VMD奇异值和FCM的转子故障特征提取与识别

摘要

In order to extract characteristics of rotor faults accurately and effectively, the paper proposes variarional mode decomposirion(VMD)and extraction approaches for singular value features, and further adopts fuzzy c means clustering (FCM)to identify rotor faults. First, vibration signal decomposition is conducted by applying VMD algorithm of high decomposition precision and less problems in mode aliasing,from which an initial feature vector matrix is formed.And then fault feature vector is acquired by singular value decomposition to the initial feature vector matrix.Finally,form the clustering center by means of fuzzy C-means clustering, calculate simdilar approach degree to realize rotor fault classification under different working conditions. This approach is applied to validate rotor laboratory bench vibration data, the analysis results showed that the method can effectively distinguish the rotor fault signals under different working conditions and achieve the ideal fault diagnosis results.%为了准确、有效地提取转子故障特征,提出了变分模态分解(VMD)和奇异值特征提取的方法,并采用模糊C均值聚类(FCM)进行转子故障识别.首先,利用分解精度高、模态混叠问题少的VMD算法进行振动信号分解,形成初始特征向量矩阵,然后对该向量矩阵进行奇异值分解,将求得奇异值作为故障特征向量,最后通过模糊C均值聚类形成聚类中心,并计算海明贴近度以实现不同工况下的转子故障分类.将此方法进行转子实验台振动数据验证,实验结果表明:该方法能够有效实现不同工况下转子故障信号的区分,取得了理想的故障诊断结果.