CN107607998B - A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system - Google Patents
A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system Download PDFInfo
- Publication number
- CN107607998B CN107607998B CN201710874077.5A CN201710874077A CN107607998B CN 107607998 B CN107607998 B CN 107607998B CN 201710874077 A CN201710874077 A CN 201710874077A CN 107607998 B CN107607998 B CN 107607998B
- Authority
- CN
- China
- Prior art keywords
- matrix
- magnetic resonance
- observation vector
- obtains
- factor matrices
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The present invention discloses a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system.This method comprises: obtaining the Magnetic Resonance data at N number of moment;The Magnetic Resonance data at preceding N-1 moment and rear N-1 moment are taken to form two observation vectors respectively;Two observation vectors are merged, observing matrix is formed;Operation is carried out to observing matrix in conjunction with total least square method and invariable rotary parametric method, obtains rotation operator;Rotation operator is the function of average attenuation time;The average attenuation time is calculated by rotation operator;Using the average attenuation time, Magnetic Resonance data conversion will be surveyed into the form of sinusoidal signal plus noise;Fast Fourier transform is carried out to sinusoidal signal and obtains initial amplitude.Method and system disclosed by the invention improve the extraction accuracy of parameter.
Description
Technical field
The present invention relates to Magnetic Resonance fields, believe more particularly to a kind of nuclear magnetic resonance water detector magnetic resonance response
Number parameter extracting method and system.
Background technique
Water resource shortage is a significant problem in worldwide.In the 45th, the whole world Davos world in 2015
" the global risk report in 2015 " issued on Economic Forum is pointed out, load of the people to flood, arid, water pollution and insufficient water
Sorrow has been over the worry to nuclear weapon or global disease, and water crisis has been global first Risks factor.
Nuclear magnetic resonance underground water Detection Techniques are a kind of lossless detection technologies of direct look for water.With traditional exploring subsurface
The geophysical method of water is compared, and nuclear magnetic resonance underground water Detection Techniques have high-resolution, high efficiency, informative reconciliation
The advantages that uniqueness.1996, the commercialization version i.e. method for the first nuclear magnetic resonance underground water Detection Techniques in the world of being born
The nuclear magnetic induction system (NUMIS, Nuclearmagnetic induction system) that geologic survey office, state develops.China has
Unit is closed using the NUMIS and NUMIS of the France boughtplus(NUMISplusThe NUMIS upgrade version released for France in 1999)
Multiple water shortage provinces such as the Inner Mongol and area have carried out related experiment, wherein NUMIS detect depth capacity be 100 meters,
NUMISplusIt is 150 meters, and has been successfully found underground water, promotes the development of China's nuclear magnetic resonance underground water Detection Techniques.Mesh
Before, Jilin University, China has voluntarily developed the nuclear magnetic resonance water detector of model JLMRS, also has reached in performance indicator
The technical level of French NUMIS.
The basic principle of nuclear magnetic resonance water detector is the hydrogen made by transmitting coil to underground water excitation energy in underground water
Energy level transition occurs for proton electron outside nucleus, is then received when electron outside nucleus is moved back from high level to low-lying level and is released with receiving coil again
The energy put is rung to obtain magnetic resonance response (Magnetic Resonance Sounding, MRS) signal by the magnetic resonance
Induction signal can inverting obtain the relevant information of underground water.
The complex signal model of nuclear magnetic resonance water detector Magnetic Resonance is as follows:
Wherein, n is sampling instant, E0For initial amplitude, the size and underground water content of initial amplitude are directly proportional, also include
The information such as underground reservoir depth, thickness, unit volume water content.T2For average attenuation time (also referred to as relaxation time), put down
Equal die-away time reflects the information of underground reservoir average pore.The initial phase of Magnetic ResonanceIt reflects
The conductivity in water-bearing layer is a constant, ω0=2 π fl(flFor your frequency of Rameau) be earth's magnetic field angular frequency.It need to will only extract
The initial amplitude E arrived0With average die-away time T2The two parameters, underground water content can be obtained, depth, contain by carrying out inverting
The information such as water layer porosity.Since itself sensitivity of nuclear magnetic resonance water detector is higher, the useful signal of Magnetic Resonance
It is extremely faint, it is highly prone to the interference from the natural noises such as the man-made noises such as power line and iron ore rock layers.How from making an uproar
Initial amplitude E is extracted in the Magnetic Resonance of sound0With average die-away time T2The two parameters just become a key and ask
Topic.
First de-noising is mostly used to improve the extraction of the parameter of the Magnetic Resonance of nuclear magnetic resonance water detector at present
Signal-to-noise ratio, then carry out the strategy of curve parameter fitting.The noise-eliminating method of use have very much, as the addition method, digital filtering, small echo,
Adaptive-filtering, Higher Order Cumulants etc., after denoising, the signal-to-noise ratio of the Magnetic Resonance of nuclear magnetic resonance water detector
It is centainly promoted, recycles the NUMERICAL MATCH METHOD FORs such as linear fit, nonlinear fitting, fit magnetic resonance response curve, after
And find out the parameter of water detector Magnetic Resonance.
