变压器局部放电监测逐层最优小波去噪算法

针对用于局部放电监测的去除白噪声算法会造成去噪脉冲信号波形畸变,脉冲幅值等波形参数产生较大误差,不利于进一步采用脉冲波形分析去除脉冲干扰的问题。为此根据局部放电信号在小波域上的分布特点,提出了各尺度信号分解和重构的最优小波选择方法,并给出了各尺度小波阈值的计算方法。仿真信号的最优小波去噪结果显示去噪信号具有波形畸变率低和幅值误差小的特点;实测信号的最优小波去噪结果证明提出的最优小波去噪算法能有效去除局部放电监测信号中的噪声,在局部放电在线监测应用中具有良好的去噪效果。

Optimum Wavelet De-noising Algorithm for Partial Discharge Online Monitoring of Transformers

White noise is a type of principle noise existing in detected signals of partial discharge(PD) online monitoring devices for electrical power transformers.De-noising algorithms,used already for partial discharge online monitoring,are possible to influence the waveform distortion of de-noised PD pulses.Errors may also happen on magnitude values of de-noised PD pulses.The PD pulse distortion and magnitude error show the disadvantages of these de-noising algorithms when the PD pulse analysis is used for removal of pulse-shaped noises.Therefore,an optimum wavelet selection approach was presented for selecting wavelets based on each scale for decomposition and reconstruction of signals,and based on the properties of PD signal and white noise appearing on each wavelet scale.The methods of threshold selection on different wavelet scales were also presented.The de-noising results of simulative signals show that the optimum wavelet de-noising approach is capable of obtaining the low waveform distortion and small magnitude errors of de-noised signals.The de-noised results of field-measured signals show that the optimum wavelet de-noising approach can remove noises existing in PD pulses effectively.The proposed method shows its strong de-noising capability in practical PD online monitoring systems.