基于Hilbert-Huang变换的大地电磁信号谱估计方法

Magnetotelluric signal spectrum estimation method based on Hilbert-Huang transform

由于大地电磁(MT)信号的非平稳特性,基于傅里叶变换的谱估计方法会带来偏差。本文提出了一种基于HHT的边际谱估算方法,即边际谱是由Hilbert谱沿时间轴积分得到,它代表了整个信号在时间跨度上的幅值累积效应。文中证明了边际谱的线性性质,详细讨论了提高Hilbert谱估计精度的两种方法,并对仿真MT信号和实测MT信号进行了数值实验。理论与实验证明,基于HHT的谱估计方法更符合MT信号的实际情况,能够最小化大地电磁信号非平稳性带来的估算偏差,且可抑制较强的白噪声,其频率分辨率和估算精度均高于傅里叶变换方法。

Due to the unstable characteristics of magnetotelluric signal,the spectrum method based on the Fourier Transform could bring in deviation. A marginal spectrum estimation method based on HHT was proposed in this paper the marginal spectrum was obtained by Hilbert spectrum integrating along the time axis,it represents amplitude accumulation effect for the whole signal over the time span. In this paper the linear property of the marginal spectrum was proved,two methods which were used to improve the accuracy of the Hilbert spectrum estimation were discussed in detail.The numerical experiments for simulation MT signal and measured MT signal were conducted. The theory and experiments proved that spectrum estimation method which is based on the HHT fits the MT signal better,it could minimize the estimation deviation which was caused by the unstationarity of the MT signal,it also could restrain strong white noise,its frequency resolution and estimation accuracy are higher than Fourier Transform method.