The parameter extraction technical spirit of the Magnetic Resonance of existing nuclear magnetic resonance water detector is based on classical signal
The strategy of processing.However no matter any noise-eliminating method can only all inhibit noise, and noise can not be completely eliminated, by
This necessarily gives extraction result to bring error, and subsequent NUMERICAL MATCH METHOD FOR can further generate error, and thus two-stage error must
The extraction accuracy of the parameter of the Magnetic Resonance of nuclear magnetic resonance water detector can so be reduced.
Summary of the invention
The object of the present invention is to provide a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system,
Improve the extraction accuracy of the two parameters of average attenuation time and initial amplitude.
To achieve the above object, the present invention provides following schemes:
A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method, comprising:
Obtain the actual measurement Magnetic Resonance data at N number of moment;
In the actual measurement Magnetic Resonance data, the actual measurement Magnetic Resonance data at N-1 moment before taking
The first observation vector is formed, the actual measurement Magnetic Resonance data at rear N-1 moment is taken to form the second observation vector;Described
The signal subspace of one observation vector is identical as the dimension of signal subspace of second observation vector;
First observation vector and second observation vector are merged, observing matrix is formed;
Operation is carried out to the observing matrix in conjunction with total least square method and invariable rotary parametric method, rotation is obtained and calculates
Son;The rotation operator is letter of the signal subspace of second observation vector relative to first observation vector
The space angle in work song space, the rotation operator are the function of average attenuation time;
The average attenuation time is calculated by the rotation operator;The average attenuation time is for reflecting that underground contains
The information of water layer average pore;
Using the average attenuation time, by the actual measurement Magnetic Resonance data conversion at sinusoidal signal plus noise
Form;
Fast Fourier transform is carried out to the sinusoidal signal and obtains initial amplitude, the initial amplitude includes underground water content
Information.
Optionally, the combination total least square method and invariable rotary parametric method carry out operation to the observing matrix,
Rotation operator is obtained, is specifically included:
Auto-correlation computation is carried out to the observing matrix, obtains the autocorrelation matrix of the observing matrix;
Eigenvalues Decomposition is carried out to the autocorrelation matrix of the observing matrix, obtains two characteristic values;
Determine that characteristic vector corresponding to the larger characteristic value in described two characteristic values is signal subspace feature vector;
The signal subspace feature vector includes first part and second part;The first part and the first observation vector phase
Corresponding, the second part is corresponding with second observation vector;
Factor matrices ψ is calculated according to the first part and second part;The factor matrices ψ is to guarantee to make described second
Part is equal to the matrix of the first part multiplied by the result of the factor matrices ψ;
Eigenvalues Decomposition is carried out to the factor matrices ψ, obtains the rotation operator.
Optionally, described that factor matrices ψ is calculated according to the first part and second part, it specifically includes:
By the first part and second part construction autocorrelation matrix and singular value decomposition is carried out, obtains left unusual square
Battle array;
Factor matrices ψ is calculated using the submatrix of the left singular matrix.
Optionally, described that autocorrelation matrix is constructed by the first part and second part and carries out singular value decomposition, it obtains
To left singular matrix, specifically include:
The first part and the second part are constructed into autocorrelation matrix, obtain signal subspace autocorrelation matrix;
Singular value decomposition is carried out to the signal subspace autocorrelation matrix, obtains left singular matrix and right singular matrix;
The left singular matrix and the right singular matrix are 2 × 2 dimension matrixes.
Optionally, the submatrix using the left singular matrix calculates factor matrices ψ, specifically includes:
The left singular matrix U is decomposed into four 1 × 1 dimension minor matrixs, respectively one column minor matrix U of the first row11, first
Two column minor matrix U of row12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
According to the two column minor matrix U of the first row12With the two column minor matrix U of the second row22Calculate the factor matrices ψ
Ψ=- U12U22 -1。
The invention also discloses a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction systems, comprising:
Module is obtained, for obtaining the actual measurement Magnetic Resonance data at N number of moment;
Observation vector constructs module, the reality for the N-1 moment before in the actual measurement Magnetic Resonance data, taking
It surveys Magnetic Resonance data and forms the first observation vector, take the actual measurement Magnetic Resonance data group at rear N-1 moment
At the second observation vector;The signal subspace that the signal subspace and second observation vector of first observation vector are opened
The dimension in space is identical;
Observation vector merging module, for first observation vector and second observation vector to be merged, shape
At observing matrix;
Rotation operator computing module is used in conjunction with total least square method and invariable rotary parametric method to the observing matrix
Operation is carried out, rotation operator is obtained;The rotation operator is that the signal subspace of second observation vector is relatively described
The space angle of the signal subspace of first observation vector, the rotation operator are the function of average attenuation time;
Average attenuation time computing module, for the average attenuation time to be calculated by the rotation operator;It is described
The average attenuation time is used to reflect the information of underground reservoir average pore;
Form transformation module turns the actual measurement Magnetic Resonance data for utilizing the average attenuation time
Change the form of sinusoidal signal plus noise into;
Initial amplitude computing module obtains initial amplitude for carrying out fast Fourier transform to the sinusoidal signal, described
Initial amplitude includes underground water content information.
