Magnetotelluric signal-noise separation method and system based on multi-resolution singular value decomposition

Abstract

The invention discloses a magnetotelluric signal-noise separation method and system based on multi-resolution singular value decomposition. The method comprises the following steps of: acquiring actually measured magnetotelluric data and segmenting the data; respectively constructing a Hankel matrix for each segment of a magnetotelluric signal, decomposing one layer by using a singular value to obtain approximate signals and detail signals with different resolutions, and dividing the magnetotelluric data into useful signal segments or strong interference data segments by using a difference value between the standard deviation of the approximate signals and the standard deviation of the detail signals with different resolutions; decomposing the approximate signal of each strong interference data segment by using a multi-resolution singular value decomposition (MRSVD) algorithm to obtain a large-scale noise contour, and subtracting the corresponding large-scale noise contour from the signal of the strong interference data segment to obtain a useful signal segment; and reconstructing the useful signal segments and the denoised data segments to obtain a magnetotelluric useful signal. According to the magnetotelluric signal-noise separation method, the MRSVD algorithm is introduced for decomposition, the decomposition error is small, and more low-frequency useful signals can be finely reserved.

Description

technical field [0001] The invention belongs to the technical field of magnetotelluric signal processing, and in particular relates to a method and system for separating magnetotelluric signal and noise based on multi-resolution singular value decomposition. Background technique [0002] Magnetotelluric (Magnetotelluric, MT) is a Cagniard resistance sounding method proposed by Tikhonow AN and Cagniard L in 1950, using natural magnetotelluric signals as field sources. The natural magnetotelluric signal is produced by the comprehensive action of various field sources with different strengths, different distances, and different attributes, and has its distinctive characteristics. Generally speaking, the natural magnetotelluric signal is weak and the frequency band of the signal is wide, especially with the expansion of the scope of human activities and the influence of terrain, the collected data will inevitably be interfered by various noises, making the time of the magnetotel...

AI Technical Summary

Problems solved by technology

In recent years, some emerging modern digital signal processing techniques, such as mathematical morphological filtering, far reference method, narrow-band filtering method, coherence method, least square method, Robust method, synchronous time series dependence, signal subspace enhancement, signal-to-noise identification and recursive Analysis coefficient decomposition, sparse decomposition, wavelet transform and comprehensive algorithm, variational mode decomposition, empirical mode decomposition, etc. have all been applied to this field. Some methods have improved the quality of magnetotelluric signals from different angles and suppressed strong interference noise. However, the low-frequency useful signal is seriously lost

[0004] Among them, empirical mode decomposition has been widely applied to magnetotelluric denoising, and its research results have gradually deepened the understanding of magnetotelluric signals and noise, but the modal aliasing and end-point effects in the decomposition process have seriously affected the low-frequency part. influences

This leads to a decline in the quality of the processed magnetotelluric signal data, seriously affecting the reliability and interpretability of the magnetotelluric data

Images