多周期全波形激电抗干扰数据处理方法及在大规模探测中的应用分析

激电法是金属矿产勘查中一种十分重要的电法勘探分支方法，但是各种电磁干扰的存在限制了激电法在大规模探测中的应用.近年来，国内外先后实现了三维分布式全波形激电探测仪器系统的研发和推广，全波形数据记录为激电信号的抗干扰处理提供了新的空间.本文针对多周期全波形采样的激电数据提出了一套基于统计分析的时间序列抗干扰数据处理方法，主要包括：经验模态分解用于分离低频趋势项干扰；相关分析用于消除突发性强噪声干扰；稳健统计用于多周期时间序列叠加；分段稳健平均和低频相对相位谱用于时/频域激电参数提取.将上述数据处理方法应用于由国产分布式电法系统实测的三维全波形激电数据，并与线性拟合、均值叠加等常处理方法进行对比，发现新方法可以有效识别和压制激电数据中的强噪声干扰，提高大供电极距和低频点探测时的激电数据质量，从而进一步推动激电法在深部矿产资源勘查中的应用.

Anti-interference processing of multi-period full-waveform induced polarization data and its application to large-scale exploration

Induced Polarization (IP) is one of the most important electrical methods in exploration of metallic ore deposits. However the existence of various electromagnetic interferences restricts its application in large-scale exploration. To solve this problem, 3D disturbed full-waveform IP instrument systems have been developed in China and elsewhere in the world, of which data processing needs further improvement. In this paper, a complete anti-interference processing method based on statistical analysis for the multi-period full-waveform IP data is proposed. Its steps are as follows:Empirical mode decomposition is used to separate low-frequency trend term interference. Correlation analysis is made to eliminate the sudden strong noise interference. Robust statistics are performed to multi-period time series stack. The subsection robust average and low frequency relative phase spectrum are employed to extract the time/frequency domain IP parameters. This data processing method is applied to the three-dimensional full-waveform IP data acquired by a domestic distributed electrical system and compared with the common processing methods such as linear fitting and mean stacking. Results show that the new method can effectively identify and suppress the strong noise interference in IP data, improve the quality of IP data when the current electrode space is large and the frequency is low, and further promote the application of the IP method in exploration of mineral resources in the deep subsurface.