一种改进的峭度图方法及其在复杂干扰下轴承故障诊断中的应用

摘要

快速峭度图是一种常用的滚动轴承故障诊断方法,但由于峭度指标对冲击过于敏感,在干扰较复杂的工况中,该方法往往无法正确识别出最优的共振频带进行包络解调.然而,解调信号的包络谱对噪声具有一定的免疫能力,而且包络谱中通常会清晰的出现故障特征频率及其倍频成分,呈现出典型的周期性脉冲特点.因此,提出应用相关峭度定量地刻画窄带信号的包络谱幅值,即以频域相关峭度值生成峭度图,用于最优频带的自适应地识别.同时,基于相关峭度的指向性,可以将该方法应用于轴承的复合故障诊断.最后通过实验分析,验证了该方法对轴承微弱故障和复合故障诊断的有效性.%Fast Kurtogram is one of the most useful methods in fault diagnosis of rolling element bearings.However,in some cases of complex interferences,it cannot exactly recognize the optimal resonance frequency band for envelope demodulation due to that the kurtosis index is too sensitive to impulsive noise.In fact,the envelope spectrum of demodulated signals in frequency domain has a certain immunity ability to noise,the bearing fault characteristic frequency and its harmonics often appear clearly with typical periodic impulse features in the envelope spectrum.Here,the frequency domain correlated kurtosis was proposed to quantitatively describe envelope spectrum amplitudes of narrow-band signals and generate the kurtogram.Moreover,the proposed method can be applied in the compound fault detection based on the directivity of correlated kurtosis.Two cases of real bearing fault signals were employed to verify the effectiveness and robustness of the proposed method in bearing weak fault diagnosis and compound fault diagnosis.