Underground water magnetic resonance longitudinal relaxation time detection method

Abstract

The invention belongs to the technical field of geophysical exploration, and relates to an underground water magnetic resonance longitudinal relaxation time detection method, which comprises the following steps: laying a magnetic resonance detection coil on the ground, fixing a short emission time, changing a current amplitude, emitting a plurality of short pulses, and collecting a short pulse observation signal generated by each short pulse; fixing a long emission time, re-emitting a series of long pulses with different current amplitudes, and collecting a long pulse observation signal generated by each long pulse; inverting the short pulse observation signal to obtain water content and transverse relaxation time distribution of underground water at different depths; and taking the water content and the transverse relaxation time distribution of the underground water at different depths as prior data of Bayesian inversion, and analyzing the long pulse observation signal by adopting Bayesian inversion to obtain longitudinal relaxation time distribution. According to the invention, a complex excitation sequence form is not needed, and a longitudinal relaxation signal of a short relaxation sample with a fast attenuation speed can be captured. The method is suitable for a high-power scene of field ground magnetic resonance detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of geophysical exploration, and more specifically, it is a method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance. Background Art

[0002] Surface nuclear magnetic resonance technology is a non-invasive geophysical method that directly detects groundwater quantitatively and evaluates the pore size distribution. There are three time constants in the nuclear magnetic resonance response that reflect the relaxation characteristics of the aquifer, namely the longitudinal relaxation time , the transverse relaxation time and the observed relaxation time . In porous media, the relaxation time is approximately proportional to the average pore size and can be used to infer the pore characteristics of the aquifer. The research on the acquisition methods of the transverse relaxation time and the observed relaxation time in surface nuclear magnetic resonance is relatively sufficient. In order to comprehensively reflect the pore characteristics of groundwater, it is necessary to achieve the detection of the longitudinal relaxation time in surface nuclear magnetic resonance.

[0003] In the existing technology, a complex excitation sequence form is often used, and the observed relaxation time parameter is sensitive to the inhomogeneity of the geomagnetic field between formation rocks and cannot directly reflect the pore characteristics of groundwater. In addition, it is difficult to capture the longitudinal relaxation signal of samples with fast attenuation speed and short relaxation. Summary of the Invention

[0004] The embodiments of the present application provide a method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance, which solves the problem of difficult to capture the longitudinal relaxation signal of samples with fast attenuation speed and short relaxation.

[0005] The present invention is implemented as follows: A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance, the method includes: Laying a nuclear magnetic resonance detection coil on the ground, fixing a short emission time, changing the current amplitude to emit a plurality of short pulses, and collecting the short pulse observation signals generated by each short pulse; Fixing a long emission time, re-emitting a series of long pulses with different current amplitudes, and collecting the long pulse observation signals generated by each long pulse; Inverting the short pulse observation signals to obtain the water content and transverse relaxation time distribution of groundwater at different depths; Taking the water content and transverse relaxation time distribution of groundwater at different depths as the prior data of Bayesian inversion, and using Bayesian inversion to analyze the long pulse observation signals to obtain the longitudinal relaxation time distribution.

[0006] Further, the inverting the short pulse observation signals to obtain the water content and transverse relaxation time distribution of groundwater at different depths includes: Establishing the objective function of the regularized inversion as: , wherein, is a short - pulse inversion model representing the water content and transverse relaxation time distribution of groundwater at different depths, is the short - pulse magnetic resonance FID signal obtained by forward calculation, is a smoothness matrix describing depth correlation, is a regularization parameter used to balance data fitting and model smoothness, is the data normalization term, is the short - pulse observed signal; The water content and transverse relaxation time distribution of groundwater at different depths are obtained by iteratively solving the objective function.

[0007] Furthermore, the calculation process of the short - pulse magnetic resonance FID signal includes: Calculating the transverse magnetization vector under a short pulse: , where is the transverse magnetization vector at spatial point under a short pulse, is the net magnetization vector, is the magnetization vector rotation angle at spatial point ; Calculating the short - pulse magnetic resonance FID signal in the time domain according to the transverse magnetization vector under a short pulse.

[0008] Furthermore, using Bayesian inversion to analyze the long - pulse observed signal to obtain the longitudinal relaxation time distribution includes: Calculating the fitting error , represents the long - pulse magnetic resonance FID signal obtained by forward calculation, is the long - pulse observed signal; Randomly perturbing a region and re - sampling the longitudinal relaxation time distribution in the current Bayesian inversion model to obtain a new Bayesian inversion model; Calculating the fitting error of the new Bayesian inversion model and judging whether to accept the new Bayesian inversion model in the form of simulated annealing; Continuously perturbing and re - sampling the longitudinal relaxation time distribution in the current Bayesian inversion model to form state jumps of the Markov chain until the maximum number of iterations is reached.

