Research On Magnetotelluric Noise Suppression Based On Adaptive Sparse Representation And K-SVD Dictionary Learning

With the rapid development of social economy,it is becoming more and more important to obtain resources from the depths of the earth.As a mainstream exploration geophysical method,the magnetotelluric sounding method plays a vital role in the analysis of geological structure and the search for deep bare ore bodies.Because the natural magnetotelluric signal has the characteristics of wide frequency band and weak signal,it is very easy to be interfered by human electromagnetic noise when collecting.Therefore,how to eliminate the strong interference in the measured magnetotelluric data and extract useful magnetotelluric signals has become an important research topic in the field of electromagnetic exploration.This paper introduces sparse representation and dictionary learning to magnetotelluric data processing,and mainly conducts the following three aspects of research work:(1)Aiming at the problems of insufficient adaptability and low processing efficiency of traditional sparse representation algorithms.The fuzzy entropy parameter is combined with the segmented orthogonal matching and tracking algorithm,and the sparse representation of the traditional preset over-complete dictionary is improved,and an adaptive segmented orthogonal matching and tracking algorithm is proposed.Use fuzzy entropy parameters to control the sparsity of the sparse representation algorithm,improve the efficiency of data processing and the accuracy of noise suppression.(2)Aiming at the problem that the traditional sparse representation algorithm cannot adaptively update the complete dictionary according to the data to be processed.Starting from the noise form of the magnetotelluric data,K-SVD dictionary learning is introduced to the magnetotelluric noise suppression,so as to realize the real-time update of the super-complete dictionary according to the data to be processed.The experimental results show that the magnetotelluric noise suppression method based on K-SVD dictionary learning can effectively eliminate the complex electromagnetic interference in the measured magnetotelluric data and retain low-frequency useful signals.(3)Aiming at the problem of insufficient feature learning in the K-SVD dictionary learning algorithm,the sparse coding algorithm is optimized and the dictionary update strategy is further improved.A magnetotelluric data processing method based on improved K-SVD dictionary learning is proposed.The experimental results show that the method has greatly improved the ability to learn the characteristics of the signal to be processed.