基于CEEMDAN和ICA的油气勘探大地电磁噪声消除方法

Magnetotelluric noise elimination for oil and gas exploration based on CEEMDAN and ICA

大地电磁(MT)信号中的噪声会严重影响大地电磁勘探的观测数据,导致后续反演、解释工作出现严重偏差,影响油气勘探效果。为此,提出了一种基于带自适应噪声的完全集合经验模态分解(CEEMDAN)和独立分量分析(ICA)的噪声消除方法。该方法将经验模态分解(EMD)中的CEEMDAN方法与盲源分离(BSS)中的ICA方法进行有效结合。首先,利用改进的端点检测技术识别电磁信号中的有噪声信号分段;其次,利用CEEMDAN方法对其进行分解,提取具有代表性的固有模态分量(IMF)分量并进行ICA处理,达到消除噪声的目的;然后,利用获得的独立分量对有用MT信号进行逆向重构;最后,将未受噪声污染的MT信号与去噪后的有用MT信号进行拼接,获得最终的消噪后的完整MT信号。对合成信号和实测MT数据的实验结果表明,该方法能有效消除MT信号中的噪声。

The noise in magnetotelluric signals seriously affects the observation results of magnetotelluric exploration,which leads to serious deviations in subsequent inversion and interpretation and affects the effect of oil and gas exploration.Therefore,this paper presents a denoising method based on complete ensemble empirical mode decomposition(CEEMD)with adaptive noise(CEEMDAN)and independent component analysis(ICA).This method effectively combines the CEEMDAN method in EMD with the ICA method in blind source separation(BSS).Firstly,the improved endpoint detection technology is used to identify the noisy signal segments in the electromagnetic signal.Then,the extracted noisy signal segments are decomposed by CEEMDAN,and the representative components of the intrinsic mode function(IMF)are extracted and processed by ICA to eliminate the noise.After that,the obtained independent components are used for the reverse reconstruction of useful MT signals.Finally,the MT signal without noise pollution is spliced with the useful MT signals after denoising to obtain the final complete denoised MT signal.The results of experiments on synthetic signals and measured MT signals show that the noise in MT signals can be effectively eliminated by this method.