基于信噪辨识的矿集区大地电磁噪声压制

为了避免形态滤波方法在大地电磁强干扰分离中的"过处理"、进一步保留大地电磁低频段的有用信息,提出基于信噪辨识的矿集区大地电磁噪声压制方法.首先,从信号处理的角度剖析矿集区典型强干扰与天然大地电磁微弱信号之间的定量辨识关系,利用形态分形维数和形态膨胀谱熵对大地电磁信号与强干扰进行信噪辨识.然后,结合形态滤波技术和阈值法,仅对辨识出明显不是天然大地电磁信号的异常波形进行噪声压制.最后,重构大地电磁有用信号,并对算法进行性能评价.仿真结果表明,形态分形维数和形态膨胀谱熵能较好地定量辨识大地电磁信号与强干扰,大地电磁信号中一些缓变化的低频信息得到了更为精细的保留;与形态滤波整体处理相比,本文所提方法获得的卡尼亚电阻率曲线更为光滑、连续,视电阻率值相对稳定,其结果更为真实地反映了测点本身所固有的大地电磁深部构造信息.

Magnetotelluric noise suppression base on signal-to-noise identification in ore concentration area

It is a very challenging task to carry out the high-precision signal-to-noise separation for magnetotelluric sounding data in ore concentration area. In order to avoid "over processing" of morphological filtering method in the strong interference separation, and further reserve the useful information of magnetotelluric sounding data in low frequency band, a new method of magnetotelluric noise suppression base on signal-to-noise identification is proposed in this paper. First of all, according to the different complexity of morphological characteristics between the strong interference types of ore concentration area and natural magnetotelluric signal, we combine mathematical morphology, fractal dimension and morphological spectrum, introducing the robust characteristic parameters-morphological fractal dimension and morphological spectrum entropy to quantitatively indentify magnetotelluric signal and strong interferences. Then, we combine morphological filtering technology with threshold method to suppress noise, which is not natural magnetotelluric signal but abnormal waveform by identification. Finally, we reconstruct the useful signal of magnetotelluric and evaluate algorithm performance. Simulation results show that the proposed method can be more finely reserve the slow change information in low frequency band, and the reconstructed time series of magnetotelluric are approach to the essence characteristics of natural magnetotelluric signal. When it suppresses the adjacent source interference in ore concentration area, the method can effectively avoid the "over processing" of the morphology filtering technology in the noise suppression, and the integral morphology of Cagniard resistivity-phase curve is more smooth and continuous. Moreover, the apparent resistivity values are relatively stable. The results are more truly reflect the inherent deep structure information of measured data itself, and the overall data quality of the low frequency band has been improved significantly.