Research On Denoising Method Of Full-wave Magnetic Resonance Sounding Signal Based On Singular Spectrum Analysis

Water is an indispensable medium for the continuation of the vitality of all things,it is a very precious natural resource that must be possessed by industrial production,agricultural development,economic growth and environmental maintenance and protection.However,the underground fresh water resources that people can use at present are very limited,Therefore,how to effectively detect the underground water resources has become an important subject worth studying by the current scientists and technicians.Magnetic resonance sounding is a geophysical method for direct detection of groundwater.It has the characteristics of nondestructive detection,convenience,efficiency,accuracy,and so on.It is widely used in the fields of water resources exploration and evaluation,water gushing in tunnels,monitoring of geological hazards in mine water inrush and so on.However,MRS signal is weak and the magnitude is nanovolt(nV).During the practical application of the field,the detection signals collected by the instrument system are inevitably disturbed by the random natural noise and the power frequency harmonics in the environment,the quality of the received signal is affected by the quality of the received signal,thus the accuracy of the subsequent inversion interpretation results is reduced.Therefore,the noise suppression of MRS signals has become a technical bottleneck restricting the effective application of MRS detection methods.How to use effective algorithm to carry out the research of MRS signal extraction in complex noise environment has important practical application value.First,on the basis of the analysis of the characteristics of the MRS signal and the noise singularity spectrum,it is concluded that the energy characteristics of the MRS signal and the noise are different,and the contribution to the eigenvalues in the singular spectrum analysis is different.The singular spectrum analysis algorithm,which only includes Gauss random white noise,strong power frequency harmonic,and three different signal noise characteristics including power frequency harmonic and Gauss random white noise,is simulated.In order to verify the noise suppression performance of the singular spectrum analysis algorithm,we carry out a contrast experiment with the classic de-noising algorithms,such as modeling de-noising algorithm and ICA filtering algorithm,to verify the superiority of the proposed singular spectrum analysis and de noising algorithm.Secondly,two key parameters,window length and eigenvalue,are discussed and discussed in the singular spectrum analysis algorithm.Different window length is selected to carry on the simulation experiment and analysis,and the suitable window length selection standard is finally summed up through the comparison and analysis of the experiment.Combined with the complexity of the actual field noise environment,When the length of the window is determined,Several experiments are carried out in the three cases where the signal is the main component,the noise is the main component,the signal and the noise are roughly equal.Finally,the rational criterion of eigenvalue selection is summed up.Finally,in order to verify the performance of singular spectrum analysis algorithm for MRS signal de-noising,laboratory simulated signal experiments and field measured data processing experiments were carried out respectively.The simulation results show that the proposed algorithm can effectively remove the complex environmental noise.The SNR of MRS signal is up to 24 dB after de-noising.The fitting error of the initial amplitude and the average relaxation time is within the range of 4% and 7%.The validity of the algorithm is fully confirmed.The results of the measured data processing further prove the practicability of the algorithm.