[0009] Furthermore, during the process of using Bayesian inversion to analyze the long - pulse observed signal to obtain the longitudinal relaxation time distribution, the water content and transverse relaxation time distribution of groundwater at different depths remain unchanged.

[0010] Further, the calculation process of the long-pulse nuclear magnetic resonance FID signal includes: searching for the slice with the closest combination of emission time and relaxation parameters in the interpolation library according to the emission time and the relaxation parameters obtained by Bayesian inversion, and obtaining the transverse magnetization vector under the long pulse by one-dimensional interpolation of the slice; calculating the long-pulse nuclear magnetic resonance FID signal in the time domain according to the transverse magnetization vector under the long pulse.

[0011] Further, the interpolation library is formed by calculating the Bloch equations under different combinations of emission time, longitudinal relaxation time, and transverse relaxation time with dense parameter intervals by the Runge-Kutta method, and obtaining the look-up table relationship between the transverse magnetization vector and the excitation field strength.

[0012] Further, the short emission time means that the emission current duration in one pulse period is less than 50 ms; the long emission time means that the emission current duration in one pulse period is greater than 200 ms.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: Based on the relaxation effect of the hydrogen proton magnetization vector during the excitation stage, this method can invert the longitudinal relaxation time distribution of groundwater only by using the free induction decay signal (FID) obtained by single-pulse excitation, without the need for a complex excitation sequence form, and can capture the longitudinal relaxation signals of samples with fast attenuation speed and short relaxation. It is applicable to high-power scenarios of field ground nuclear magnetic resonance detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flowchart of a method for detecting the longitudinal relaxation time of groundwater nuclear magnetic resonance provided by an embodiment of the present application; Figure 2 are the FID detection data under short-pulse (a) and long-pulse (b) excitations provided by an embodiment of the present application; Figure 3 are the groundwater inversion results provided by an embodiment of the present application. (a) is the water content inversion result, (b) is the transverse relaxation time inversion result, and (c) is the longitudinal relaxation time inversion result; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the present invention.

[0016] Groundwater nuclear magnetic resonance exploration utilizes the principle of nuclear magnetic resonance to detect groundwater. Its working principle is based on the nuclear magnetic resonance phenomenon of hydrogen nuclei in water. When the hydrogen nuclei in groundwater are excited by radio frequency pulses in a specific magnetic field environment, nuclear magnetic resonance signals are generated. By receiving and analyzing these signals, the equipment can obtain relevant information about groundwater, such as the location, thickness, and water content of the aquifer. Groundwater nuclear magnetic resonance exploration equipment generally includes: Transmission system: Used to generate and transmit radio frequency pulses with specific frequencies and intensities to the ground to excite the nuclear magnetic resonance of hydrogen nuclei in groundwater. Receiving system: Responsible for receiving the signals generated by the nuclear magnetic resonance of underground hydrogen nuclei, and amplifying, filtering, etc. these signals for subsequent analysis. Control system: Controls the transmission system and the receiving system, coordinates the work of each part, and ensures the smooth progress of the detection process. The transmission system is connected to the transmitting coil or long wire, and the receiving system is connected to the receiving coil. The embodiments of this application achieve the emission of pulses and the reception of observation signals with the help of a groundwater nuclear magnetic resonance detection system.

[0017] See Figure 1 The flowchart of a method for detecting the longitudinal relaxation time of groundwater nuclear magnetic resonance shown, provides a method for detecting the longitudinal relaxation time of groundwater nuclear magnetic resonance, including: S1 Lay a nuclear magnetic resonance detection coil on the ground, fix a short emission time, change the magnitude of the current amplitude to emit multiple short pulses, and collect the short pulse observation signals generated by each short pulse; The detection coil here includes the receiving coil and the transmitting coil, and there is no restriction on the laying form of the receiving coil and the transmitting coil. The short emission time is relative to the long emission time, referring to the duration of the emission current in a pulse period. In the embodiments of this application, the short emission time is controlled within 50 ms, and the magnitudes of the current amplitudes of multiple short pulses are variable, and the current amplitudes are distributed logarithmically within a certain range, for example, controlled within a range of 1 A - 200 A.