Optionally, the rotation operator computing module, specifically includes:
Auto-correlation computation submodule obtains the observing matrix for carrying out auto-correlation computation to the observing matrix
Autocorrelation matrix;
The First Eigenvalue decomposes submodule, carries out Eigenvalues Decomposition for the autocorrelation matrix to the observing matrix, obtains
To two characteristic values;
Feature vector determines submodule, for determining Characteristic Vectors corresponding to the larger characteristic value in described two characteristic values
Amount is signal subspace feature vector;The signal subspace feature vector includes first part and second part;Described first
Part is corresponding with first observation vector, and the second part is corresponding with second observation vector;
Factor matrices computational submodule, for calculating factor matrices ψ according to the first part and second part;It is described because
Matrix number ψ is the matrix for guaranteeing that the second part is made to be equal to the first part multiplied by the result of the factor matrices ψ;
Second Eigenvalue decomposes submodule, for carrying out Eigenvalues Decomposition to the factor matrices ψ, obtains the rotation and calculates
Son.
Optionally, the factor matrices computational submodule, specifically includes:
Singular value decomposition unit, for constructing autocorrelation matrix by the first part and second part and carrying out singular value
It decomposes, obtains left singular matrix;
Factor matrices computing unit, for calculating factor matrices ψ using the submatrix of the left singular matrix.
Optionally, the singular value decomposition unit, specifically includes:
Auto-correlation subelement obtains signal for the first part and the second part to be constructed autocorrelation matrix
Subspace autocorrelation matrix;
Singular value decomposition subelement obtains a left side for carrying out singular value decomposition to the signal subspace autocorrelation matrix
Singular matrix and right singular matrix;The left singular matrix and the right singular matrix are 2 × 2 dimension matrixes.
Optionally, the factor matrices computing unit, specifically includes:
Matrix decomposition subelement, for the left singular matrix U to be decomposed into four 1 × 1 dimensions minor matrixs, respectively first
One column minor matrix U of row11, two column minor matrix U of the first row12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
Factor matrices computation subunit, for according to the two column minor matrix U of the first row12Small square is arranged with second row two
Battle array U22Calculate the factor matrices ψ
Ψ=- U12U22 -1。
The specific embodiment provided according to the present invention, the invention discloses following technical effects: one kind disclosed by the invention
In the technical solution of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system, the preceding N-1 moment is utilized
Magnetic Resonance data form the first observation vector and the Magnetic Resonance data at rear N-1 moment composition second is seen
Invariable rotary relationship between direction finding amount determines rotation operator, and utilization rotation operator and the relationship between the average attenuation time,
Rotation operator is calculated to extract the average attenuation time in conjunction with total least square method and invariable rotary parametric method, is realized and is being gone
The average attenuation time is directly calculated while except noise, is improved the precision of parameter extraction, is reduced operand.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the method flow of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
Figure;
Fig. 2 is that the calculating of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention rotates calculation
The method flow diagram of submethod;
Fig. 3 is the calculating factor square of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
The method flow diagram of battle array method;
Fig. 4 is to extract averagely to decline in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
Subtract the emulation experiment curve of time parameter;The wherein curve that (a) partially changes for the mean value of average attenuation time with signal-to-noise ratio,
(b) curve partially changed for the variance of average attenuation time with signal-to-noise ratio;
Fig. 5 is that initial vibration is extracted in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
The emulation experiment curve of width parameter;Wherein (a) is partially the curve that changes with signal-to-noise ratio of mean value of initial amplitude, (b) is partially
The curve that the variance of initial amplitude changes with signal-to-noise ratio;
Fig. 6 is the system structure of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system embodiment of the present invention
Figure.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
When two parameters of most critical are average attenuations in the Magnetic Resonance x (n) that nuclear magnetic resonance water detector obtains
Between T2With initial amplitude E0.The present invention uses total least square (TLS, Total least squares) and invariable rotary parameter
Method (ESPRIT, Estimating Signal Parameter via Rotational Invariance Techniques) mentions
Take average die-away time T2, initial amplitude is calculated using fast Fourier transform (FFT, Fast FourierTransformation)
E0。
In the actual measurement Magnetic Resonance data x (n) that nuclear magnetic resonance water detector obtains, it can generally be attached with white Gaussian
Noise η (n) then surveys Magnetic Resonance data x (n) expression are as follows:
X (n)=s (n)+η (n) (2)
Wherein n is sampling instant, and s (n) is the true Magnetic Resonance removed after noise;E0For initial amplitude,
The size of the initial amplitude is directly proportional to underground water content, also contains underground reservoir depth, thickness, unit volume water content
Etc. information;T2For average attenuation time, also referred to as relaxation time, the average attenuation time reflects underground reservoir average pore
Information;For initial phase, which reflects the conductivity in water-bearing layer, is a constant;ω0For the angular frequency in earth's magnetic field
Rate, ω0=2 π fl, flFor Rameau that frequency.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
Fig. 1 is the method flow of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
Figure.
