The Study Of Fuzzy Cluster Method To Process The EM Noise Signals Of MT Data

In economically developed areas,there are often serious correlated noise interferences,and it is more and more difficult to accurately estimate the impedance of the earth's electromagnetic tensor.If you want to obtain more accurate inversion interpretation results,you need to study the method to improve the quality of the MT impedance tensor under strong noise conditions.This paper analyzes several typical noises and compares the advantages and disadvantages of traditional data processing methods.This thesis studies a new method of noise recognition and suppression,starting from the spectrum screening perspective in the data processing process.In the spectrum screening process,the “fuzzy clustering” analysis method is introduced;the spectrum is filtered by the characteristic that the natural MT field signal ratio tends to be fixed,the MT measured data is divided into signals and noise,and the noise spectrum is deleted to improve the power spectral density matrix estimation quality.The theoretical MT data and the measured MT data are processed by the new method studied in this paper.The results show that the fuzzy clustering method can identify strong correlated noise.The apparent resistivity and phase curves given by the fuzzy clustering method combined with the Robust method are smoother than those obtained by the Robust method alone,and the signal-to-noise ratio is significantly improved.The research results show that the fuzzy clustering method can significantly improve the processing results of magnetotelluric data.