Research On Signal Extraction Method For Magnetic Resonance Sounding In Complex Noise Environment

With the rapid development of industry,agriculture and urbanization in China,the demand for water resources exploration continues to grow.Magnetic resonance sounding(MRS),as a newly geophysical method,has become an important modern geophysical method for water exploration and has been widely concerned by international researchers due to its directly non-invasive quantitative detection of the aquifer location,water content and porosity.Compared with other geophysical methods such as induced polarization and transient electromagnetic method,MRS has the ability of fast and efficient detection of hydrogen proton abundance and water content in groundwater,and shows a broad application prospect in the field of hydrology and water resources survey and water source disaster monitoring and early warning.However,as weak signal of nano-volt level,MRS signal is vulnerable to various electromagnetic noises in the field,which leads to low signal-to-noise ratio of the original data.The existing noise processing methods and MRS instrument can not fully meet the technology requirements in complex noise environment.Therefore,the interference suppression and reliable acquisition of MRS signal parameters become the urgent problems for MRS method in the strong complex noise environment.The electromagnetic noise encountered in MRS groundwater detection includes spike pulse noise,power frequency harmonic noise,random noise and other non-stationary noise.The spike pulse noise has the characteristics of large amplitude,short duration and wide frequency coverage;The frequency of power harmonic noise is fixed and its amplitude is relatively stable;Random noise is composed of various unpredictable factors,there is no uniform law,and it has randomness in statistical characteristics.In general,the acquired MRS data needs to be filtered by combining various de-noising methods,and then the envelope of the de-noised signal is extracted and the signal parameters are fitted.However,when a variety of noises coexist,the noise interference is complex and its intensity is large,the fitting accuracy of signal parameters is low,which will directly reduce the accuracy of the inversion interpretation,and even get the wrong inversion results.Moreover,complex noise interference is easy to cause the saturation failure of MRS acquisition system.In order to solve the problem of complex and strong noise interference in MRS detection,this thesis takes noise suppression and accurate acquisition of MRS signal parameters as the research direction,and focuses on the accurate extraction of MRS signal parameters,spike noise suppression affecting the extraction accuracy of signal parameters,the development of receiving system and field cases in complex noise to further improve MRS noise suppression and signal parameter extraction theory.The main research contents are as follows:1.In terms of the problem that traditional extraction method of MRS signal parameters is limited in capability and low in precision when multiple noises coexist,the interference is complex and its intensity is large,the characteristics of different types of electromagnetic noise and their effects on MRS signal are analyzed.Based on time-frequency analysis and the principle of short-time Fourier transform,the extracting method of MRS signal parameters based on discrete STFT in frequency domain is proposed.This method only focuses on the relationship between the amplitude of the frequency spectrum at Larmor frequency and MRS signal parameters,which can greatly reduce the complex noise influence and obtain reliable extraction result of signal parameters.It provides a novel strategy for the MRS signal parameter extraction in groundwater detection in complex noise environment.2.In terms of the problem that the spike noise seriously affects the accuracy of MRS signal parameter extraction due to the superposition of spike noise frequency spectrum and MRS signal frequency spectrum when the spike noise is serious,the synchrosqueezing wavelet transform and the improved threshold method are investigated to suppress spike noise.Specifically,the SWT principle is described,and the SWT frequency resolution in processing MRS signals is analyzed and compared.In order to overcome the shortcomings of traditional threshold method,an improved threshold processing method is proposed.Furthermore,in detail the implementation of the improved threshold method and heuristic threshold calculation are described in the process of suppressing spikes.According to the proposed SWT-IMT method,the suppressing experiments of simulated spikes and field spikes with different amplitudes and durations are carried out.The results show that the SWT-IMT method can effectively eliminate the spike noise,and its effect is better than that of the continuous wavelet transform method and the traditional time-domain threshold identification and substitution method.3.In terms of the problem that the accuracy of traditional parameter extraction method depends heavily on the denoising effect when extracting MRS signal in time domain,an optimized extraction method of MRS signal parameters based on discrete short time Fourier