[0018] The obtained short pulse observation signal is the free induction decay signal (FID).

[0019] S2 Fix a long emission time, re - emit a series of long pulses with different current amplitudes, and collect the long pulse observation signals generated by each long pulse; In the embodiments of this application, the long emission time is controlled to be greater than 200 ms, that is, the duration of the emission current in a pulse period is controlled to be greater than 200 ms.

[0020] S3 Invert the short pulse observation signals to obtain the water content and transverse relaxation time distribution of groundwater at different depths; S4 Use the water content and transverse relaxation time distribution of groundwater at different depths as the prior data for Bayesian inversion, and adopt Bayesian inversion to analyze the long pulse observation signals to obtain the longitudinal relaxation time distribution.

[0021] The evolution law of hydrogen protons in groundwater at different emission times is as follows: The evolution process of hydrogen protons in groundwater follows the Bloch equation and can be expressed as: , In the formula, is the gyromagnetic ratio, is the net magnetization vector, is the longitudinal relaxation time, is the transverse relaxation time, is the external magnetic field, , and are the directional components of the magnetization vector ; The relaxation effect of the magnetization vector during external excitation will cause the attenuation of its modulus value, and the attenuation degree is affected by the longitudinal relaxation time and the transverse relaxation time; when the relaxation characteristics and the external excitation conditions are fixed, the longer the emission time, the greater the attenuation degree of the modulus value. Utilizing this feature, in the embodiments of the present application, the short-pulse observation signal is inverted to obtain the water content and the transverse relaxation time distribution of groundwater at different depths.

[0022] The obtained water content and transverse relaxation time distribution of groundwater at different depths are used as the prior data for Bayesian inversion, that is, the water content, longitudinal relaxation time, and transverse relaxation time of groundwater at different depths are included in the Bayesian inversion. The long-pulse observation signal is analyzed by Bayesian inversion to obtain the longitudinal relaxation time distribution.

[0023] In one embodiment, inverting the short-pulse observation signal to obtain the water content and transverse relaxation time distribution of groundwater at different depths includes: Establishing the objective function of the regularization inversion as: , In the formula, is the short-pulse inversion model representing the water content and transverse relaxation time distribution of groundwater at different depths, is the short-pulse nuclear magnetic resonance FID signal obtained by forward calculation, is the smoothness matrix describing the depth correlation, is the regularization parameter used to balance data fitting and model smoothness, is the data normalization term, is the short-pulse observation signal; Iteratively solving the objective function to obtain the water content and transverse relaxation time distribution of groundwater at different depths.

[0024] For short-pulse excitation, there is an explicit analytical solution for the transverse magnetization vector at the end of the excitation: , In the formula, is the transverse magnetization vector at the spatial point under a short pulse, is the net magnetization vector, is the magnetization vector rotation angle at the spatial point .

[0025] Through the above formula, the transverse magnetization vector under a short pulse can be obtained, and the short-pulse nuclear magnetic resonance FID signal in the time domain is calculated based on the transverse magnetization vector under the short pulse.

[0026] In one embodiment, since there is no explicit analytical solution between the transverse magnetization vector and the relaxation parameters under long-pulse excitation, Bayesian inversion is used to analyze the long-pulse observed signal to obtain the longitudinal relaxation time distribution of groundwater at different depths.

[0027] The Bayesian inversion framework can be expressed as: , In the formula, is the prior probability density function of the Bayesian inversion model, represents the likelihood function, is the probability density function of the long-pulse observed signal, represents the posterior probability density function of the longitudinal relaxation time of the Bayesian inversion model.

[0028] The Markov chain Monte Carlo sampling method is used to solve this Bayesian problem. The specific iterative process includes: calculating the fitting error , represents the long-pulse nuclear magnetic resonance FID signal obtained by forward calculation, is the long-pulse observed signal; Randomly perturb a region in the longitudinal relaxation time distribution of the current Bayesian inversion model and re-sample to obtain a new Bayesian inversion model; Calculate the fitting error of the new Bayesian inversion model, and judge whether to accept the new Bayesian inversion model in the form of simulated annealing; Continuously perturb and re-sample the longitudinal relaxation time distribution of the current Bayesian inversion model to form the state jump of the Markov chain until the maximum number of iterations is reached.

