基于MODWPT平方包络峭度谱的轴承声信号故障诊断方法

Application of MODWPT Square Envelope Kurtosis Spectrum in Fault Diagnosis of the Rolling Bearing Acoustic Signal

针对噪声干扰条件下的轴承声信号故障诊断问题,可以通过基于最大重叠离散小波包变换(MODWPT)的平方包络峭度谱法对轴承进行故障诊断。该方法首先对原始非平稳信号用MODWPT分解为若干个子频带分量之和,再对各子频带分量做平方包络峭度谱,快速定位原始非平稳信号当中冲击成分显著的频带范围,最后对目标频带做带通滤波并进行包络解调可得到故障诊断结果。通过实测轴承声信号数据验证,该方法可以有效地对轴承进行故障诊断。

In order to diagnose the Rolling Bearing under complex track-side noise environment,the original non-stationary signal can be decomposed into the sum of several sub-band components by maximum overlap discrete wavelet packet transform(MODWPT),and then square envelope Kurtosis spectrum is applied to each sub-band component,the frequency range of the fault impulse signal in the non-stationary signal is located quickly.Finally,the target frequency band is subjected to band-pass filtering and envelope demodulation to obtain the fault diagnosis result.Through the real data verification,the method can accurately locate the resonance frequency band of signal of the Rolling Bearing,and give the accurate fault information.