基于信息熵优化变分模态分解的滚动轴承故障特征提取

Bearing fault feature extraction based on VMD optimized with information entropy

针对变分模态分解(Variational Mode Decomposition,VMD)的参数需事先人为确定的问题以及如何选取包含故障特征信息的本征模态分量(Intrinsic Mode Function,IMF)的问题,提出了基于信息熵的参数确定方法和基于信息熵的IMF选取方法。该方法首先对原始故障信号进行变分模态分解,通过信息熵最小值原则对其参数进行优化,获得既定的若干IMF分量;在优化参数时获得信息熵最小值所在的IMF,选取其为有效IMF分量进行包络解调分析,提取轴承故障特征频率。通过轴承仿真信号和实际数据分析,表明该方法能够提取滚动轴承早期故障信号的微弱特征,并实现故障的准确判别。

Aiming at problems of the variational mode decomposition(VMD)’s parameters needing to be determined in advance and how to choose intrinsic mode functions(IMFs)containing fault feature information,a method to determine VMD parameters and the other one to choose IMFs both based on information entropy were proposed.With these methods,the original fault signal was decomposed using VMD,and the VMD parameters were optimized with the minimum information entropy principle to obtain several IMFs.The IMF corresponding to the minimum information entropy was chosen as the effective IMF to do the envelope demodulation analysis,and extract a bearing fault feature frequency.Through analyzing bearing simulated signals and actual signals,it was shown that the proposed methods can be used to effectively extract weak features of the early fault signal of a rolling bearing,and realize accurate fault diagnosis.