改进EMD-小波分析的转子振动信号去噪方法

摘要

针对低信噪比转子振动信号的去噪问题，提出了EMD和小波分析相结合的去噪方法。首先对含噪信号进行EMD处理，利用连续均方误差准则对IMF分量进行高低频区分。然后设定不同的阈值，利用小波分析对高低频IMF分量进行有用信号的提取，将提取的信号和低频IMF分量叠加重构，实现降噪。为了验证所提方法的有效性，进行了数字仿真和转子振动信号降噪分析，结果表明，所提方法整体上优于EMD和小波阈值去噪方法。%A method based on EMD and wavelet analysis was proposed for denoising of low SNR rotor vibration sig-nals. Firstly, the noise signal was decomposed by EMD, and the IMF components were classified into high-frequency compo-nents and low-frequency components according to consecutive mean square error (CMSE) criterion. Then, different thresh-olds were determined and the useful signals of high and low frequency IMF components were extracted respectively based on the wavelet analysis. The extracted signals were superimposed with the low-frequency IMF components and reconstruct-ed so that the noise of IMF was removed efficiently. Numerical simulation and rotor vibration signal analysis were carried out to evaluate the performance of the proposed method. The results show that the proposed method is better than the individ-ual EMD or wavelet threshold denoising algorithm.