小波阈值去噪的改进方法

利用小波方法去噪,是小波应用于工程实践一个重要的方面;本文指出了采用常见的阈值确定方法对含有较强高频分量的信号进行小波阈值去噪时,去噪效果与实时性存在着矛盾,不利于信号特征的提取。提出了一种新型的阈值确定方法,进一步提出了以小波去噪后特征频率分量的功率谱密度值的下降程度为依据的待定因子确定方法。以Mallat算法为例,对新型阈值确定方法进行了分析和比较,指出运用该方法进行小波去噪处理,增强了算法的实时性,同时信号拟关注的信息损失较小,去噪效果比较理想。交-交变频器输出电流的小波去噪实验结果表明,该方法行之有效。

A Modified Method of Wavelet Threshold De-noising

Using wavelet de-noising is an important application of wavelet analysis in engineering. When the usual methods for getting thresholding are used for disposing the signal that contains high frequency component,the conflict exists between the denoising effect and the real time in the wavelet denoising process. This makes against the feature extraction of signal. A novel method for getting thresholding is put forward. Furthermore,in the wavelet denoising process the decreasing degree of the power spectrum density of the feature frequency component acted as the basis,a method for getting the factor is also put forward. The Mallat algorithm as the example,the novel method is analyzed. In the wavelet denoising process using this method,the real time of the algorithm is increased,and the attention information of the signal is less lost and the wavelet denoising effect is ideal. The wavelet denoising test results of the output current of the cycloconverter show that the above methods are effective.