基于EEMD和自相关函数峰态系数的轴承故障诊断方法

Fault diagnosis method for rolling bearings based on EEMD and autocorrelation function kurtosis

针对滚动轴承故障冲击信号周期性强且易被强烈的背景噪声所淹没的特点,提出了基于EEMD和自相关函数峰态系数的轴承故障诊断方法。首先,对采集到的复杂振动信号进行EEMD分解,根据自相关函数峰态系数和峭度准则重构IMF分量以突出故障特征信息;然后,利用谱峭度自动确定带通滤波器的最佳中心频率和带宽;最后,将滤波后的信号进行包络解调分析并与理论故障特征频率对比。通过轴承故障的仿真和实验研究,验证了该方法的有效性和可行性。

Considering that fault shock signals of rolling bearings have the features of periodicity and easily immerging in background noise, a fault diagnosis method based on the EEMD and autocorrelation function kurtosis was proposed. Bearing fault signal was decomposed by EEMD method, and according to the autocorrelation function kurtosis and the kurtosis criterion, the IMF components, which contain much more fault information, were chosen to reconstruct a new composite signal. By virtue of the spectral kurtosis analysis of the new composite signal, a band-pass filter was designed. The new composite signal was filtered by the band-pass filter, further envelope demodulated and then compared with the theoretical failure frequency. A case study on bearing faults simulations and experiments verifies the effectiveness and feasibility of the method proposed. © 2017, Editorial Office of Journal of Vibration and Shock. All right reserved.