基于单通道盲源分离算法的局部放电特高频信号去噪方法

Partial Discharge Ultra-High Frequency Signal De-Noising Method Based on Single-Channel Blind Source Separation Algorithm

为有效抑制现场检测局部放电特高频信号背景噪声中的周期性窄带干扰和高斯白噪声干扰,提出一种基于单通道盲源分离算法的去噪方法。首先对局部放电特高频信号进行时频联合分析,获得源信号数量;然后对检测到的局部放电信号进行奇异值分解,将得到的重构奇异值子矩阵重新组合成多通道信号,并采用特征矩阵近似联合对角化方法进行盲源分离,从局部放电信号中分离出噪声干扰;最后采用l_1范数最小化方法进行源信号估计,得出去噪后的局部放电特高频信号。使用该方法对模拟试验和现场实测信号进行去噪处理,并与现有方法的去噪结果进行对比。结果表明,与现有方法相比,该方法可更有效抑制周期性窄带和高斯白噪声干扰,且去噪后的局部放电特高频信号波形不发生明显畸变。

In order to effectively suppress the periodical narrowband and Gaussian white noise interference in the background noise of field test partial discharge(PD)ultra-high frequency(UHF)signal,a de-noising method based on single-channel blind source separation algorithm is proposed.Firstly,the PD UHF signal is time-frequency joint analyzed for obtaining the number of source signals.Secondly,by the singular value decomposition of PD signal,the calculated reconstructing singular value sub-matrices are recombine as multiple channel signals.Then,the backgrounds noises are separate multiple channel PD UHF signals by the joint approximate diagonalization of eigen-matrics blind source separation algorithm.At last,the source PD signal is estimated by the l1-norm minimization method,and the de-noised PD UHF signal is obtained.The de-noising method presented in this paper was applied on the laboratory and field measured signals,and the de-noising results were compared with other existing de-noising methods.The results show that the proposed de-noising method can suppress periodical narrowband and Gaussian white noise interference better compared with existing method.In addition,the de-noising PD UHF signal waveform is distorted unapparent.