基于Bayes Bootstrap统计降噪方法的磁共振测深信号检测

来源期刊:中南大学学报(自然科学版)2014年第9期

论文作者:张海如 欧阳缮 王国富 张法全

文章页码:3144 - 3150

关键词:磁共振测深;统计降噪;信号检测;Bayes Bootstrap模型

Key words:magnetic resonance sounding; statistical noise reduction; signal detection; Bayes Bootstrap model

摘    要:为从受强噪声干扰的实测磁共振测深(MRS)信号中提取有效特征参数,针对导电层状模型MRS信号,利用各导电层MRS信号分量取对数后与采样时间具有严格线性关系的特性,提出从各导电层MRS信号分量中提取2个最优估计点来重建MRS信号的思想,即用2个最优估计点代替整体统计误差最小时的估计值,以提高重建精度。该方法先将多周期各导电层MRS信号分量序列转化为二维矩阵,再用Bayes Bootstrap模型对各列信息进行误差估计,最后,根据误差估计结果,统计出2个最优有效数据点。研究结果表明:本文方法提取MRS信号特征参数的平均误差率远比传统方法的小,能从信噪比为1 dB的强噪声环境中,以5.5%的平均误差率提取出有效特征参数;能从同时受到工频干扰、奇异噪声干扰和平稳随机噪声干扰的含噪MRS信号中提取有效特征参数。此外,本文方法能从较少周期的实测数据中提取出MRS信号的有效特征参数,可提高MRS仪器地质探测效率。

Abstract: In order to extract a set of feature parameters of magnetic resonance sounding (MRS) signal, an algorithm of statistical noise reduction was presented over layered conductive earth, which took full use of the linear relationship aong the logarithmic MRS signal of each layered electrical conductive media and its sampling time. It estimated two optimal points from the real inspected signal sequence to reconstruct the MRS signal. Since it used only two optimal statistical points instead of the set of parameters with the minimum global statistical error, the algorithm could enhance the reconstruction accuracy of the MRS signal. The algorithm translated the multi-period MRS signal of each layered electrical conductive media sequence into two-dimensional matrix at first. Then, it structured the Bayes Bootstrap model to estimate the error of each column vector. Finally, according to these results of error analysis, the two optimal points were acquired. The results show the algorithm has a much lower average error rate than the traditional ones. It can extract effective feature parameters from the acquired MSR signal under poor SNR (1 dB) at the average error rate 5.5%. Furthermore, it also works for complicated original MSR signal, which contains the stable stochastic noise, the power frequency disturbing signal and the spike noise. With it the time for the sample period can be reduced greatly, inversion precision of the MRS systems can be improved.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号