稳健统计用于扩频激电数据预处理与脉冲噪声压制  

Robust statistical methods for spread spectrum induced polarization data preprocessing and reducing impulse noise

作　　者：刘卫强[1,2] 陈儒军[2] 

机构地区：[1]中国地质科学院地球物理地球化学勘查研究所,廊坊065000 [2]中南大学地球科学与信息物理学院,长沙410083

出　　处：《地球物理学进展》2016年第3期1332-1341,共10页Progress in Geophysics

摘　　要：在扩频激电数据预处理中,传统的均值叠加与基于叠加后数据的数字滤波方法并不能对脉冲噪声进行有效压制.脉冲噪声的影响会保留到复电阻率频谱中,对激电参数的计算产生不利影响.本文针对现有数据处理方法压制脉冲噪声的不足,提出将稳健统计方法应用于扩频激电数据预处理,主要包括将稳健最小二乘回归用于线性趋势项消除和将稳健M估计用于周期数据叠加.通过模拟数据测试为稳健统计方法选择合适的影响函数与迭代算法,然后将其应用于实测的扩频激电数据预处理.通过对处理结果与计算误差进行对比分析,发现稳健统计方法对脉冲噪声和高斯噪声都有较好的压制作用.相比于均值叠加,稳健统计可减小数据计算误差,提高数据预处理质量;同时提高计算结果随叠加次数的收敛速度,节省观测时间.Considering traditional mean stack and digital filtering method can＇t reduce impulse noise effectively in the data processing of induced polarization based spread spectrum signal. The impact of impulse noise will remain in the complex resistivity spectrum,which will influence the calculating of induced polarization parameters.Objective of this paper is to reduce the impulse noise in spread spectrum induced polarization（ SSIP） data preprocessing. Robust statistical method was used to the SSIP data preprocessing. Robust least-squares regression was used to remove the linear trend from the original data. Robust M estimate was used to stack the data of all periods. The most appropriate influence function and iterative algorithm were chosen by test the simulated data to suppress the outliers＇ influence. Above methods were used to the practical data.The calculating results and error were analyzed. Compared to the ordinary least square method,robust statistical methods can reduce Gaussian random noise and impulse noise effectively. The robust method can improve the quality of SSIP data. It can improve the convergence rate of calculating results with the stack times and saving the measuring time.

关 键 词：扩频激电 稳健最小二乘回归 稳健M估计 线性趋势项消除 周期叠加 

分 类 号：P631[天文地球—地质矿产勘探]