基于经验小波变换的变压器振动信号特征提取

为有效提取变压器振动信号特征,提出了一种基于经验小波变换(EWT)的信号特征提取方法。首先利用EWT方法将不同工况的变压器振动信号分别分解为若干经验小波函数(EWF)分量;然后计算各分量Hilbert谱,通过时频表示直观反映不同工况变压器振动信号的频率特征信息;最后计算不同工况振动信号各EWF分量与原信号的相关系数,并提取相关度高的分量,根据其能量构建信号的特征矢量,实现对不同工况变压器振动信号特征提取的量化处理。仿真试验表明,该方法能有效提取变压器振动信号特征,且根据提取的特征矢量能够正确识别变压器绕组所属的不同工况。

Feature Extraction for Vibration Signals of Power Transformer Based on Empirical Wavelet Transform

In order to realize effective feature extraction for vibration signals of power transformers, a method of signal feature extraction based on empirical wavelet transform(EWT)is proposed. Firstly, transformer vibration signals in different working conditions are decomposed into several empirical wavelet functions(EWFs)through the method of EWT. Secondly, the Hilbert spectrum of each EWF is calculated, and the frequency characteristics of transformer vibration signals in different working conditions are shown in time-frequency representation. Finally, the correlation coefficient of each EWF and the original signal is calculated to extract components of high correlation. The eigenvectors of signals are built according to the energy of components above to quantize the features of transformer vibration signals. It is shown by experiment that this method has a good effect on feature extraction for vibration signals of power transformers, and the different transformer winding conditions can be recognized correctly through the extracted eigenvectors.