基于局部均值分解和切片双谱的滚动轴承故障诊断研究

Fault diagnosis of roller bearings based on local mean decomposition and slice bispectrum

针对滚动轴承故障诊断问题,提出一种结合局部均值分解(LMD)和切片双谱的诊断新方法。首先利用LMD算法对故障信号进行自适应分解,分解后获得一组位于不同频带的乘积函数(PF)分量,然后利用所提出的峭度准则对分解结果进行筛选,筛选出峭度值最大的PF分量,并对其包络信号做切片双谱分析,从而提取出故障特征频率信息。为加快分解速度、减少分解运算量,对LMD算法中的循环迭代结束条件做出改进,并利用模拟信号验证了LMD算法的信号分解能力以及切片双谱的噪声抑制和非二次相位耦合谐波剔除能力。最后,运用提出的诊断方法对实测轴承内圈、外圈故障振动信号进行分析,诊断效果良好,证明该方法具有一定的可靠性。

To effectively extract fault features of roller bearings, a new method based on local mean decomposition （LMD） and slice bispectrum was proposed. Original fault signals were decomposed into a series of product function （PF） components within different frequency bands, the decomposition results were screened with the proposed kurtosis criterion, then the PF component with the maximum kurtosis value was chosen, the fault feature type could be judged by analyzing the slice bispectrum computed for the envelope signal of the chosen PF component. In order to reduce the computation amount and accelerate the decomposition rate, the loop iteration ending condition of local mean decomposition was improved, then the decomposing capacity of LMD and the ability of noise suppression and eliminating the non-quadratic phase coupling harmonic components of slice bispectrum were verified with simulation signals. Bearing＇s inner ring and outer ring fault signals were analyzed with the proposed method and the good results showed that this method has a certain level of reliability.