基于多尺度形态分解谱熵的电机轴承预测特征提取及退化状态评估

Motor bearing forecast feature extracting and degradation status identification based on multi-scale morphological decomposition spectral entropy

由于预测特征提取与退化状态评估直接关系故障预测可信性,结合数学形态学与信息熵理论,针对电机滚动轴承,提出基于多尺度形态分解谱熵的预测特征提取方法,用灰色关联分析对退化状态进行评估。对不同损伤程度轴承振动信号进行多尺度形态分解,分别计算其在不同尺度域内的复杂性度量能谱熵、奇异谱熵,以其作为预测特征向量。建立标准退化模式矩阵,对待检测样本信号特征向量与标准模式进行灰色关联分析,据关联度大小对样本信号退化状态进行评估。并仿真与实例数据验证该方法对电机轴承退化状态评估的有效性。

How to extract forecast features and distinguish degeneration status is the key problem in fault forecasting and it affects the reliability of fault forecasting. Combining mathematical morphology and information entropy, a method for motor bearing forecast feature extraction based on multi scale morphological decomposition spectral entropy was proposed here. On this basis, the degeneration status for bearing combining was estimated with the gray relational analysis. The multi-scale morphological decomposition was performed for bearing vibration signals with different damage levels, its complexity indicators of power spectral entropy and singular spectral entropy in different scale domains were computed. The two indicators were taken as a forecasting character-vector. Then, the standard degeneration mode matrix was built. The grey relational analysis between the character-vector and standard mode of the sample signal to be detected was conducted to distinguish the degenerate status of the sample signal according to the value of grey relevancy. The effectiveness of this proposed method for motor bearing degeneration status identification was verified via simulations and actural example data.