Low-frequency magnetotelluric signal denoising method based on shift invariant sparse coding

Abstract

The invention provides a low-frequency geomagnetic signal denoising method based on shift-invariant sparse coding. The method comprises the following steps: firstly, decomposing a noise-containing magnetotelluric time sequence into a low-frequency effective signal and a noise-containing high-frequency signal by using mathematical morphology filtering; and then dividing the high-frequency signal containing noise into a plurality of segments, autonomously learning a feature structure of human noise from the high-frequency signal containing noise by using shift invariant sparse coding to construct a learning type redundant dictionary, and performing signal-noise separation on the high-frequency effective signal containing noise by using the learned redundant dictionary to obtain a denoised high-frequency effective signal; and adding the low-frequency effective signal and the high-frequency effective signal to obtain a full-band magnetotelluric effective signal. Strong human noise in low-frequency magnetotelluric data can be effectively removed, the quality of the magnetotelluric data is remarkably improved, and the deep detection effect of a magnetotelluric method is improved.

Description

technical field [0001] The invention belongs to the field of exploration geophysical signal processing, and relates to a method for separating magnetotelluric signal-noise based on a shift-invariant sparse coding machine learning algorithm, especially for 10 -4 ~10 1 Processing methods for low frequency magnetotelluric data in the Hz range. Background technique [0002] The magnetotelluric method has the advantages of large detection depth and no need for artificial field sources, and is widely used in mineral resource exploration and deep electrical structure detection. However, the natural magnetotelluric signal has strong randomness, weak amplitude, and wide frequency range (10 -4 ~10 5 Hz), it is highly susceptible to human noise pollution. [0003] For decades, scholars at home and abroad have proposed many methods to suppress or separate the noise in the magnetotelluric signal. Among them, the far reference method and Robust method have been widely recognized and ...

AI Technical Summary

Problems solved by technology

These two types of methods can effectively protect low-frequency signals in high-quality segments, but the processing of noisy segments is still to directly extract human noise, which loses effective low-frequency signals in strong interference segments.

[0006] In general, since the frequency of the magnetotelluric signal is very wide (10 -4 ~10 5 Hz), the existing magnetotelluric data processing methods are difficult to meet the processing of data in various frequency bands, especially when processing low-frequency magnetotelluric data below 10Hz, the existing signal-to-noise separation methods have obvious limitations

For the processing of low-frequency magnetotelluric data below 1 Hz, only the general reference trace method or Robust method can be used for processing, and there are few signal-to-noise separation methods specifically for low-frequency magnetotelluric data

Images