基于噪声辅助局部波动特征分解的齿轮裂纹故障定量诊断方法

摘要

将齿轮故障机理研究与故障诊断方法相结合，提出了一种新的基于噪声辅助局部波动特征分解的齿轮裂纹定量诊断方法。首先，建立了带裂纹齿轮系统动力学模型，获得不同裂纹程度下的动力学响应，从中提取对故障敏感而与工况无关的特征参数构成特征向量矩阵，输入到支持向量回归机中，建立了特征参数与齿轮裂纹程度之间的映射关系。然后，通过这种映射关系，对实际的齿轮裂纹信号采用噪声辅助LOD方法进行特征提取，实现了齿轮裂纹的定量诊断。%A new method of quantitative diagnosis of gear crack faults was proposed based on noise-assisted LOD which combined gear fault mechanism with fault diagnosis methods.Firstly,the feature vector matrixes were constituted by feature parameters which sensitively and independent of working conditions might be extracted from a dynamics model established to obtain the dynamics re-sponses under different levels of cracks and input into the support vector regression(SVR),thus the mapping relationship among feature parameters and levels of gear cracks was created.Based on map-ping relationship,the quantitative diagnosis of gear cracks was realized by using noise-assisted LOD extracting feature parameters from actual gear crack signals.