一种基于形态提升的自适应轴承微冲击提取方法

Adaptive weak impulse extraction method of rolling bearings based on morphological lifting wavelet

在轴承早期故障中,冲击信号十分微弱,往往被背景噪声和低频谐波淹没,使得冲击信号提取有一定的难度。自适应多尺度形态滤波采用非线性滤波方法,通过不同尺度的结构元素,对冲击信号进行匹配并提取出来,但是由于微弱的冲击信号对结构元素尺度十分敏感,所以很难达到理想的提取效果。而利用形态梯度提升小波,首先将信号中的脉冲进行放大,降低低频信号和部分噪声的干扰,再对提升后的信号进行自适应多尺度形态滤波,这样就能显著地提取微弱的冲击信号,进而判定轴承发生的故障类型。仿真实验和轴承故障实例表明,该方法能有效提取背景噪声下的微弱冲击信号,是一种有效的微冲击提取方法。

In the incipient failure of rolling bearings, the impulse signal is extremely weak and difficult to extract from strong background noise. With the method of adaptive multi-scale morphological filtering, the impulse can be matched and extracted using variable scale of structural elements, but as the weak impulse is too sensitive to the scale, the extraction is often not ideal. Here, morphological lifting wavelet was introduced to remain the impulse and smooth the noise and low-frequency components, thus the impulse could be separated from background noise and then with adaptive multi-scale morphological filtering, the impulse could be extracted efficiently. Comparing with other two methods, the proposed method was proved to be effective for weak impulse extraction.