Referring to Fig. 1, the nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method, comprising:
Step 1: obtaining the actual measurement Magnetic Resonance data at N number of moment, respectively x (n1)、x(n2)、…、x(nN),
The value of n is successively defined as n1、n2、…、nN, then
Step 2: in the actual measurement Magnetic Resonance data, the actual measurement magnetic resonance response letter at N-1 moment before taking
Number forms the first observation vector X, and the actual measurement Magnetic Resonance data at rear N-1 moment is taken to form the second observation vector
Y;Then
Wherein,Expression is defined as;There is translation relation, W between first observation vector X and the second observation vector YxFor
White Gaussian noise in first observation vector X, WyFor the white Gaussian noise in the second observation vector X;A is direction vector, includes
The information of frequency and average die-away time;Γ is width phase information matrix, which is to contain amplitude and phase
The matrix of information, width phase information matrix only one element γ,
Rotation operator Φ is defined, i.e.,
Then have
As it can be seen that the first observation vector X and the second observation vector Y is opened into the identical signal subspace of dimension, seen by first
Direction finding amount X and the second observation vector Y signal subspaces are it is found that the second observation vector Y is revolved relative to the first observation vector X
Rotation operator Φ is turned, and rotation operator Φ is average attenuation time T2Function.The present invention is exactly to utilize the second observation
The invariable rotary relationship of vector and the first observation vector extracts average attenuation time T2。
Step 3: the first observation vector X and the second observation vector Y being merged, observing matrix Z is formed;
Wherein
For direction matrix, WzFor noise matrix;Observing matrix Z, direction matrixWith noise matrix WzIt is 2 × (N-1)
Dimension.
Step 4: operation being carried out to the observing matrix Z in conjunction with total least square method and invariable rotary parametric method, is obtained
Rotation operator.
Fig. 2 is that the calculating of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention rotates calculation
The method flow diagram of submethod.
Referring to fig. 2, which specifically includes:
Step 401: auto-correlation computation being carried out to the observing matrix Z, obtains the autocorrelation matrix R of the observing matrixzz;
Wherein I is unit matrix, σ2For the variance of white Gaussian noise.
Auto-correlation computation equally is carried out to width phase information matrix Γ, obtains the autocorrelation matrix R of width phase information matrix ΓΓΓ,
RΓΓ=E { Γ ΓH}。
Step 402: to the autocorrelation matrix R of the observing matrixzzEigenvalues Decomposition is carried out, two eigenvalue λs are obtained1With
λ2;I.e.
Wherein λiIt is characterized value, eiIt is characterized value λiCorresponding feature vector;I is characterized the corresponding serial number of value;ΛsIt is right
Angular moment battle array;EsIt is characterized value λ1Corresponding characteristic vector, ENIt is characterized value λ2Corresponding characteristic vector;
In conjunction with formula (15) and formula (16), since Magnetic Resonance is simple component, then eigenvalue λ1> λ2, then
There is λ1=Λs, λ2=σ2, eigenvalue λ1Corresponding characteristic vector Es=[e1] open into signal subspace, eigenvalue λ2Corresponding feature
Vector EN=[e2] open into noise subspace, the noise subspace and eigenvalue λ1Corresponding characteristic vector EsThe signal subspace opened
Space has orthogonality relation.It means that existence anduniquess, it is nonsingular 1 × 1 dimension non-singular matrix T, set up following formula:
Step 403: determining the larger eigenvalue λ in described two characteristic values1Corresponding characteristic vector EsFor signal subspace sky
Between feature vector;The signal subspace feature vector includes first part ExWith second part Ey;The first part ExWith institute
It is corresponding to state the first observation vector X, the second part EyIt is corresponding with the second observation vector Y;
Specifically: it can be by E by the rotational invariance of observation vectorsIt is decomposed into ExAnd EyTwo parts respectively correspond the first observation
Vector X and the second observation vector Y;I.e.
Wherein ExAnd EyIt is 1 × 1 dimension matrix.
Step 404: according to the first part ExWith the second part EyCalculate factor matrices ψ;The factor matrices ψ
To guarantee to make the second part EyIt is equal to the first part E multiplied by the result of the factor matrices ψxMatrix, i.e., using total
Body least square method seeks equation Ey=ExThe solution of Ψ.
Fig. 3 is the calculating factor square of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
The method flow diagram of battle array method.