[0029] Here, the calculation method for the short-pulse nuclear magnetic resonance FID signal in the short pulse cannot be used. For the long-pulse excitation of this application, the magnetization vector is affected by the relaxation effect during the excitation stage and an explicit analytical solution cannot be used. This application quickly obtains the transverse magnetization vector through the "slicing-interpolation" magnetization vector calculation strategy. Specifically, it includes: The Runge-Kutta method is used to calculate the Bloch equation for different emission time, longitudinal relaxation time and transverse relaxation time parameter combinations with dense parameter intervals, and the lookup table relationship between the transverse magnetization vector and the excitation field strength is obtained as a large interpolation library; For the forward response solution, the transverse magnetization vectors of all spatial points can be searched in parallel for the slice of the closest combination of emission time and relaxation parameters in the interpolation library, and then one-dimensionally interpolated from the slice according to the excitation field intensity at the spatial point.

[0030] The long pulse magnetic resonance FID signal in the time domain is calculated according to the transverse magnetization vector under the long pulse.

[0031] The method for calculating a long pulse magnetic resonance FID signal or a short pulse magnetic resonance FID signal in the time domain according to a transverse magnetization vector under a long pulse or a short pulse comprises: Spatial point kernel function is related to the transverse magnetization vector and is expressed as: , In the formula, represents the Larmor frequency, The term represents the spatial sensitivity of the receiving coil, and are the phase components of the transmitting field and the receiving field generated under elliptical excitation, represents the net magnetization vector; Assumptions and Represents a spatial point The water content at and the observed relaxation time are calculated according to Kernel function forward calculation can obtain short pulse or long pulse magnetic resonance FID signal , expressed as: , in, Indicates time.

[0032] The following is a specific application to explain the implementation process and effects of the embodiment of the present application.

[0033] A rectangular magnetic resonance detection coil is laid on the ground, with a size of 100m×100m and 1 turn; Fixed a short launch time , changing the current amplitude to emit 40 short pulses, the current amplitude is distributed in a logarithmic form between 1A and 200A; After completing 40 short pulse excitations, a long emission time is fixed. , re-emit a series of long pulses of different current amplitudes; Figure 2Among them, (a) is the short-pulse observation signal of a 40-ms short pulse, and the real part, imaginary part, and modulus are used to represent the short-pulse observation signal. Figure 2 Among them, (b) is the long-pulse observation signal of a 600-ms long pulse, and the real part, imaginary part, and modulus are used to represent the long-pulse observation signal; it can be seen that the long-pulse observation signal shows a significant amplitude attenuation compared with the short-pulse observation signal.

[0034] According to the evolution law of groundwater hydrogen protons at different emission times, the relaxation effect that occurs when the magnetization vector is externally excited will cause the modulus of the magnetization vector to decay, and the degree of decay is affected by the longitudinal relaxation time and transverse relaxation time parameters. When the relaxation characteristics and external excitation conditions are fixed, the longer the emission time, the greater the degree of decay of the modulus of the magnetization vector. The emission times are taken as 40 ms and 600 ms respectively, and the longitudinal relaxation time takes 100 values at uniform intervals within the range of 0.01 s - 1 s. For each longitudinal relaxation time value, the transverse relaxation time takes 100 values at uniform intervals within the range of to satisfy ; According to the evolution law of groundwater hydrogen protons at different emission times, the initial amplitude of the short-pulse observation signal obtained by the 40-ms short pulse is mainly related to the formation water content, while the attenuation amount of the initial amplitude of the long-pulse observation signal obtained by the 600-ms long pulse compared with the short-pulse observation signal is related to the longitudinal relaxation time at different depths. By establishing the objective function of regularization inversion and iteratively solving the objective function, the water content and transverse relaxation time distributions of groundwater at different depths are obtained.

[0035] Since there is no explicit analytical solution between the transverse magnetization vector and the relaxation parameters under long-pulse excitation, the Bayesian inversion method is used to analyze the long-pulse observation signal to obtain the longitudinal relaxation time distribution of groundwater at different depths. The Bayesian inversion model includes the transverse relaxation time, longitudinal relaxation time, and water content. The water content and transverse relaxation time distributions of the short-pulse inversion model of the short pulse are assigned to the Bayesian inversion model. According to the above iterative process of the Markov chain, the longitudinal relaxation time distribution of groundwater at different depths within 60 m underground is obtained.

[0036] Figure 3 Among them, (a) is the inversion result of the water content, Figure 3 Among them, (b) is the inversion result of the transverse relaxation time, Figure 3Among them, (c) is the inversion result of the longitudinal relaxation time. Among them, the two main aquifers are located at 20 m and 40 m underground respectively; the longitudinal relaxation time of shallow groundwater is about 0.5 s; while the longitudinal relaxation time of deep groundwater is relatively large, about 0.8 s.

