基于改进的EEMD实现地震信号去噪

Seismic Signal De-noising Based on Improved EEMD

EEMD(Ensemble Empirical Mode Decomposition)是EMD的一种新扩展,为克服EMD本身存在的模态混叠而产生。传统EEMD去噪通常是将由噪声引起的IMF(Intrinsic Mode Function)直接舍掉,然后利用剩余的IMF重构信号,然而这种作法仅仅是压制了部分噪声,去噪效果不够理想。基于此本文提出一种改进的EEMD随机噪声消除方法,该方法结合了平移不变量小波阈值滤波压制高频噪声的优点以及Savitzky-Golay滤波去除低频噪声的优势。结果表明该方法不仅可以很好地衰减随机噪声,有效提高地震资料的信噪比,而且能够较好地保持有效信号。

The EEMD is a new extension of EMD,which can overcome the disadvantage of mode mixing of EMD.The de-noising of conventional EEMD is to abandon directly IFM caused by noise,and remaining IMF is used to restructure the signal.However,this method suppresses only the partial noise and the de-noising effect is not ideal.An improved noise elimination method based on EEMD is proposed in this paper.This method combines the translation invariant wavelet thresholding filter which can suppresses high frequency noise and Savitzky-Golay filter which can remove low frequency noise effectively.The processing results of synthetic seismic record and real seismic data show that the noise attenuation can not only attenuate the random noise nearly and sufficiently improve signal to noise ratio in seismic data but also preserve well effective signal compared with traditional EEMD.