Referring to Fig. 3, which is specifically included:
Step 4041: by the first part ExWith the second part EyConstruct autocorrelation matrix RsAnd carry out singular value
It decomposes, obtains left singular matrix U;It is realized especially by following procedure:
Enable Ψ=T-1Φ T, then can be obtained:
Ey=ExT-1Φ T=ExΨ (19)
In view of first part E described in equation two sidesxWith the second part EyBoth sides error, using singular value point
Solution combines total least square method, and the first part and the second part are constructed autocorrelation matrix, obtains signal subspace sky
Between autocorrelation matrix, and to the signal subspace autocorrelation matrix carry out singular value decomposition, obtain left singular matrix and right surprise
Different matrix;The left singular matrix and the right singular matrix are 2 × 2 dimension matrixes, then
Wherein, S is singular value matrix, and U is left singular matrix, and V is right singular matrix, and S, U and V are 2 × 2 dimension matrixes.
Step 4042: calculating factor matrices ψ using the submatrix of the left singular matrix;
The left singular matrix U is decomposed into four 1 × 1 dimension minor matrixs, respectively one column minor matrix U of the first row11, first
Two column minor matrix U of row12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
According to the two column minor matrix U of the first row12With the two column minor matrix U of the second row22Calculate the factor matrices ψ
Ψ=- U12U22 -1 (22)
Step 405: Eigenvalues Decomposition being carried out to the factor matrices ψ, obtains the rotation operator Φ.
By formula (19) it is found that rotation operator Φ is the characteristic value of factor matrices ψ, as long as therefore being carried out to factor matrices ψ
Eigenvalues Decomposition, gained characteristic value are rotation operator Φ.
Step 5: the calculated value of average attenuation time being calculated by the rotation operator ΦWhen with the average attenuation
Between calculated valueReflect the information of underground reservoir average pore;
By formula (10) it is found that rotation operator Φ is average attenuation time T2Function, therefore only require to obtain rotation operator
Φ can be obtained by the calculated value of average attenuation timeI.e.
And it can use the angular frequency that rotation operator calculates earth's magnetic field, obtain the calculated value of earth's magnetic field angular frequencyI.e.
In practice, when detecting underground water with nuclear magnetic resonance technique, the actual value ω of the angular frequency in earth's magnetic field0It is known
, therefore pass through the calculated value of contrastively magnetic field angular frequencyWith the actual value ω of earth's magnetic field angular frequency0It can verification test
As a result accuracy.
Step 6: utilizing the average attenuation time, the actual measurement Magnetic Resonance data x (n) is converted into sine
The form of signal plus noise;
The calculated value of the average attenuation time obtained in steps of 5It is the true value T of average attenuation time2Approximation
Value.In practice, the true value T of average attenuation time2It is unable to get, calculation method provided by the invention can make average
The calculated value of die-away timeCloser to the true value T of average attenuation time2.The step 6 of the invention is actually to utilize
The calculated value of average attenuation timeThe form that Magnetic Resonance data x (n) is converted into sinusoidal signal plus noise will be surveyed,
Specifically:
The equal sign two sides of formula (2) simultaneously multiplied byIt can obtain
At this point, actual measurement Magnetic Resonance data are converted into the form of sinusoidal signal plus noise.
Step 7: fast Fourier transform being carried out to the sinusoidal signal and obtains initial amplitude;The initial amplitude is for reflecting
Underground water content information.
The principle that sinusoidal signal amplitude is calculated using fast Fourier transform carries out fast Fourier transform to formula (25), can
To obtain initial amplitude.Specifically:
Obtain the real part of the sinusoidal signal;
Fast Fourier transform is carried out to the real part of the sinusoidal signal, obtains the calculated value of initial amplitude.
In a kind of technical solution of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method disclosed by the invention,
Utilize the magnetic resonance response of Magnetic Resonance data composition first observation vector and rear N-1 moment at preceding N-1 moment
Signal data forms the invariable rotary relationship between the second observation vector, determines rotation operator, and using rotation operator and averagely
It is average to extract to calculate rotation operator in conjunction with total least square method and invariable rotary parametric method for relationship between die-away time
Die-away time needs not move through first de-noising the two steps of extracting parameter again compared with existing parameter extracting method, but
The average attenuation time is directly calculated while removing noise, the precision of parameter extraction is improved, reduces operand.
Emulation experiment is carried out to the method for extracting the average attenuation time in this embodiment of the invention below:
Take Magnetic Resonance data parameters: initial amplitude E0=300nv, average attenuation time T2=10, Rameau
You are frequency f0=2326Hz, initial phaseSample frequency fs=10000Hz, sampling number N=30000.
White Gaussian noise is added, different signal-to-noise ratio is taken to repeat 100 experiments.
Table 1 is to extract averagely to decline in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
Subtract the experimental result table of the emulation experiment of time parameter.
