含噪信号经验模式分解虚假分量识别方法

发布时间：2019-07-20 11:24

【摘要】：针对含噪信号Hilbert-Huang变换存在虚假分量,提出改进的奇异值分解(SVD)方法进行降噪,改进包含两个部分:一是利用重构相空间代替传统矩阵如Hankel矩阵,以去掉信号冗余,再者提出奇异值能量熵分量差分法,更易于定出重构奇异值阶次;二是提出了频谱比值法对虚假分量进行辨识,更有效辨识出虚假分量.首先利用经验模式分解(EMD)得到本征模式分量(IMF),识别并剔除趋势项,重构信号,然后进行SVD,重构降噪后的信号,消除虚假分量,最后进行时频分析.联合方法应用于含噪仿真信号,信噪比(signal noise ratio,SNR)提高了5.5%,虚假分量辨识率提高至100%,用于双跨转子故障振动信号,得到正确的时频结果,表明了所提方法识别含噪信号虚假分量的有效性.
[Abstract]:Aiming at the existence of false components in Hilbert-Huang transform of noisy signals, an improved singular value decomposition (SVD) (SVD) method is proposed to reduce noise. The improvement consists of two parts: one is to use reconstructed phase space instead of traditional matrix such as Hankel matrix to remove signal redundancy, and the singular value energy entropy component difference method is proposed to determine the order of reconstructed singular value. Second, the spectrum ratio method is proposed to identify the false components more effectively. Firstly, the intrinsic pattern component (IMF), is identified and the trend term is eliminated and the signal is reconstructed by empirical pattern decomposition (EMD). Then the signal after noise reduction is reconstructed by SVD, and the false components are eliminated. Finally, the time-frequency analysis is carried out. The joint method is applied to the noisy simulation signal, the signal-to-noise ratio (signal noise ratio,SNR) is increased by 5.5%, and the identification rate of the false component is increased to 100%. The method is applied to the fault vibration signal of the two-span rotor, and the correct time-frequency results are obtained, which shows the effectiveness of the proposed method in identifying the false component of the noisy signal.
【作者单位】：南京航空航天大学能源与动力学院;南京航空航天大学理学院;
【基金】：航空自然科学基金(2012ZD52054) 青年科学基金(61403193) 南京航空航天大学基本科研业务费科研项目(NS2014081)
【分类号】：TN911.7

【参考文献】