改进Autogram及其在滚动轴承故障诊断中的应用

An Improve Autogram Method and its Application to Fault Diagnosis of Rolling Bearing

针对自相关谱峭度(Autogram)诊断效果易受最大重叠离散小波包变换(MODWPT)预设分解层数影响的不足,本文提出一种参数自适应Autogram诊断方法。该方法将平均包络熵(MEE)最小值作为优化目标对MODWPT最佳分解层数进行搜寻,并以分解后节点平方包络自相关峭度的最大值来确定最优频带的中心频率及带宽,最后通过包络解调提取故障特征信息。研究结果表明,自适应的分解层数确定方法较好地改善了Autogram方法的故障诊断效果,该方法可以快速、准确地识别出滚动轴承的故障特征。

Aiming at the problem that the diagnostic effect of Autogram is easily affected by the decomposition level of maximum overlap discrete wavelet packet transform(MOWDPT), an improved adaptive Autogram method is proposed for of rolling bearing fault diagnosis. In this method, the minimum value of the mean envelope entropy(MEE) was firstly used to search the optimal decomposition level of MOWDPT, and then the central frequency and bandwidth of the optimal resonance band was selected by the maximum value of the kurtosis of the unbiased autocorreelation(AC) of the squared envelope of the decomposed signal, finally the fault feature information was extracted by demodulating. The research results show that the adaptive determination of decomposition level improves the fault diagnosis effect of the improved Autogram, and this method can quickly and accurately identify the fault feature of rolling bearing.