利用高阶谱重构功率谱抑制高斯有色噪声

在大地电磁测深法(MT)中,天然交变电磁场是其研究的主要对象。天然电磁场通常非常微弱,而传统的MT处理方法抑制噪声能力,尤其是抑制有色噪声能力相对较弱,从而导致大地电磁响应函数经常出现形态异常,在资料反演解释时,许多地质特征难以有效提取出来,这些问题严重阻碍了MT的实际应用效能和发展。根据任何高斯过程,其高阶统计量(高阶累积量,高阶谱)均为零的性质,通过信号的高阶谱恢复功率谱,再由功率谱估算MT响应函数,以抑制高斯有色噪声的影响。数值模拟试验显示,这种方法在抑制高斯有色噪声方面优于传统功率谱方法。

Using Higher-order Spectrum to Restructure Power Spectrum to Suppress Gaussian Colored Noise

For magnetotelluric (MT), its main research object is natural electromagnetic fields. Natural electromagnetic fields generally are very weak. However the anti-noise ability of traditional processing method is relatively weak, which lead to the shape abnormity of MT responses. When MT data is interpreted, it is very difficult to efficiently extract the geologic characteristics. All these problems badly lower MT practical efficiency. Based on that the higher-order statistics vanish for any Gaussian process, higher-order spectrum to reconstruct power spectrum to suppress Gaussian colored noise may be used. Synthetic test shows that the method presented here is superior in suppressing Gaussian colored noise.