基于广义形态学滤波和 Hilbert边际谱的滚动轴承故障诊断

摘要

Generalized morphological filter output could be good at eliminating the phenomenon of statistical bias,Hilbert marginal spectrum envelope method overcame the traditional needs to identify deficiencies bandpass filter center frequency and bandwidth.Combining the two methods mentioned, generalized morphological filter was used to complete the signal de-noising,then decomposing signals by EMD and then selecting the appropriate IMF components the partial Hilbert marginal spectrum of the signals was obtained.The bearing inner and outer ring fault diagnosis results show that the meth-od may accurately extract fault features,which determines the type and location of bearing failure ef-fectively,so it has wider applications in many fields.%广义形态滤波器可以很好地抑制输出统计偏倚的现象,Hilbert 边际谱克服了传统包络法需要确定带通滤波器的中心频率和带宽的不足,将两种方法相结合,首先利用广义形态滤波对信号进行去噪,在此基础上对信号进行经验模态分解,然后选取合适的 IMF分量得到信号的局部 Hilbert 边际谱。通过对轴承内外环进行故障诊断发现,该方法能准确地提取故障特征,从而有效地判别轴承的故障类型和部位,具有较广阔的应用前景。