The experimental result table of the emulation experiment of 1 average attenuation time parameter of table
Fig. 4 is to extract averagely to decline in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
Subtract the emulation experiment curve of time parameter;The wherein curve that (a) partially changes for the mean value of average attenuation time with signal-to-noise ratio,
(b) curve partially changed for the variance of average attenuation time with signal-to-noise ratio.
Referring to table 1 and Fig. 4, from simulation result as can be seen that when additive gaussian white noise, the embodiment in the present invention
Method can effectively extract average attenuation time parameter.When signal-to-noise ratio is higher than -5dB, extraction accuracy and extraction stability are all fine.
As signal-to-noise ratio reduces, extracts performance and declined, but the range of decrease is not quickly.When signal-to-noise ratio is -10dB, the average attenuation of extraction
The mean value error of time is 0.12%, and the variance of the average attenuation time of extraction is 0.03.Even if working as signal-to-noise ratio down to -20dB
When, the error of the mean value of the average attenuation time of extraction is also only 0.47%, and the variance of the average attenuation time of extraction is 0.25,
Still within the acceptable range.
Emulation experiment is carried out to the method for extracting initial amplitude in this embodiment of the invention below:
Take Magnetic Resonance data parameters: initial amplitude E0=300nv, average attenuation time T2=10, Rameau
You are frequency f0=2326Hz, initial phaseSample frequency fs=10000Hz, sampling number N=30000.
It selects to carry out the points of fast Fourier transform to be NF=5000, at this time spectral resolution beCalculating resolution isAnd Rameau that frequency f0=2326Hz, both
Be the integral multiple of spectral resolution, and be the integral multiple of calculating resolution, thus theoretically fast Fourier transform method in amplitude
Extracting aspect is that error, extraction error are not mainly derived from the extraction error and the influence of noise of average attenuation time.
Additive gaussian white noise takes different signal-to-noise ratio to repeat 100 experiments.
Table 2 is that initial vibration is extracted in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
The experimental result table of the emulation experiment of width parameter.
The experimental result table of the emulation experiment of 2 initial amplitude parameter of table
Fig. 5 is that initial vibration is extracted in nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method embodiment of the present invention
The emulation experiment curve of width parameter;Wherein (a) is partially the curve that changes with signal-to-noise ratio of mean value of initial amplitude, (b) is partially
The curve that the variance of initial amplitude changes with signal-to-noise ratio.
Referring to table 2 and Fig. 5, from simulation result as can be seen that when additive gaussian white noise, fast Fourier transform method can
Effectively to extract initial amplitude parameter, extraction accuracy and stability are all preferable.
Fig. 6 is the system structure of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system embodiment of the present invention
Figure.
Referring to Fig. 6, the nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system, comprising:
Module 801 is obtained, for obtaining the actual measurement Magnetic Resonance data at N number of moment;
Observation vector constructs module 802, for the N-1 moment before in the actual measurement Magnetic Resonance data, taking
Actual measurement Magnetic Resonance data form the first observation vector, take the actual measurement Magnetic Resonance number at rear N-1 moment
According to forming the second observation vector;The letter that the signal subspace and second observation vector of first observation vector are opened
The dimension in work song space is identical;
Observation vector merging module 803, for first observation vector and second observation vector to be merged,
Form observing matrix;
Rotation operator computing module 804 is used in conjunction with total least square method and invariable rotary parametric method to the observation
Matrix carries out operation, obtains rotation operator;The rotation operator is that the signal subspace of second observation vector is opposite
The space angle of the signal subspace of first observation vector, the rotation operator are the function of average attenuation time;
Average attenuation time computing module 805, for the average attenuation time to be calculated by the rotation operator;Institute
The average attenuation time is stated for reflecting the information of underground reservoir average pore;
Form transformation module 806, for utilizing the average attenuation time, by the actual measurement Magnetic Resonance data
It is converted into the form of sinusoidal signal plus noise;
Initial amplitude computing module 807 obtains initial amplitude for carrying out fast Fourier transform to the sinusoidal signal, institute
Stating initial amplitude includes underground water content information.
The rotation operator computing module 804, specifically includes:
Auto-correlation computation submodule obtains the observing matrix for carrying out auto-correlation computation to the observing matrix
Autocorrelation matrix;
The First Eigenvalue decomposes submodule, carries out Eigenvalues Decomposition for the autocorrelation matrix to the observing matrix, obtains
To two characteristic values;
Feature vector determines submodule, for determining Characteristic Vectors corresponding to the larger characteristic value in described two characteristic values
Amount is signal subspace feature vector;The signal subspace feature vector includes first part and second part;Described first
Part is corresponding with first observation vector, and the second part is corresponding with second observation vector;
Factor matrices computational submodule, for calculating factor matrices ψ according to the first part and second part;It is described because
Matrix number ψ is the matrix for guaranteeing that the second part is made to be equal to the first part multiplied by the result of the factor matrices ψ;
Second Eigenvalue decomposes submodule, for carrying out Eigenvalues Decomposition to the factor matrices ψ, obtains the rotation and calculates
Son.
