双极性阈值叠加及分段线性拟合方法消除低频噪声研究  

Study on De-noising in Low Electromagnetic Data Based on Bipolar Preset Stacking and Piecewise Linear Fitting

作　　者：徐汶东[1] 王艳[2] 

机构地区：[1]安徽理工大学计算机科学与工程学院,安徽淮南232001 [2]广东省电力设计研究院,广东广州510663

出　　处：《聊城大学学报（自然科学版）》2013年第3期31-35,共5页Journal of Liaocheng University:Natural Science Edition

基　　金：国家地球物理探测仪器工程技术研究中心开放基金(20130026);安徽理工大学博士基金(11199)

摘　　要：尖峰干扰和直流漂移的存在严重影响低频电磁法数据的质量及解释结果的可信度.在双极性叠加(次数为N)方法压噪失效的情况下,针对尖峰干扰及非线性漂移的特点,基于双极性叠加方法提出双极性阈值叠加和分段线性拟合的方法去噪.仿真结果表明,双极阈值叠加方法可以有效消除尖峰干扰,阈值M为0.1时,误差在0.12V左右,信噪比可提高N(1-M)倍.对低频非线性直流漂移,采用分段线性拟合方法去噪,可使功率谱逼近真实谱的误差在1%左右,满足实际应用要求.双极性阈值叠加压制和分段线性拟合压噪方法简单灵活,压噪效果明显,为低频电磁弱信号的提取提供了解决方法,同时为低频噪声的压制奠定了研究基础.the data quality and credibility of explanation is affected by spike and trend noises in low frequency electromagnetic prospecting, which makes the classical bipolar preset stacking method （N times） fail to de-noise. This contribution presents a method of de-noising which consists on bipolar pre- set stacking method and piecewise linear fitting. Results of the simulation indicate that the bipolar preset stacking method can remove spike noises effectively. When the value M preset is equal to 0.1, the error is about 0.12 V, and the SNR is improved to ~/N（1--M）times. for a nonlinear trend , the error of its power spectral approaching its real spectrum is about 1~ after de-noised by the piecewise linear fitting method, which meets the demand of practical application. The proposed methods present a easy, flexi- ble, and effective way for de-noising the spike and trend noise in the low frequency electromagnetic data, and establish research basis for de-noising of the low frequency noises.

关 键 词：尖峰干扰 直流漂移 低频电磁数据 双极性阈值叠加 分段线性拟合 

分 类 号：P631[天文地球—地质矿产勘探]