基于CEEMD与SVM的离心泵转子不对中故障诊断方法研究  

Research on fault diagnosis method of centrifugal pump rotor misalignment based on CEEMD and SVM

作　　者：肖幸鑫 宋礼威 张翊勋 董亮[1] 张宇航 XIAO Xingxin;SONG Liwei;ZHANG Yixun;DONG Liang;ZHANG Yuhang(Research Center of Fluid Machinery Engineering and Technology,Jiangsu University,Zhenjiang 212000,China;State Key Laboratory of Nuclear Power Monitoring Technology and Equipment,CGN Engineering Co.,Ltd.,Shenzhen 518172,China)

机构地区：[1]江苏大学流体机械工程技术研究中心,江苏镇江212000 [2]中广核工程有限公司核电监控技术与装备国家重点实验室,广东深圳518172

出　　处：《流体机械》2022年第7期85-92,共8页Fluid Machinery

基　　金：国家自然科学基金项目(51879122,51779108,51779106);镇江市重点研发计划项目(GY2017001,GY2018025);过程装备与控制工程四川省高校重点实验室开放基金项目(GK201614,GK201816);江苏高校优势学科建设工程项目(PAPD);江苏省“六大人才高峰”高层次人才项目(GBZB-017)。

摘　　要：为了更好地判断离心泵转子不对中故障,通过互补经验模态分解(CEEMD)结合支持向量机(SVM)对转子不对中故障进行识别,搭建离心泵故障模拟实验台,利用电涡流振动位移传感器采集离心泵转子位移信号,使用CEEMD算法分解离心泵在正常状态与故障状态下信号,通过相关系数法和阈值,选取有效内涵模态分量(IMF)分量进行信号重构,计算重构信号的时域特征参数并组成特征向量,最后利用SVM对故障进行识别分类。结果表明,采用CEEMD方法可以有效提取出离心泵转子不对中时的故障特征。采用SVM方法对重构后的信号提取的特征向量进行训练,故障识别准确率可以达到93%,能够有效识别离心泵转子不对中故障。In order to better judge the occurrence of rotor misallocations of centrifugal pump,complete ensemble empirical mode decomposition(CEEMD)and support vector machine(SVM)were used to identify rotor misalignment.By setting up a centrifugal pump fault simulation experimental platform,eddy current vibration displacement sensor was used to collect the centrifugal pump rotor displacement signals;CEEMD algorithm was used to decompose the centrifugal pump signals in normal state and fault state;and the Intrinsic Mode Functions(IMF)were selected to reconstruct the signals through the correlation coefficient method and threshold value;the time domain characteristic parameters of reconstructed signals were calculated and were formed to characteristic vectors.Finally,the faults were identified and classified by SVM.The results show that the CEEMD method can effectively extract the fault characteristics of the centrifugal pump when the rotor is misaligned.SVM method is used to train the characteristic vector extracted from the reconstructed signals,and the fault identification accuracy can reach 93%,which can effectively identify the occurrence of rotor misalignment fault of centrifugal pump.

关 键 词：离心泵 转子不对中 CEEMD 时域特征参数 SVM 故障诊断 

分 类 号：TH3[机械工程—机械制造及自动化] TH17