中文摘要：

针对电机性能退化状态分析中性能退化特征难以有效提取且不能准确反映性能退化趋势的问题,通过电机失效机理分析,得到电机振动信号与性能退化状态的对应关系,给出电机振动信号的测量方案,提出一种基于电机振动信号分析的电机性能退化特征提取方法.上述方法采用基于Hilbert-Huang变换的边际能量谱分析来获取振动能量在整个频域上的大小和分布,根据电机自身性能的退化,在支持向量机（SVM）方法的基础上将两分类中的各个频率能量看作两个相互独立分布未知的总体,提出一种新的权重贡献分配法对频域能量进行自学习提取,最后利用提取的频域能量建立能够表征与反映电机长期性能变化趋势的振动特征频率能量作为性能退化特征参数,从而解决性能退化特征的有效提取问题.试验结果分析表明,利用从振动信号中提取的退化特征确实能够反映电机的性能状态退化,并且采用所提出的由特征频率能量所建立的性能退化参数相比全频域能量能准确地反映电机性能状态退化趋势,从而解决了准确反映性能退化趋势的问题,利用提出的电机性能退化特征提取方法建立的性能退化参数能够对电机长期退化中的性能状态进行分析及预测.

英文摘要：

According to the result of failure mechanism analysis of electric motor, the correspondence between the vibration signal and performance degraded failure of electric motor was found, and a degradation characteristics ex traction method of rotary mechanical properties by using a combination of Hilbert Huang Transform （HHT） and Support Vector Machine （SVM） was proposed based on the vibration signals of electric motor to solve the problem that it is difficult to effectively extract degradation characteristics for electric motor performance degradation status a nalysis. First, the size and distribution of vibration energy in the whole frequency domain were obtained by the mar ginal energy spectrum analysis based on Hilbert Huang transforms. Then, according to the degradation of rotary me chanical properties, self learning extraction of the frequency domain energy was conducted by using the weight con tribution allocation method based on support vector machine. Finally, the frequency domain energy which can charac terize and reflect the change trend of long term performance of electric motor vibration characteristics was estab lished as degradation characteristic parameters by using frequency domain energy extraction. Example analysis shows that the degradation parameters established by characteristics frequency energy extraction can more effectively reflect the state of electric motor performance degradation compared with the whole frequency domain energy.