局域均值分解和小波阈值在大地电磁噪声压制中的应用

发布时间：2018-05-24 18:55

本文选题：局域均值分解 + 小波阈值　；参考：《振动与冲击》2017年05期

【摘要】：大地电磁测深法是基于电磁感应原理,利用天然交变电磁场来研究地下岩层的电学性质及其分布特征。然而,天然电磁场频带范围宽、信号微弱,在实际测量中大地电磁信号极易受到各种电磁噪声干扰,严重影响了后续的电磁法反演解释水平。针对这一难题,将局域均值分解(LMD)的自适应性和小波分析的多分辨性相结合,提出基于局域均值分解和小波阈值的大地电磁噪声压制方法。将含噪信号进行LMD分解得到若干阶乘积函数(PF)分量;根据大地电磁信噪特征保留PF_1分量,仅对其余各阶PF分量选取合适的小波阈值进行降噪处理;叠加重构获得大地电磁有用信号。通过计算机模拟典型强干扰,研究不同小波函数、分解层数及阈值方式下算法的去噪性能,并将其应用于矿集区实测大地电磁数据处理。实验结果表明,所提方法能较好地提取出叠加在微弱大地电磁信号上的大尺度强干扰的轮廓特征,视电阻率曲线更为光滑、连续,低频段的大地电磁数据质量得到了明显改善。
[Abstract]:The magnetotelluric sounding method is based on the principle of electromagnetic induction and uses the natural alternating electromagnetic field to study the electrical properties and distribution characteristics of underground strata. However, because of the wide frequency band and weak signal of natural electromagnetic field, the magnetotelluric signal is easily disturbed by various electromagnetic noises in the actual measurement, which seriously affects the interpretation level of subsequent electromagnetic inversion. In order to solve this problem, combining the self-adaptability of local mean decomposition (LMD) with the multi-resolution of wavelet analysis, a magnetotelluric noise suppression method based on local mean decomposition and wavelet threshold is proposed. The noisy signal is decomposed into several order product function components by LMD decomposition, and the PF_1 component is retained according to the magnetotelluric signal noise feature, and only the proper wavelet threshold is selected for the other PF components to be de-noised. Superposition reconstruction can obtain magnetotelluric useful signals. Through computer simulation of typical strong interference, the de-noising performance of different wavelet function, decomposition layer number and threshold method is studied, and applied to the processing of magnetotelluric data measured in ore concentration area. The experimental results show that the proposed method can extract the contour features of large scale strong interference superimposed on weak magnetotelluric signals, the apparent resistivity curve is more smooth and continuous, and the quality of magnetotelluric data in low frequency band has been improved obviously.
【作者单位】：湖南师范大学物理与信息科学学院;中南大学地球科学与信息物理学院有色金属成矿预测与地质环境监测教育部重点实验室;湖南文理学院物理与电子科学学院;
【基金】：国家自然科学基金(41404111;41304098) 国家863计划(2014AA06A602) 湖南省自然科学基金(2015JJ3088) 中国博士后科学基金(2015M570687)
【分类号】：P631.325

【参考文献】