基于改进型CEEMDAN和RBF神经网络的往复式压缩机阀片故障诊断研究.pdfVIP

机械工程师

MECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEER

MECHANICALENGINEER

基于改进型CEEMDAN和RBF神经网络的

往复式压缩机阀片故障诊断研究

谢雪松，谢路杨，都腾飞，李仲树，李泽军，刘新

（江苏长江智能制造研究院有限责任公司,江苏常州213164）

摘要：

利用改进的CEEMDAN和RBF神经网络相结合的方法识别压缩机气阀故障。首先利用改进型CEEMDAN对实验

采集的振动信号进分解，该方法克服了传统方法在处理模态混叠和虚假分量方面的不足，并提出基于改进型CEEMDAN

和奇异谱熵相结合的信号处理手段，对振动信号构造特征向量，最后将特征向量输入RBF神经网络进行模式识别训练。研

究结果表明：基于改进型CEEMDAN和奇异谱熵的信号处理方法可以有效地提取出往复式压缩机阀片故障的特征，而

RBF神经网络算法在对压缩机阀片故障进行模式识别时准确率高达99.58%。

关键词：

故障诊断；往复泵；信号处理；特征提取；模式识别

中图分类号：文献标志码：A文章编号：员园园圆原圆猿猿猿（圆园20）11原园090原园4

TP183

ResearchonFaultDiagnosisofReciprocatingCompressorValvePlateBasedonImprovedCEEMDANand

RBFNeuralNetwork

XIEXuesong,XIELuyang,DUTengfei,LIZhongshu,LIZejun,LIUXin

（JiangsuChangjiangIntelligentManufacturingResearchInstituteCo.,Ltd.,Changzhou2