基于经验模态分解和小波阈值的冲击信号去噪

Denoising of Shock Signal Based on EMD and Wavelet Threshold

摘要:

冲击信号是非线性的并且容易受到噪声污染；为研究冲击信号去噪的问题,针对经验模态分解(Empirical Mode Decomposition,EMD)去噪和小波阈值去噪方法存在的不足,提出了基于EMD的小波阈值去噪方法；单纯的EMD去噪方法会在去除高频噪声的同时压制高频的有效信息；EMD与小波阈值去噪相结合,利用连续均方误差准则确定含噪较多的高频固有模态函数(Intrinsic Mode Function, IMF),对高频IMF分量进行小波阈值去噪,以分离并保留这些分量中的有效信息,同时保持低频IMF分量不变；对模拟数据和实际冲击信号进行去噪处理,结果表明,基于EMD的小波阈值去噪方法的去噪效果优于单纯的EMD去噪方法和小波阈值去噪方法。

Abstract:

Shock signal is easily interfered by noise. As Empirical Mode Decomposition(EMD) denoising method and wavelet threshold denoising method both have their disadvantages, a new denoising method is proposed when they are combined. When removing high-frequency noise, EMD denoising method also suppresses effective high-frequency information. In the proposed denoising method, continuous square error rule is used to define IMF components with higher frequency, and the wavelet threshold method is applied in those components. Meanwhile the other IMF components retains the same. Firstly the performance of proposed denoising method is validated by simulation, then the proposed method is applied in shock signal. The results show that the denoising effect of the proposed method is better than both single methods.