论文部分内容阅读
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.
In the preprocessing of SSI data, impulse noise can not be effectively suppressed by the traditional average overlay and the digital filtering method based on the superposed data.The influence of impulse noise will be retained in the complex resistivity spectrum, In order to overcome the shortcomings of existing data processing methods to suppress impulsive noise, a robust statistical method is proposed for the preprocessing of spread-spectrum data, including robust least-squares regression for linear trend term elimination and robust M estimation is applied to the superposition of periodic data.The appropriate influence function and iterative algorithm are selected for the robust statistical method through the simulation data test and then applied to the preprocessing of the measured spread spectrum electrified data.Through the comparative analysis of the processing results and the calculated errors , And found that the robust statistical method has a better suppression of impulse noise and Gaussian noise.Compared with the mean superposition, robust statistics can reduce the data calculation error and improve the quality of data preprocessing, meanwhile improve the convergence rate of calculation results with the number of superpositions , Save observation time.