基于小波和分形分析的GIS局部放电信号特征提取

Signal Characteristic Extraction of Partial Discharge in GIS Based on Wavelet and Fractal Analysis

结合小波包分频和分形技术原理,提出一种气体绝缘金属开关装置(GIS)局部放电信号特征的提取方法。文中指出了小渡变换和分形理论相结合用于特征提取的有效性;求出不同故障类型局部放电信号的分维数,利用其大小进行特征提取;对于不能以此参数区分的放电信号,利用小波包对其进行频带分解,通过设定阈值找出不同故障类型与特征频带的分维数的相关性,从而实现故障特征提取。针对GIS局部放电的5种典型放电模型,通过试验获得大量放电样本数据,利用上述特征提取方法对其进行区分,获得了较好效果,结果表明利用小波包分形维数能够较好地进行局部放电信号特征提取。

By combining wavelet packet transform with fractal theory, a new method is put forward to extract characteristics of partial discharge （PD） signal in a gas insulated switchgear （GIS）. To begin with, the effectiveness of the new algorithm used to extract the characteristics of PD signal is described, which can be realized by the value of the fractal dimensions of PD among those types of insulation defect. However, as there are several types of insulation defect with similar values of fraetal dimensions, wavelet packet transform and fraetal analysis should be combined to identify the complicated situation. In accordance with the threshold value of each type of insulation defect in each frequency band, the correlation between the type of insulation defect and the fractal dimensions can be figured to perform extraction. Moreover, five different discharge models are designed to simulate partial discharges occurring inside GIS and a large number of discharge samples are obtained by the model tests. It is further proved that high extraction discriminative effect can be attained with the method proposed.