【摘要】 应用信息论中的 MDL(Minimum Description Length)判据,针对动态电能质量扰动信号的压缩和消噪问题,提出了一种新的利用离散小波变换和局部余弦变换的数据压缩和消噪方法,该方法采用 MDL 判据结合压缩评价因子进行估计信号模型的选择。针对不同噪声水平和信号类型,该算法具有数据自适应能力,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分解系数的最佳个数,并能根据信号动态选择小波基或局部余弦基。算例结果表明该方法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。更多还原

【Abstract】 Adopting minimum description length (MDL) criterion, the authors proposed a novel data compression and noise reduction method for the compression and noise reduction of dynamic power quality disturbance signal by use of discrete wavelet transform and local cosine transform, in which the selection of estimated signal model was conducted by MDL criterion combined with compression efficiency index. For various noise level and types of signal, the proposed method possesses data adaptive ability, the optimal numbers of retained wavelet decomposition coefficients can be determined while not any transcendental parameter setting, e.g., threshold value setting, and subjective judgement are needed. In addition, according to the signals the wavelet bases or local cosine bases can be dynamically chosen. The results of calculation examples show that the proposed algorithm can properly tally the signal fidelity with signal compression efficiency.更多还原