联合同步挤压小波变换和多尺度排列熵的局部放电类型识别

Partial discharge recognition based on synchrosqueezing wavelet transform and multi-scale permutation entropy

摘要:

为稳定提取变压器局部放电信号的特征,提出一种基于同步挤压小波变换和多尺度排列熵的局部放电特征提取方法,再通过ＧＫ模糊聚类方法对局部放电信号的特征进行识别分类。首先,通过同步挤压小波变换对4种典型变压器故障产生的局部放电信号进行分解,将其分解为一组含有局部放电特征信息的模态分量；然后,通过多尺度排列熵量化各模态分量的局部放电特征信息,使用各模态分量多尺度排列熵的平均值作为识别特征向量；最后,利用模糊聚类得到的局部放电样本标准聚类中心,采用欧式贴近度进行局部放电识别分类。将提出的方法应用于变压器局部放电的实验数据上,并与基于小波分解方法和经验模态分解的识别方法进行对比分析,实验结果表明,所提出的方法具有更好的分类性,对变压器局部放电分类具有更高的识别精度，平均识别精度达到93.60%。

Abstract:

In order to extraction of stabilize the characteristics of partial discharge signal of transformer, a method of partial discharge feature extraction based on synchrosqueezing wavelet transform(SWT) and Multi-scale permutation entropy is proposed, and then the characteristics of partial discharge signal are identified and classified by GK fuzzy clustering method. Firstly, the partial discharge signals generated by four typical Transformer faults are decomposed into a set of modal components containing the characteristic information of local discharge. Then, the partial discharge characteristic information of each modal component is quantified by multi-scale permutation entropy, and the average of the multi-scale permutation entropy of each modal component is used as the identification feature vector. Finally, using the standard cluster center of partial discharge samples obtained by GK fuzzy clustering, European proximity was used to classify partial discharge. The proposed method is applied to the experimental data of transformer partial discharge and compared with the recognition method based on wavelet decomposition and empirical modal decomposition. The experimental results show that the proposed method has better classification, and the classification of transformer partial discharge has higher recognition accuracy, the average recognition accuracy is 93.60%.