测控技术2025，Vol.44Issue(7)：26-34,9.DOI:10.19708/j.ckjs.2025.01.207
基于类不平衡学习的离心泵故障诊断研究
Centrifugal Pump Fault Diagnosis Based on Class Imbalance Learning
摘要
Abstract
The"class imbalance"problem exists between the fault data and normal data collected during the operation of rotating machinery,leading to a decrease in the accuracy of the data-driven fault diagnosis model.To address this problem,the centrifugal pump is taken as an object,and the accurate fault diagnosis of the cen-trifugal pump is realized by a"two-step"approach.Firstly,the high-quality expansion of fault samples of cen-trifugal pumps is realezed based on Wasserstein generative adversarial network with gradient penalty(WGAN-GP)model.Secondly,by using the convolutional neural network(CNN)method of deep learning,the fault diag-nosis model of centrifugal pump is designed,and three sets of centrifugal pump sample sets with different bal-anced ratios and balanced sample sets are constructed to complete the accurate fault diagnosis of centrifugal pump.The experimental results show that the sample set expanded by the WGAN-GP model has a positive ben-efit for centrifugal pump fault diagnosis and can effectively improve the accuracy of centrifugal pump fault diag-nosis.关键词
离心泵/类不平衡数据/故障诊断/生成对抗网络Key words
centrifugal pump/class-imbalanced data/fault diagnosis/GAN分类
信息技术与安全科学引用本文复制引用
陈志辉,曹思民,李耀武,赵雪岑,马剑,黄俊杰..基于类不平衡学习的离心泵故障诊断研究[J].测控技术,2025,44(7):26-34,9.基金项目
基础研究计划基金(113JCJQ2023114001) （113JCJQ2023114001）
