轴承故障信号的平均组合差值形态滤波分析

Average combination difference morphological filter analysis on fault signals of rolling bearing

为了提取受谐波和随机噪声干扰的信号中的冲击故障特征,提出基于平均组合差值形态滤波(ACDIF)和Teager能量峭度(TEK)的滚动轴承故障诊断方法.将对冲击成分具有不同抑制方式的4种基本形态算子两两合并加强抑制效果,组合作差反向提取出正、负冲击,构造出一组新的组合差值形态算子(CDIF),通过比较分析选择其中2种CDIF的平均值作为最终滤波输出.针对滤波过程中结构元素(SE)的选择问题,采用TEK作为评价指标筛选最佳结构元素长度,有效提高了滤波处理的效率和精确度.数值仿真和轴承外圈故障振动信号的试验结果表明,利用该方法能够有效地滤除随机噪声和谐波干扰,提取强背景噪声下的冲击故障特征,滤波效果优于传统方法.

A new fault diagnosis method based on average combination difference morphological filter(ACDIF)and Teager energy kurtosis(TEK)was proposed in order to extract the impulse components embedded in vibration signals consisting of much noise and harmonics.The four basic operations with different suppression modes on impact components were combined in pairs for enhancing the effect of pulse suppression.Positive and negative impact was extracted in reverse by combination difference so that several new combination difference(CDIF)operations were constructed.Two CDIFs were selected by comparing the performance of all CDIFs,and the average of two selected CDIFs was regarded as final output.The length of SE was defined by a new criterion named TEK aiming at the selection problem of structure element(SE).Then the effectiveness of ACDIF for fault feature extraction was improved.The experimental results on simulation and bearing vibration signals demonstrate that the method can effectively suppress interference and extract periodic impacts from vibration signals.The comparison results indicate that the method outperforms other methods for bearing fault diagnosis.