Research Of The Methods About MRS Signal Denotsing Based On Wavelet Analysis And Statistical Theory

In recent years, Surface Magnetic Resonance Sounding (MRS) technology hasgot rapid development; this technology is the most direct geophysical method todetect underground water. Essentially, this method makes energy level transition ofhydrogen proton in underground water. When the extranuclear electrons drop to a lowlevel, the coin receives the electromagnetic wave released by the extranuclearelectrons, then the information of underground water could be known throughprocessing. However, because the amplitude of the MRS signal is very small, thesignal is easy to be disturbed by plenty and complex noise, at this time, the signal tonoise ratio (SNR) will become lower, lead to great errors for the estimates ofgroundwater information, therefore, how to effectively inhibit interference andimprove the signal to noise ratio becomes critical. According to noise of differenttypes in the signal, this paper puts forward the corresponding de-noising strategy.After comparing the de-noising method proposed by this paper with existingde-noising technology, the occasions in which the de-noising method proposed by thispaper can apply are determined, and the effect of noise elimination and analysis ofresult are given. The reliability and validity of de-noising method proposed by thispaper are proved by using the simulation data and experimental data respectively.For the power frequency harmonic noise and random noise, the de-noisingmethod of wavelet threshold is be used to eliminate the noise in synthetic MRS signal,the feasibility of this method is be verified, and the method is used in the noisesuppression of the measured MRS signal, through the analysis of experiment results,complete wavelet threshold and Sqtwolog wavelet threshold could make the signal oflow SNR showing the characteristics of strong signal, the noise can be suppressedeffectively and SNR is significantly increased. The other four kinds of waveletthreshold can play the role of making the signal curve smoother. The method ofadaptive notch filter combined with reconstruction of wavelet modulus maximum value is also used to filter out the above two kinds of noise. The experimental resultsshow that the method can remove the noise effectively and the SNR is improvedgreatly. When both the SNR and the integrity of useful signal are taken into account,the method is compared with the method of wavelet threshold can play a better role.For eliminating the spike noise, the de-noising method of the wavelet highdecomposition coefficient threshold and the statistical de-noising method based onGrubbs criterion and Dixon criterion are used, the second kind of method includingGD method and GDT method. Compared with the traditional stacking method ofsingular value threshold, the methods above all improve the utilization rate of data,can suppress the spike noise more effectively and improve the SNR greatly. thestacking of GD method and the stacking of GDT method can get a more stable signalcurve.Methods researched in this paper can effectively restrain the noise in the MRSsignal and improve the SNR, increase credibility of the following inversion andinterpretation for the MRS data and make the calculation of hydrogeologicalparameters more accurate.