基于小波混合阈值方法的电能质量信号去噪

基于小波变换的阈值去噪是滤除采样信号中白噪声的有效方法.在实际应用中,阈值的选取和对小波系数的处理方法是影响其去噪效果的2个重要因素.软硬阈值方法各有其优缺点.将两者结合起来的混合阈值方法可提高去噪算法的性能.在分析软硬阈值方法各自的优缺点后,结合正态分布的规律和信号、噪声的小波系数在不同层上的分布特点,提出了能自适应确定临界尺度的软硬阈值结合的混合阈值去噪方法.对采样信号进行小波变换,根据各层小波系数能量的最小值,确定临界尺度.对小于或等于临界尺度的层数上的小波系数用软阈值方法处理,对大于临界尺度的层数上的小波系数用硬阈值方法处理,这样处理能减小信号能量损失.对尺度系数和处理后的小波系数进行重构,得到去噪后的信号.仿真实验表明.混合阈值方法去噪性能稳定,在不同信噪比下都能有效地去除白噪声,较好地保留了原信号的局部特征,且方法简单,计算量小.

Power quality signal denoising based on wavelet hybrid threshold method

The threshold denoising based on wavelet transform is an effective method for eliminating white noise from sampled signal,of which the setting of threshold and the processing of wavelet coefficients are two important affecting factors. The combination of soft and hard threshold methods can improve denoising performance. According to the distribution of signal and noise wavelet coefficients at different levels,a hybrid threshold denoising method is proposed,which combines soft and hard thresholds and adaptively determines the critical scale based on the normally distributed rule. The sampled signals are processed by wavelet transform and the critical scale is then determined according to the minimum energy values of wavelet coefficients at different levels. To reduce the loss of signal energy,the wavelet coefficients at and below the critical level are processed with the soft threshold method and the wavelet coefficients above the critical level are processed with the hard threshold method. The denoised signal is obtained by reconstructing the level coefficients and the processed wavelet coefficients. Simulation results show that,the simple hybrid threshold method has stable denoising performance with less calculation load and effectively eliminates white noise under different SNRs,remaining the local characteristics of original signals.