基于VMD-SVD的单向阀微弱故障诊断方法

针对大型往复式机械高压隔膜泵单向阀振动信号中的微弱故障特征难以提取,且磨损击穿故障、卡阀故障及正常状态振动信号难以识别的问题,提出一种基于变分模态分解和奇异值分解的单向阀微弱故障特征提取及诊断方法。首先对振动信号进行VMD分解,再借助能量百分比和方差贡献率筛选出包含丰富故障信息的模态IMF分量,其次对筛选后的IMF分量构建初始特征矩阵,然后利用SVD对特征矩阵进行分解,得到特征矩阵的奇异值作为表征微弱故障信号的特征向量,最后应用多变量预测模型对单向阀微弱故障进行诊断。工程实验结果表明,该方法能有效地提取出单向阀微弱故障信号特征并能对单向阀状态类型进行有效识别。

Weak Fault Diagnosis Method for Check Valve Based on VMD-SVD

Aiming at the problem that the weak fault feature in the vibration signal of the check valve of the large reciprocating mechanical high-pressure diaphragm pump is difficult to extract,and the wear breakdown fault,the stuck valve fault,and the normal state vibration signal are difficult to identify,this paper proposes a weak fault feature extraction and diagnosis method for check valve based on variational mode decomposition(VMD)and singular value decomposition(SVD).Firstly,the vibration signal is decomposed by VMD,and then the energy percentage and variance contribution rate are used to select the intrinsic mode function(IMF)components with rich fault information.Secondly,the initial feature matrix is constructed for the selected IMF components.Thirdly,the feature matrix is decomposed by SVD to achieve the singular value,which is used as the feature vector to characterize the weak fault signal.Finally,the variable predictive model based class discrimination(VPMCD)is used to classify the weak wear breakdown fault,the weak stuck fault,and the normal state vibration signal of the check valve.The results of engineering experiments show that the proposed method can effectively extract the weak fault signal features of the check valve and identify its status type.