Signal Processing The Generalized S-transform And Its Application In The Processing Of Magnetotelluric Sounding Data

Electromagnetic method is usually used to study the structure and the conductivity of the earth internal abnormal medium in geophysics.So far,Fourier transform is the longest method in processing of the actual electromagnetic data,and also is the most common method.With the deepening of the oil and gas exploration work,the difficult in exploration is increasing,so other methods are more concerned in order to adapt to the requirements of the survey for geologists.The earth electromagnetic(MT)signal is a kind of nonlinear,the non-stationary random signal.At present the methods of processing the magnetotelluric sounding data commonly are:the least square method,far reference method,Robust method,etc.These methods are used Fourier transform to analysis the spectrum of the electromagnetic time series,because the Fourier analysis is a global transformation,and is without any of the frequency of resolving power in the time domain,and also is without any time distinguishing ability in frequency domain,which makes the signal noise time-varying filter difficulty in some degree.This paper introduced S transform and generalized S transform methods,and the application results in different field at home and abroad.Firstly this paper studied S transform and generalized S transform methods,and the basic principle of S transformation,and the relation of GST frequency and FFT frequency.The paper also studied the data processing feasibility of theory magnetotelluric sounding material,and the common noise frequency characteristics of electromagnetic signal in time domain.Based on the generalized S transform,the paper designed processing process and method of the MT data.This paper proposed the time-frequency ratio filter using the characteristics of the generalized transform S of MT data,and this paper put forward three kinds of threshold method to control noise and process the measured data.Compared with the different results,the GST method showed the superiority.S transformation was proposed by Stoekwell and others.Because of the fixed basic wavelet of the S transform and demanding wide frequency information in electromagnetic,and especially for more than 100 HZ and strength low 1 HZ human interference noise,the paper needed to improve the S transformation further to achieve satisfactory effects.This paper introduced the stack window to reduce the influence of boundary effect.The time-frequency analysis method of the improved generalized S transform is better than short-time window Fourier transform and wavelet transform.On one hand,with analysis of the electric field and magnetic field spectrum of time series,the method can accurate the noise positioning of the earth electromagnetic material,through the definition of different time-frequency window and different time-frequency filter to filter the noise.On the other hand,this method can get more spectrum coefficient,so as to improve the quality of power spectrum statistics.Through the simulation signal analysising and processing theory data of stability and instability time series,and processing of the noise with the different degree of noise interference,this method can accurate the noise positioning.Through the definition of different time-frequency window and different time-frequency filter and the noise filtering,the apparent resistivity and phase estimate quality has improved,compared with the processing of the result of the SSMT and GST.GST method proved the feasibility and effectiveness,but results were also slightly different according to the different filter method.