Supervised transient electromagnetic signal noise reduction method based on non-negative matrix factorization

Abstract

The invention discloses a supervised transient electromagnetic signal noise reduction method based on non-negative matrix factorization. First, in a training phase, short-time Fourier transform and non-negative matrix factorization processing are carried out on the pure signal, and atomic dictionaries representing respective characteristics of the signals are obtained; next, in a noise reduction phase, the noisy signal is processed by using the atomic dictionaries and the noise reduction model, and an initially estimated transient electromagnetic signal is obtained; and finally, the steps arerepeated for many times, the late-stage data of the initially estimated transient electromagnetic signal and the original early-stage and medium-stage data of the noisy signal are respectively accumulated,respective arithmetic mean values are solved, the arithmetic mean values are then spliced, anda final complete transient electromagnetic signal is estimated.The method can be applied to the workof exploring underground targets such as coal mines, metal mines and oil and gas fields by the transient electromagnetic instrument.

Description

technical field [0001] The invention relates to the field of transient electromagnetic signal processing, in particular to a non-negative matrix decomposition supervised transient electromagnetic signal noise reduction method. Background technique [0002] As an important geophysical prospecting method, the transient electromagnetic method is based on the principle of electromagnetic induction, using an ungrounded return line or a grounded line source to send a primary field underground, and under its excitation, a secondary field is excited in the underground geological body. , by analyzing and processing the secondary field signal, the purpose of detecting underground geological bodies is achieved. The transient electromagnetic method has high detection and resolution capabilities, and has been widely used in the exploration of underground targets such as coal mines, metal mines, and oil and gas fields. [0003] When working in the field, the signal collected by the trans...

AI Technical Summary

Problems solved by technology

[0003] When working in the field, the signal collected by the transient electromagnetic method will be interfered by various noises, such as: sky electrical noise, power frequency interference, etc., especially In the late stage of the transient electromagnetic signal, when the noise is serious, it will even overwhelm the clean transient electromagnetic signal, and these noises seriously affect the subsequent processing of the transient electromagnetic signal

In this regard, in the early stage of transient electromagnetic signal processing, people have studied the characteristics of various noises, and proposed a variety of noise reduction methods, such as wavelet transform, Kalman filter, singular value decomposition and stacked noise reduction autoencoder, etc., these The method has shortcomings such as lack of adaptability, unsatisfactory noise reduction results, complicated processing steps, and complex models, which are not conducive to subsequent inversion work

Images