性和效率。通过计算流体动力学(CFD)对二维圆柱绕流进行分析

比较了POD-RBF与POD-LSTM模型在瞬态流场预测性能上的差异。进一步将POD-LSTM模型应用于离心泵瞬态流场预测

详细分析了离心泵叶轮、蜗壳及密封装置的预测效果。计算结果表明:相较于POD-RBF模型

POD-LSTM模型在预测距离训练集较远时刻的流场时性能较优

预测精度较高

压力场的平均相对偏差仅为0.96%; 与传统CFD方法相比

POD-LSTM模型在预测离心泵压力场和y方向速度场时的平均相对偏差分别为0.06%、6.07%

计算时间仅为传统CFD方法的0.01%

显著降低了计算成本; POD-LSTM模型的预测结果与CFD模拟结果的一致度较高

验证了其在离心泵流场预测中的精准性。研究可为流体机械领域数字孪生体的构建提供新的技术路径。

Abstract

Given the complexity of transient flow field prediction in fluid mechanics and the limitations of the existing proper orthogonal decomposition-radial basis function(POD-RBF)model in time-dependent predictions

a proper orthogonal decomposition-long short-term memory network(POD-LSTM)model is introduced to enhance prediction accuracy and efficiency. Computational Fluid Dynamics(CFD)is employed to analyze the flow around a two-dimensional cylinder

comparing the performance of the POD-RBF and POD-LSTM models in predicting transient flow fields. Furthermore

the POD-LSTM model is applied to the transient flow field prediction of a centrifugal pump

with a detailed analysis of the prediction effects on the impeller

volute

and sealing device. The computational results indicate that

compared to the POD-RBF model

the POD-LSTM model performs better in predicting flow fields at times distant from the training set

achieving higher prediction accuracy with an average relative error of only 0.96% in the pressure field. In comparison to traditional CFD methods

the POD-LSTM model demonstrates an average relative error of 0.06% in predicting the pressure field and 6.07% in predicting the y-direction velocity field of the centrifugal pump

with a computation time of only 0.01% of that of traditional CFD methods

thus significantly reducing computational costs. The consistency between the prediction results of the POD-LSTM model and the CFD simulation results validates its accuracy in predicting the flow field of centrifugal pumps. This research presents a novel technical approach for developing digital twins in the field of fluid mechanics.

关键词

Keywords

references

庄存波, 刘检华, 熊辉, 等. 产品数字孪生体的内涵、体系结构及其发展趋势 [J]. 计算机集成制造系统, 2017, 23(4): 753-768.

ZHUANG Cunbo, LIU Jianhua, XIONG Hui, et al. Connotation, architecture and trends of product digital twin [J]. Computer Integrated Manufacturing Systems, 2017, 23(4): 753-768.

TUEGEL E J, INGRAFFEA A R, EASON T G, et al. Reengineering aircraft structural life prediction using a digital twin [J]. International Journal of Aerospace Engineering, 2011, 2011: 154798.

WANG Zhicun, YANG Shuchi, CHEN P C. Nonlinear gust reduced order modeling based on FUN3D and Volterra theory [C]//2018 AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference. Reston, VA, USA: AIAA, 2018: AIAA 2018-1211.

WARWICK G. GE advances analytical maintenance with digital twins [EB/OL].(2015-10-19)[2024-01-16]. https://aviationweek.com/special-topics-pages/optimizing-engines-through-lifecycle/ge-advances-analytical-maintenance.

李亚洁, 刘强, 李炜. 基于离心泵数字孪生流场云图的叶轮故障智能诊断方法研究及应用 [J/OL]. 北京航空航天大学学报.(2023-04-20)[2024-02-25]. https://doi.org/10.13700/j.bh.1001-5965.2022.0997.

LI Yajie, LIU Qiang, LI Wei. Research and application of intelligent diagnosis method for impeller faults based on digital twin flow field cloud map of centrifugal pump [J/OL]. Journal of Beijing University of Aeronautics and Astronautics.(2023-04-20)[2024-02-25]. https://doi.org/10.13700/j.bh.1001-5965.2022.0997.

王方, 甘甜, 王煜栋, 等. 航空发动机燃烧室数字孪生体系关键技术 [J]. 航空动力学报, 2023, 38(7): 1546-1560.

WANG Fang, GAN Tian, WANG Yudong, et al. Key technology of digital twin system for aero-engine combustors [J]. Journal of Aerospace Power, 2023, 38(7): 1546-1560.

YANG Jun, HUANG Yanping, WANG Dianle, et al. Fast prediction of compressor flow field in nuclear power system based on proper orthogonal decomposition and deep learning [J]. Frontiers in Energy Research, 2023, 11: 1163043.

田琳琳, 赵宁, 钟伟, 等. 基于小生境遗传算法的风电场布局优化 [J]. 南京航空航天大学学报, 2011, 43(5): 650-654.

TIAN Linlin, ZHAO Ning, ZHONG Wei, et al. Placement optimization of wind farm based on niche genetic algorithm [J]. Journal of Nanjing University of Aeronautics Astronautics, 2011, 43(5): 650-654.

FENTAYE A D, ZACCARIA V, KYPRIANIDIS K. Aircraft engine performance monitoring and diagnostics based on deep convolutional neural networks [J]. Machines, 2021, 9(12): 337.

张人会, 陈学炳, 郭广强, 等. 低比转数离心泵叶轮内流场重构与模态分析 [J]. 农业机械学报, 2018, 49(12): 143-149.

