基于DWT的自适应电能质量信号压缩及消噪

Adaptive power quality signals compression and denoising based on DWT

应用信息论的AMDL(ApproximateMinimumDescriptionLength)判据,针对动态电能质量扰动(DPQD)信号压缩和消噪问题,提出了一种新的利用离散小波变换(DWT)的数据压缩和消噪算法。针对不同噪声水平和不同信号类型,这种算法具有数据自适应能力,其不需要进行任何先验的参数设置(例如阈值设置)和主观判断就能确定保留小波系数的最佳个数,并能根据信号动态选择小波基。计算实例表明MDL判据能够在信号保真度与信号压缩效率之间找到最佳的契合点。

This paper introduces a new data compression and denoising technique for short time voltage change signal via discrete wavelet transform (DWT) and information-theory based approximate minimum description length criterion (AMDL). This algorithm has data-adapted capability for different noise levels and signal sorts. Without any parameter setting (such as thresholding) or subjective judgments, it can select the 'best' number of wavelet-retained coefficients and 'best' wavelet basis according to dymamic signals. Results of computation prove that the MDL criterion can give the best compromise between the fidelity of the signal and the efficiency of signal compression.