基于ENLMD与Hilbert包络谱的轴承运行状态故障特征提取

Fault Feature Extraction of Bearing Operation Status Based on ENLMD and Hilbert Envelope Spectrum

针对局部均值分解(Local Mean Decomposition,LMD)方法实现过程中只部分抑制了端点效应和模态混叠的问题提出集成噪声重构局部均值分解(Ensemble Noise-reconstructed Local MeanDecomposition,ENLMD)方法。该方法对经过LMD分解后待处理的PF分量(Product Function)的集合进行多次重构采样和重构再分解得到最终平均PF分量;然后引入敏感评判指标选取有效的PF分量进行Hilbert包络谱,进而提取滚动轴承的故障特征,识别其运行状态;最后,利用模拟轴承故障仿真信号和凯斯西储大学滚动轴承测试集完成了ENLMD方法与LMD、ELMD等的对比试验。实验结果表明:ENLMD可取得比LMD、ELMD更好的效果,可用于滚动轴承运行状态的故障特征提取。

Ensemble Noise-reconstructed Local Mean Decomposition(ENLMD)solves the problem of only partially suppressing end effect and mode aliasing in Local Mean Decomposition(LMD).First,ENLMD decomposes the set of Product Function(PF)components to be processed after LMD decomposition by refactoring sampling and reconstruction and decomposition to get the final average PF components.Then the sensitive evaluation index is introduced to select the effective PF component for Hilbert envelope spectrum,and then the fault characteristics of rolling bearing are extracted to identify its operation state.Finally,ENLMD is compared with LMD,ELMD and so on by using simulated bearing fault simulation signal and Casey Reserve University rolling bearing test set.The experimental results show that ENLMD can achieve better results than LMD and ELMD,and can be used for fault feature extraction of rolling bearing.