基于互相关的有效奇异值消噪方法-计算机工程与应用.PDF

Computer Engineering and Applications 计算机工程与应用 2015 ，51（15） 221 基于互相关的有效奇异值消噪方法 潘峥嵘，谯自健，张 宁，戴 芮 PAN Zhengrong, QIAO Zijian, ZHANG Ning, DAI Rui 兰州理工大学 电气与信息工程学院，兰州 730050 College of Electrical and Information Engineering, Lanzhou University of Technology, Lanzhou 730050, China PAN Zhengrong, QIAO Zijian, ZHANG Ning, et al. Improved singular value de-noising method based on cross- correlation. Computer Engineering and Applications, 2015, 51 （15）：221-224. Abstract: In order to improve the Signal to Noise Ratio （SNR）, aiming at the problem how to determine the effective reconstructing number of singular values to de-noise, this method of singular value decomposition based on cross-correlation is proposed. These signals are obtained by reconstruction of every single singular value and cross-correlation coefficients are calculated to determine the effective singular values to reconstruct original signal. This method is validated by the signal including the different frequency components and compared with the similar methods, the results show that this method is not only simple but also better. Key words: Signal to Noise Ratio（SNR）; singular value; de-noising; cross-correlation 摘 要：为了提高信号的信噪比，针对奇异值分解消噪方法中用于重构的有效奇异值个数难以确定问题，提出了一 种基于互相关的有效奇异值选取方法，进而达到消噪的目的。该方法利用单个奇异值重构得到的分量信号，求互相 关，从而确定用于重构的有效奇异值个数。通过采用含有不同频率成分的带噪信号对该方法进行验证，并与同类奇 异值选择方法进行对比，结果表明，该方法不仅简单易行而且效果较好。 关键词：信噪比；奇异值；消噪；互相关 文献标志码：A 中图分类号：TP206.3 ；TP391.9 doi ：10.3778/j.issn. 1002-8331.1312-0212 1 引言 量变化较平缓的信号无法准确确定用于重构的奇异值； [1] 文献[13]提出一种基于特征均值的SVD 信号消噪算法， 信号消噪一直是信号处理和故障诊断 中一个非常 [2] 将小于特征均值的奇异值全部归零达到消噪目的，该方 重要的步骤 。奇异值分解（Singular Value Decomposi- [3-4] 法会丢失大量有用信号或者消噪效果不佳；文献[14]提 tion ，SVD）具有良好的代数和几何不变性 ，其分解的 [5] 出一种奇异值差分谱理论能有效消