transform(STFT)is proposed.The method can directly extract the initial amplitude,average relaxation time and initial phase of MRS signal in frequency domain according to Larmor frequency spectrum value,effectively reduce the influence of complex noise on the extraction accuracy of signal parameters.In order to obtain high-precision signal parameters from the single-sided amplitude in frequency domain,this method needs to overcome the fence effect in frequency domain and introduces the single-sided amplitude correction coefficient,and the influence factors of amplitude correction coefficient are analyzed.In order to extract signal parameters,according to the MRS signal to be extracted,a reference signal is firstly designed and the corresponding amplitude correction coefficient is calculated.Secondly,the noisy MRS signal is segmented by discrete STFT to calculate the corresponding single-sided amplitude spectrum,and the equation relationship between MRS signal parameters and single-sided amplitude spectrum is derived.Then,the amplitude correction coefficient and the single-sided amplitude close to the Larmor frequency in each segment are used to calculate the single-sided amplitude at the Larmor frequency.Finally,the initial amplitude and average relaxation time of MRS signal are obtained by optimization method.According to the initial amplitude and average relaxation time,the free attenuation signal is further constructed and the initial phase of MRS signal is optimized by gradually changing the phase of the constructed signal.The experimental results show that the proposed method based on discrete STFT is superior to the traditional extraction method,it can effectively obtain MRS signal parameters in complex noise environment.4.In terms of the problem that the denoising effect of noise suppression algorithm and the extraction accuracy of signal parameters are affected by the original data,an anti power frequency MRS acquisition system is developed.Specifically,the low-noise preamplifier circuit,band-pass filter circuit and power frequency notch circuit are designed for MRS acquisition system.Through the test and comparison,the system can effectively suppress a large number of out-band noise and in-band harmonic noise at specific power frequency,and improve the quality of raw MRS data.In order to further verify the actual effect of the MRS system and the methods,the field cases of groundwater exploration are carried out.The practicability of the system and the methods is verified by the spike noise suppression,signal parameter extraction and inversion interpretation of field data.The main innovations are as follows:1.Spike noise attenuation method based on SWT-IMT of MRS signal is proposed.In this method,the noisy MRS signal is decomposed into the SWT coefficients of different scales using SWT method,and the SWT coefficients are re-estimated using the improved threshold algorithm to achieve the attenuation suppression of the SWT coefficient of spike noise.Finally,the MRS signal is reconstructed by inverse SWT.Compared with continuous wavelet transform and threshold substitution method in time domain,the SWT-IMT method has better frequency resolution,spike recognition ability and suppression effect,and can effectively retain the signal and have good practicability in suppressing noise process.This algorithm can avoid the influence of signal and noise aliasing at Larmor frequency on the effective extraction of signal parameters,and is conducive to the application of MRS instrument in the complex environment with spike noise.2.MRS signal parameter extraction method is proposed based on discrete STFT and optimization in frequency domain.The traditional extraction of MRS signal parameters generally uses synchronous detection method,linear fitting or nonlinear fitting.However,the parameter extraction accuracy of the traditional method is seriously affected by the time domain filtering effect.In the thesis,the proposed extracting method of MRS signal parameters based on discrete STFT overcomes the shortcomings of the traditional method to be affected by residual noise in time domain in the extracting process of signal parameters.Compared with the traditional method,the proposed method can extract MRS signal parameters with higher accuracy in frequency domain,and has the advantage of approaching the ideal narrowband filter.The method provides a method support and a novel way to obtain MRS signal parameters from low SNR data under complex interference.3.MRS acquisition system design scheme under strong power frequency harmonic interference is proposed.Ensuring the low noise of acquisition circuit,the narrow-band-pass filter circuit is combined with notch circuit with specific power frequency to develop an anti power frequency full wave acquisition system.The system can effectively suppress the power frequency noise,prevent the saturation distortion of the amplifier circuit,and enhance the reliability of the system detection.Compared with the traditional broadband acquisition system,the system has better anti-interference ability and improves the quality of raw recording data.The MRS acquisition system provides an effectively technical equipment for MRS method in complex power frequency noise interference.