【摘要】 受电机所处工作环境中诸多因素的影响，轴承故障振动数据通常会混杂大量噪声，使得故障特征被无效噪声信息所淹没。为了将轴承故障冲击特征信息从含噪信号中提取出来，提出了一种CEEMDAN与改进形态差值滤波结合的故障诊断方法。在诊断初始阶段利用CEEMDAN对故障信号加以处理，得到相应的固有模态函数(IMF);用归一化互信息及峭度值作为评判标准，筛选所需的IMFs分量信号，并以此为基础完成信号重构；利用改进形态差值滤波实现对重构信号的去噪处理；求取处理后的信号频谱并加以探究，提取故障特征信息，完成对故障的有效诊断。由实例验证结果可知，该方法可在背景噪声干扰下对故障特征频率进行较好的定位，能够作为滚动轴承故障诊断的有效方法。更多还原

【Abstract】 Under the influence of many factors in the working environment of the motor, the bearing fault vibration data are usually mixed with a lot of noise, which makes the fault features overwhelmed by invalid noise information. In order to extract the bearing fault impact feature information from the noise-containing signals, a fault diagnosis method combining CEEMDAN and improved morphological difference filtering was proposed. In the initial stage of diagnosis, CEEMDAN was used to process the fault signal and obtain the corresponding intrinsic mode function(IMF). The normalized mutual information and the craggy value were used as the judging criteria to filter the required IMFs component signal and complete the reconstruction of the signal based on them. The improved morphological difference filtering was used to realize the denoising of the reconstructed signal. The processed signal spectrum was obtained and explored to extract the fault characteristic information and complete the effective diagnosis of the fault. Based on the example verification results, it could be seen that the method could be better located in the background noise interference on the fault characteristics of the frequency, can be used as an effective method of rolling bearing fault diagnosis.更多还原