基于经验模态分解的自适应滤波算法在局部放电窄带干扰抑制中的应用

Application of adaptive filtering algorithm based on empirical mode decomposition to suppress DSI in PD detection

自适应滤波算法是当前抑制窄带干扰的有效方法。对于单频率的窄带干扰,设置自适应滤波器的参数比较容易,但是对于局放监测中多个频率且频率范围很宽的窄带干扰,设置自适应滤波的参数就会变得很困难。根据经验模态分解EMD(Emp iricalMode Decomposition)的分频特性,将EMD引入自适应滤波算法,提出了一种基于EMD的自适应滤波算法。局放信号中多个频率的窄带干扰经EMD分解之后,会分解到不同的模态函数中,从而将多频率的窄带干扰转化成了多个单频率的窄带干扰,在此基础之上对固有模态函数进行自适应滤波,可以较容易地解决自适应滤波器参数设置的问题,并能获得比普通自适应滤波更好的效果。仿真及实际数据的处理验证了该算法的有效性。

Adaptive filter is an effective method for suppressing Discrete Spectrum Interference （DSI）. For DSI with single frequency, it is easy to set the parameters of the filter. However, it will become difficult when faces with DSI of multi-frequency, which often occurs in partial discharge monitoring systems. Based on the frequency splitting characteristics of Empirical Mode Decomposition （ EMD）, this paper incorporates it into the adaptive filtering algorithm and a novel algorithm was proposed. After EMD of the noisy signals, DSIs of different main frequency could be decomposed into different Intrinsic Mode Functions （IMF）, and then multi-frequency DSI would turn to some single-frequency DSIs. Based on this transformation, problems about settings of adaptive filter can he solved. Results obtained from simulation signals and on-site data confirm its validity.