基于EMD与ICA的地震信号去噪技术研究

Random noise denoising technique based on EMD and ICA

去除随机噪声是地震资料处理的重要环节,而目前的多数去噪技术都不同程度存在去噪效果差、损害有效信号等问题。为此,利用经验模态分解(Empirical Mode Decomposition,简写为EMD)能将信号自适应分解为不同尺度振动模态的优点及独立分量分析(Independent Component Analysis,简写为ICA)能提取独立源信号的优势,构造了一种EMD与ICA相结合的新的去噪算法,很好地实现了地震有效信号和随机噪声的分离,在提高去噪效果的同时提高了有效信号的保幅效果。将该算法应用于仿真实验和实际资料去噪,结果都明显优于总体经验模态分解(Ensemble Empirical Mode Decomposition,简写为EEMD)去噪结果,地震资料的信噪比和分辨率都大大提高。

Noise attenuation is an important step in seismic data processing.However,most noise attenuation technologies have some problems,such as poor denoising effect and damaging effective noise,more or less.As we all know,Empirical Mode Decomposition(EMD)can self-adaptively decompose signal into multi-scale vibration mode and Independent Component Analysis(ICA)can extract independent source signals.By using these advantages,we constructed a new denoising algorithm by combining EMD with ICA,which can well separate effective signals from random noise and improve the amplitude-preservation effect while improving denoising effect.The algorithm was applied on numerical simulation and actual data for denoising.As a consequent,the results are obviously better than that of Ensemble Empirical Mode Decomposition(EEMD).Meanwhile,the S/N and resolution of seismic data are greatly improved.