基于全映射复合多尺度散布熵的滚动轴承故障诊断

Fault Diagnosis for Rolling Bearings Based on Full-Mapping Composite Multiscale Dispersion Entropy

为有效提取滚动轴承振动数据中的非平稳故障特征,将复合多尺度散布熵(CMDE)中的不同映射方式进行集成,形成了一种新的测量轴承振动信号复杂度和自相似度的方法,即全映射复合多尺度散布熵(FCMDE)。在此基础上,提出了基于FCMDE和k近邻(KNN)的滚动轴承故障诊断方法,利用FCMDE计算轴承振动信号的熵值并提取轴承的故障特征,将高维故障特征输入KNN分类器中进行滚动轴承的故障识别,采用西储大学和江南大学轴承数据集的验证结果表明,FCMDE方法能够有效识别滚动轴承的故障类型,准确率分别达到了100%和95.83%。

In order to effectively extract the non-stationary fault features from vibration data of rolling bearings,different mapping methods in composite multiscale dispersion entropy(CMDE)are integrated,and a new method is proposed for measuring the complexity and self-similarity of bearing vibration signals,which is called full-mapping composite multiscale dispersion entropy(FCMDE).On this basis,a fault diagnosis method for the bearings is proposed based on FCMDE and k-nearest neighbor(KNN),FCMDE is used to calculate the entropy of bearing vibration signals,the fault features of the bearings are extracted,and the high-dimensional fault features are input into KNN classifier to identify the fault of the bearings.The validation results using bearing datasets from Western Reserve University and Jiangnan University indicate that the FCMDE method can effectively identify the fault types of the bearings,and the accuracy rates reach 100%and 95.83%respectively.