基于改进标准化最大似然估计的最小描述长度降噪方法

Minimum description length denoising based on improved normalized maximum likelihood

针对现有各种降噪方法存在的缺点,提出了一种改进标准化最大似然估计的最小描述长度降噪方法。该方法增加编码过程中对集合本身码长计算,降噪中自适应确定降噪阈值。通过仿真信号和实际某轴承故障信号降噪分析,结果表明所提方法可以有效消除噪声并尽可能保留有用信号成分,降噪后信号的信噪比高于VisuShrink降噪和BayesShrink降噪等方法。基于改进标准化最大似然估计的MDL降噪方法进一步完善了MDL降噪理论,提升了其降噪效果。

Aiming at defects of different denoising methods,a minimum description length （MDL ） denoising method based on improved normalized maximum likelihood （INML）was proposed.This method encodes the model class, adds to the original code length and adaptively defines the threshold in signal denoising.Numerical simulation signals and experimental signals of rolling element bearings were used to test and compare the performances of the proposed method with VisuShrink,BayesShrink and RNML etc.The results show that the INML-based noise cancellation method has more effective denoising performance and higher SNR.It can not only eliminate random noise,but also preserve interested components of signals.The INML method enriches the theory of MDL denoising and enhances its performance.