改进希尔伯特-黄变换的滚动轴承故障诊断

Rolling Bearing Fault Diagnosis Based on Improved Hilbert-Huang Transform

针对希尔伯特-黄变换中经验模态分解方法存在的端点效应和虚假固有模态函数的问题,提出一种改进希尔伯特-黄变换方法并将此方法应用于滚动轴承故障诊断中。首先,利用最小二乘支持向量机和镜像延拓相结合的方法来抑制端点效应;其次,采用敏感固有模态函数选择算法选出反映故障特征的敏感固有模态函数;最后,利用敏感固有模态函数的包络谱进行故障诊断。通过仿真分析和应用实例可看出,该方法能够有效提取出滚动轴承故障信号的特征信息并准确诊断出引起滚动轴承的故障原因。

Empirical mode decomposition has boundary effect and false intrinsic mode function problem which effect its application. In this paper,an improved Hilbert-Huang transform method is put forward and the method is used in the rolling bearing fault diagnosis. a method to depress the boundary effect is developed bases on least squares support vector machine and mirror extension;the sensitive intrinsic mode function selection algorithm is used to select the sensitive intrinsic mode function which reflect fault feature;the envelope spectrum of sensitive intrinsic mode function is used to fault diagnosis. the simulation analysis and application example show that this method can effectively extract the feature information of rolling bearing fault signal and accurately diagnose the cause of the fault caused by rolling bearings.