[0037] In the embodiment of the present application, the observation signal obtained by only one short-pulse excitation and the observation signal obtained by only one long-pulse excitation can be used to invert the longitudinal relaxation time distribution of groundwater, without a complex excitation sequence form, and is applicable to high-power scenarios of field ground magnetic resonance detection.

[0038] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance, characterized in that The method includes: Laying a magnetic resonance detection coil on the ground, fixing a short emission time, changing the magnitude of the current amplitude to emit multiple short pulses, and collecting the short pulse observation signals generated by each short pulse; Fixing a long emission time, re-emitting a series of long pulses with different current amplitudes, and collecting the long pulse observation signals generated by each long pulse; Inverting the short pulse observation signals to obtain the water content and transverse relaxation time distribution of groundwater at different depths; Taking the water content and transverse relaxation time distribution of groundwater at different depths as the prior data for Bayesian inversion, and using Bayesian inversion to analyze the long pulse observation signals to obtain the longitudinal relaxation time distribution.

2. The groundwater nuclear magnetic resonance longitudinal relaxation time detection method according to claim 1, characterized in that, The inverting the short pulse observation signals to obtain the water content and transverse relaxation time distribution of groundwater at different depths includes: Establishing the objective function for regularization inversion as: ， In the formula, is a short-pulse inversion model representing the water content and transverse relaxation time distribution of groundwater at different depths, is the short-pulse nuclear magnetic resonance FID signal obtained by forward calculation, is a smoothness matrix describing depth correlation, is a regularization parameter used to balance data fitting and model smoothness, is a data normalization term, is the short-pulse observation signal; Iteratively solving the objective function to obtain the water content and transverse relaxation time distribution of groundwater at different depths.

3. The groundwater nuclear magnetic resonance longitudinal relaxation time detection method according to claim 2, wherein The calculation process of the short pulse magnetic resonance FID signal includes: Calculate the transverse magnetization vector under short pulses: , where is the transverse magnetization vector under short pulses at the spatial point , is the net magnetization vector, is the magnetization vector tipping angle at the spatial point ; Calculating the short pulse magnetic resonance FID signal in the time domain according to the transverse magnetization vector under the short pulse.

4. A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance according to claim 1, characterized in that Using Bayesian inversion to analyze the long pulse observation signals to obtain the longitudinal relaxation time distribution includes: Calculate the fitting error , represents the long-pulse nuclear magnetic resonance FID signal obtained from the forward calculation, and is the long-pulse observation signal; Randomly perturbing a region and re-sampling the longitudinal relaxation time distribution in the current Bayesian inversion model to obtain a new Bayesian inversion model; Calculate the fitting error of the new Bayesian inversion model , and determine whether to accept the new Bayesian inversion model in the form of simulated annealing; Continuously perturbing and re-sampling the longitudinal relaxation time distribution in the current Bayesian inversion model to form the state jump of the Markov chain until the maximum number of iterations is reached.

5. A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance according to claim 4, characterized in that, During the process of using Bayesian inversion to analyze the long pulse observation signals to obtain the longitudinal relaxation time distribution, the water content and transverse relaxation time distribution of groundwater at different depths are kept unchanged.

6. The groundwater nuclear magnetic resonance longitudinal relaxation time detection method according to claim 4, characterized in that The calculation process of the long pulse magnetic resonance FID signal includes: Searching for the slice with the closest combination of emission time and relaxation parameters in the interpolation library according to the emission time and the relaxation parameters obtained by Bayesian inversion, and obtaining the transverse magnetization vector under the long pulse by one-dimensional interpolation of the slice; Calculating the long pulse magnetic resonance FID signal in the time domain according to the transverse magnetization vector under the long pulse.

7. A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance according to claim 6, characterized in that, The interpolation library is formed by calculating the Bloch equation under different combinations of emission time, longitudinal relaxation time and transverse relaxation time with dense parameter intervals by the Runge-Kutta method, and obtaining the look-up table relationship between the transverse magnetization vector and the excitation field strength.

8. A method for detecting the longitudinal relaxation time of groundwater by nuclear magnetic resonance according to claim 1, characterized in that The short emission time refers to that the emission current duration in a pulse period is less than 50 ms; the long emission time refers to that the emission current duration in a pulse period is greater than 200 ms.

Images