电能质量扰动的Block-Thresholding去噪方法

Block-Thresholding Approach for Power Quality Disturbance Denoising

提出一种基于block-thresholding阈值估计量的电能质量扰动小波去噪算法。在小波域,各个尺度携带信号信息的小波系数其分布具有"簇聚"性质,即大部分系数成簇聚集在信号突变位置。所提算法将各个尺度的小波系数分成若干块,针对各个块进行阈值处理;而不像传统的小波阈值去噪算法,如Donoho等提出的VisuShrink那样预先确定一个阈值,对所有小波系数逐项比较进行去留处理。将所提算法与传统阈值去噪方法进行比较研究,仿真和实验结果表明所提算法在全局适应性和空间适应性方面的优越性。

In this paper, a novel block-thresholding approach for power quality disturbances denoising is proposed. In wavelet domain, it shows that large detail coefficients come as groups; they cluster around the areas where the function changes significantly. Unlike the wavelet thresholding procedure such as Donoho＇s VisuShrink, which adopts a term-by-term threshold rule, the proposed approach estimates the ＂average＂ wavelet coefficient within a block, and making a decision about retaining or discarding it. The calculation and experiment results indicate that, in adaptivity and spatial adaptivity senses, the proposed approach performs better than term-by-term-based denoising algorithms.