Research On Powerline Harmonic Noise Reduction And Oscillating Signal Extraction Method Of Magnetic Resonance Sounding

With fast population growth,rapid economic and social development,as well as accelerating urbanization,the water demand in our conutry continues to rise.In order to avoid the industrial shutdown,food crisis and ecosystem deterioration caused by the shortage of water resources,the development and utilization of groundwater has become the focus of attention.Magnetic Resonance Sounding(MRS)has become widely used in the field of groundwater exploration because it has direct sensitivity to hydrogen atoms and can realize non-invasive detection and quantitative evaluation of groundwater.However,due to the weak MRS signal(usually n V level)caused by the low intensity of the applied background field,and the serious noise interference mainly consisted of powerline harmonics in the measurement environment,the low signal-to-noise ratio has become the main reason that restricts the usage of MRS detection in urban and rural areas.To deal with the detection environment with rapid noise changes and complex noise sources,and solve the problems of ubiquitous and serious interference of powerline harmonic noise,difficulty in correctly separating the aliased co-frequency noise with signal,and low accuracy of signal extraction in MRS detection process,this paper carries out the research on powerline harmonic noise reduction and high-sampled oscillating signal extraction supported by the China NSFC under Grant 41722405.The main research results are as follows:(1)Based on the principle of MRS detection,the mathematical expression of induced electromotive force of MRS signal in receiving coil is established,and the difference between the MRS oscillating signal and envelope signal is expounded.The influence of different levels of powerline harmonic noise on signal extraction and data inversion interpretation is studied;the impact of low pass filter parameter selection,frequency offset of reference signal and initial phase of MRS signal on traditional synchronous detection method in signal extraction process is analyzed;and the error caused by residual noise and initial value selection in data fitting method is discussed.Finally,the necessity of in-depth research of MRS data processing methods is emphasized.(2)To solve the problem of serious interference and intensive sources of powerline harmonic noise,a frequency-domain symmetry method is proposed.The imaginary part spectrum generated by the powerline harmonic noise after Fourier transform and demodulation can be used to estimate the real part spectrum which is aliased with MRS Signal to eliminate the noise.The numerical results show that the frequency-domain symmetry method performs better than the conventional modeling method,and can eliminate the powerline harmonic noise more accurately.In addition,it is found that when the harmonic frequency is symmetric with respect to the Larmor frequency,the frequency-domain symmetry method may fail,be inefficient or do not require spectrum correction,but the possibility of such special cases is often not high due to the influence of geographical location.(3)To solve the problem of signal distortion caused by the aliasing of co-frequency noise and MRS signal,a correlation modeling method is proposed.Based on the correlation detection principle of orthogonal vector phase-frequency detector,a mathematical model is established for the detection result of MRS signal containing the co-frequency noise,and then the mathematical transformations are performed on the model to eliminate the interference item of the co-frequency noise.At last,the MRS signal is recovered.The simulation results show that the correlation modeling method can effectively eliminate different levels of co-frequency noise and improve the accuracy of MRS signal extraction results.(4)To solve the problems that the conventional envelope detection method will introduce errors and the residual noise will reduce the accuracy of signal extraction,a shaping Prony method to extract MRS oscillating signal is proposed.By deriving the corresponding relationship between the complex exponential form of MRS signal and the parameters of shaping Prony component,the useful component can be uniquely determined within the effective frequency band and then the signal parameters can be calculated.The simulation results show that the number of decomposition plays a decisive role in the performance of the shaping Prony method,and a reasonable number of decomposition can balance the extraction accuracy and calculation time of the method.The shaping Prony method is free from the powerline harmonic noise,but the increase of random noise will reduce the method performance.When the signal-to-noise ratio of MRS signal and random noise is not less than-13 d B,the relative error of the extracted parameters is not more than 10%.Compared with conventional synchronous detection method,the shaping Prony method can provide more accurate signal parameter estimation results.(5)The methods proposed in this paper are used to improve the conventional MRS data processing flow,and the actual data acquisition and interpretation experiments were carried out in the field experimental test site of Shaoguo Town,Changchun City.Comparing the data processing results as well as the corresponding inverted water content information of the conventional process and the improved process,the results show that the improved process has a higher degree of noise suppression,a more accurate MRS signal extraction,and a more consistent inversion interpretation with the drilling geological data than the conventional process.