ZHANG Renhui, CHEN Xuebing, GUO Guangqiang, et al. Reconstruction and modal analysis for flow field of low specific speed centrifugal pump impeller [J]. Transactions of the Chinese Society for Agricultural Machinery, 2018, 49(12): 143-149.

CHEN Xuebing, ZHANG Renhui, JIANG Lijie, et al. Adaptive POD surrogate model method for centrifugal pump impeller flow field reconstruction based on clustering algorithm [J]. Modern Physics Letters: B, 2021, 35(7): 2150126.

张人会, 徐啸宇, 郭广强. 液环真空泵叶轮与壳体型线的协同优化 [J/OL]. 西华大学学报(自然科学版).(2024-01-04)[2024-03-25]. http://kns.cnki.net/kcms/detail/51.1686.N.20240103.1424.002.html.

ZHANG Renhui, XU Xiaoyu, GUO Guangqiang. Collaborative optimization of impeller and shell profile of liquid ring vacuum pump [J/OL]. Journal of Xihua University(Natural Science Edition).(2024-01-04)[2024-03-25]. http://kns.cnki.net/kcms/detail/51.1686.N.20240103.1424.002.html.

ZHOU Yang, JIANG Ming, YUAN Xiaolin, et al. Fault prediction of molecular pump based on DE-Bi-LSTM [J/OL]. Fusion Science and Technology.(2023-11-29)[2024-03-11]. https://doi.org/10.1080/15361055.2023.2275089.

WANG Chenyang, JIANG Wanlu, YUE Yi, et al. Research on prediction method of gear pump remaining useful life based on DCAE and Bi-LSTM [J]. Symmetry, 2022, 14(6): 1111.

DENG Zhiwen, CHEN Yujia, LIU Yingzheng, et al. Time-resolved turbulent velocity field reconstruction using a long short-term memory(LSTM)-based artificial intelligence framework [J]. Physics of Fluids, 2019, 31(7): 075108.

刘汉儒, 袁一放, 马岩, 等. 基于本征正交分解的串列叶栅多目标气动优化研究 [J]. 工程热物理学报, 2023, 44(6): 1546-1558.

LIU Hanru, YUAN Yifang, MA Yan, et al. Multi-objective aerodynamic optimization of tandem cascade based on proper orthogonal decomposition [J]. Journal of Engineering Thermophysics, 2023, 44(6): 1546-1558.

ARMELLINI A, CASARSA L, MUCIGNAT C. Flow field analysis inside a gas turbine trailing edge cooling channel under static and rotating conditions [J]. International Journal of Heat and Fluid Flow, 2011, 32(6): 1147-1159.

ZHOU Lei, WEN Jiahao, WANG Zhaokun, et al. High-fidelity wind turbine wake velocity prediction by surrogate model based on d-POD and LSTM [J]. Energy, 2023, 275: 127525.

LUMLEY JL. The structure of inhomogeneous turbulent flows [C]//Proceedings of the International Colloquium on the Fine Scale Structure of the Atmosphere and Its Influence on Radio Wave Propagation. Moscow, USSR: Nauka, 1967:166-178.

SIROVICH L. Turbulence and the dynamics of coherent structures: Ⅲ dynamics and scaling [J]. Quarterly of Applied Mathematics, 1987, 45(3): 583-590.

HOCHREITER S, SCHMIDHUBERJ. Long short-term memory [J]. Neural Computation, 1997, 9(8): 1735-1780.

SAK H, SENIOR A, BEAUFAYS F. Long short-term memory recurrent neural network architectures for large scale acoustic modeling [C]//Proc. Interspeech 2014. Florence, Italy: IEEE, 2014: 338-342.

GRAVES A. Supervised Sequence Labelling with Recurrent Neural Networks [M]. Berlin, Heidelberg: Springer, 2012.

GERS F A, SCHMIDHUBER J, CUMMINS F. Learning to forget: continual prediction with LSTM [J]. Neural Computation, 2000, 12(10): 2451-2471.

武频, 孙俊五, 封卫兵. 基于自编码器和LSTM的模型降阶方法 [J]. 空气动力学学报, 2021, 39(1): 73-81.

WU Pin, SUN Junwu, FENG Weibing. Reduced order model based on autoencoder and long short-term memory network [J]. Acta Aerodynamica Sinica, 2021, 39(1): 73-81.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Digital Twin-Driven Remote Operation and Maintenance Method for Electromechanical Equipment

Analysis on Transient Hydraulic Characteristics of Cavitation Process in Centrifugal Pumps

Quality Measurement Criteria and Optimization Algorithm of Tetrahedral Mesh for Centrifugal Pumps

Modification for Turbulence Model Considering Rotation and Curvature with Applications

Cavitation State Recognition of Centrifugal Pump Based on Deep Learning

Related Author

ZHANG Yingfeng

ZHANG Gen

LIU Xin

SUN Peilu

俞志君 2

付强 1

朱荣生 1

袁寿其 1

Related Institution

Department of Mechanical and Electrical Engineering, Yuncheng University

School of Mechatronic Engineering, Xi’an Technological University

Key Laboratory of Industrial Engineering and Intelligent Manufacturing of Ministry of Industry and Information Technology, Northwestern Polytechnical University

Research Center of Fluid Machinery Engineering and Technology, Jiangsu University

Zhijiang College, Zhejiang University of Technology

⁰