形态滤波器用于滚动轴承故障信号的特征提取

Morphological Filters in Feature Extraction for Rolling Bearing Defect Signals

以滚动轴承故障振动信号为例,在讨论形态滤波算子和形态结构元素的基础上,设计了多尺度混合形态滤波器来滤除信号中的噪声及无关的谐波成分。对滤波后的信号再进行多尺度形态闭开的差值滤波处理,得到了明显的周期性故障冲击特征。仿真数据和实验信号的分析表明,设计合适的形态滤波器既可以抑制噪声,又能够提取故障特征信息。形态滤波器算法构造灵活,计算简单,适于故障信号的在线处理和特征分析。

The vibration signals with defects in rolling bearing were measured and the morphological filters were presented that conclude morphological operators and structuring elements. The multi-scale morphological filters with mixed operators were designed to remove the noises and harmonic components. The morphological difference filter with closing and opening operation were constructed to extract the periodical impulses. Then the bearing defect characteristic information was very obvious by means of the frequency spectrum analysis. The simulated data and the experimental signals were processed and the conclusion indicates that the morphological filter with proper operators and structuring elements is ideal for smoothing noises and extracting useful features. The morphological filters are suitable for signal processing and defect diagnosis on-line.