The factor matrices computational submodule, specifically includes:
Singular value decomposition unit, for constructing autocorrelation matrix by the first part and second part and carrying out singular value
It decomposes, obtains left singular matrix;
Factor matrices computing unit, for calculating factor matrices ψ using the submatrix of the left singular matrix.
The singular value decomposition unit, specifically includes:
Auto-correlation subelement obtains signal for the first part and the second part to be constructed autocorrelation matrix
Subspace autocorrelation matrix;
Singular value decomposition subelement obtains a left side for carrying out singular value decomposition to the signal subspace autocorrelation matrix
Singular matrix and right singular matrix;The left singular matrix and the right singular matrix are 2 × 2 dimension matrixes.
The factor matrices computing unit, specifically includes:
Matrix decomposition subelement, for the left singular matrix U to be decomposed into four 1 × 1 dimensions minor matrixs, respectively first
One column minor matrix U of row11, two column minor matrix U of the first row12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
Factor matrices computation subunit, for according to the two column minor matrix U of the first row12Small square is arranged with second row two
Battle array U22Calculate the factor matrices ψ
Ψ=- U12U22 -1。
In a kind of technical solution of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system disclosed by the invention,
Utilize the magnetic resonance response of Magnetic Resonance data composition first observation vector and rear N-1 moment at preceding N-1 moment
Signal data forms the invariable rotary relationship between the second observation vector, determines rotation operator, and using rotation operator and averagely
It is average to extract to calculate rotation operator in conjunction with total least square method and invariable rotary parametric method for relationship between die-away time
Die-away time needs not move through first de-noising the two steps of extracting parameter again compared with existing parameter extracting method, but
The average attenuation time is directly calculated while removing noise, the precision of parameter extraction is improved, reduces operand.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For system disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Used herein a specific example illustrates the principle and implementation of the invention, and above embodiments are said
It is bright to be merely used to help understand method and its core concept of the invention;At the same time, for those skilled in the art, foundation
Thought of the invention, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as limitation of the present invention.
Claims (10)
1. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method, comprising: the actual measurement magnetic for obtaining N number of moment is total
Vibration response signal data;It is characterized by further comprising:
In the actual measurement Magnetic Resonance data, the actual measurement Magnetic Resonance data composition at N-1 moment before taking
First observation vector takes the actual measurement Magnetic Resonance data at rear N-1 moment to form the second observation vector;Described first sees
The signal subspace of direction finding amount is identical as the dimension of signal subspace of second observation vector;
First observation vector and second observation vector are merged, observing matrix is formed;
Operation is carried out to the observing matrix in conjunction with total least square method and invariable rotary parametric method, obtains rotation operator;Institute
State the signal subspace that the signal subspace that rotation operator is second observation vector is opened relative to first observation vector
The space angle in space, the rotation operator are the function of average attenuation time;
The average attenuation time is calculated by the rotation operator;The average attenuation time is for reflecting underground reservoir
The information of average pore;
Using the average attenuation time, by the actual measurement Magnetic Resonance data conversion at the shape of sinusoidal signal plus noise
Formula;
Fast Fourier transform is carried out to the sinusoidal signal and obtains initial amplitude, the initial amplitude is believed comprising underground water content
Breath.
2. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method according to claim 1, feature
It is, the combination total least square method and invariable rotary parametric method carry out operation to the observing matrix, obtain rotation and calculate
Son specifically includes:
Auto-correlation computation is carried out to the observing matrix, obtains the autocorrelation matrix of the observing matrix;
Eigenvalues Decomposition is carried out to the autocorrelation matrix of the observing matrix, obtains two characteristic values;
Determine that characteristic vector corresponding to the larger characteristic value in described two characteristic values is signal subspace feature vector;It is described
Signal subspace feature vector includes first part and second part;The first part is opposite with first observation vector
It answers, the second part is corresponding with second observation vector;
Factor matrices ψ is calculated according to the first part and second part;The factor matrices ψ is to guarantee to make the second part
It is equal to the matrix of the first part multiplied by the result of the factor matrices ψ;
Eigenvalues Decomposition is carried out to the factor matrices ψ, obtains the rotation operator.
3. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method according to claim 2, feature
It is, it is described that factor matrices ψ is calculated according to the first part and second part, it specifically includes:
By the first part and second part construction autocorrelation matrix and singular value decomposition is carried out, obtains left singular matrix;
Factor matrices ψ is calculated using the submatrix of the left singular matrix.
4. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method according to claim 3, feature
It is, it is described that autocorrelation matrix is constructed by the first part and second part and carries out singular value decomposition, obtain left unusual square
Battle array, specifically includes:
The first part and the second part are constructed into autocorrelation matrix, obtain signal subspace autocorrelation matrix;
Singular value decomposition is carried out to the signal subspace autocorrelation matrix, obtains left singular matrix and right singular matrix;It is described
Left singular matrix and the right singular matrix are 2 × 2 dimension matrixes.
5. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method according to claim 3, feature
It is, the submatrix using the left singular matrix calculates factor matrices ψ, it specifically includes:
The left singular matrix U is decomposed into four 1 × 1 dimension minor matrixs, respectively one column minor matrix U of the first row11, the first row two
Column minor matrix U12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
According to the two column minor matrix U of the first row12With the two column minor matrix U of the second row22Calculate the factor matrices ψ
Ψ=- U12U22 -1。
6. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system, comprising: module is obtained, it is N number of for obtaining
The actual measurement Magnetic Resonance data at moment;It is characterized by further comprising:
Observation vector constructs module, the actual measurement magnetic for the N-1 moment before in the actual measurement Magnetic Resonance data, taking
Resonance response signal data forms the first observation vector, takes the actual measurement Magnetic Resonance data composition the at rear N-1 moment
Two observation vectors;The signal subspace that the signal subspace and second observation vector of first observation vector are opened
Dimension it is identical;
Observation vector merging module is formed and is seen for merging first observation vector and second observation vector
Survey matrix;
Rotation operator computing module, for being carried out in conjunction with total least square method and invariable rotary parametric method to the observing matrix
Operation obtains rotation operator;The rotation operator is the signal subspace relatively described first of second observation vector
The space angle of the signal subspace of observation vector, the rotation operator are the function of average attenuation time;
Average attenuation time computing module, for the average attenuation time to be calculated by the rotation operator;It is described average
Die-away time is used to reflect the information of underground reservoir average pore;
Form transformation module, for utilize the average attenuation time, by the actual measurement Magnetic Resonance data conversion at
The form of sinusoidal signal plus noise;
Initial amplitude computing module obtains initial amplitude for carrying out fast Fourier transform to the sinusoidal signal, described initial
Amplitude includes underground water content information.
7. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system according to claim 6, feature
It is, the rotation operator computing module specifically includes:
Auto-correlation computation submodule, for carrying out auto-correlation computation to the observing matrix, obtain the observing matrix from phase
Close matrix;
The First Eigenvalue decomposes submodule, carries out Eigenvalues Decomposition for the autocorrelation matrix to the observing matrix, obtains two
A characteristic value;
Feature vector determines submodule, for determining that characteristic vector corresponding to the larger characteristic value in described two characteristic values is
Signal subspace feature vector;The signal subspace feature vector includes first part and second part;The first part
Corresponding with first observation vector, the second part is corresponding with second observation vector;
Factor matrices computational submodule, for calculating factor matrices ψ according to the first part and second part;The factor square
Battle array ψ is the matrix for guaranteeing that the second part is made to be equal to the first part multiplied by the result of the factor matrices ψ;
Second Eigenvalue decomposes submodule, for carrying out Eigenvalues Decomposition to the factor matrices ψ, obtains the rotation operator.
8. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system according to claim 7, feature
It is, the factor matrices computational submodule specifically includes:
Singular value decomposition unit, for constructing autocorrelation matrix by the first part and second part and carrying out singular value point
Solution, obtains left singular matrix;
Factor matrices computing unit, for calculating factor matrices ψ using the submatrix of the left singular matrix.
9. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system according to claim 8, feature
It is, the singular value decomposition unit specifically includes:
Auto-correlation subelement obtains signal subspace sky for the first part and the second part to be constructed autocorrelation matrix
Between autocorrelation matrix;
Singular value decomposition subelement obtains left unusual for carrying out singular value decomposition to the signal subspace autocorrelation matrix
Matrix and right singular matrix;The left singular matrix and the right singular matrix are 2 × 2 dimension matrixes.
10. a kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extraction system according to claim 8, feature
It is, the factor matrices computing unit specifically includes:
Matrix decomposition subelement, for the left singular matrix U to be decomposed into four 1 × 1 dimension minor matrixs, respectively the first row one
Column minor matrix U11, two column minor matrix U of the first row12, one column minor matrix U of the second row21With two column minor matrix U of the second row22
Factor matrices computation subunit, for according to the two column minor matrix U of the first row12With the two column minor matrix U of the second row22
Calculate the factor matrices ψ
Ψ=- U12U22 -1。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710874077.5A CN107607998B (en) | 2017-09-25 | 2017-09-25 | A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710874077.5A CN107607998B (en) | 2017-09-25 | 2017-09-25 | A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107607998A CN107607998A (en) | 2018-01-19 |
CN107607998B true CN107607998B (en) | 2019-02-26 |
Family
ID=61058025
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710874077.5A Active CN107607998B (en) | 2017-09-25 | 2017-09-25 | A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107607998B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109613617B (en) * | 2019-01-24 | 2020-02-07 | 吉林大学 | Underground water detection method and system based on magnetic resonance response signal parameter extraction |
CN111239837B (en) * | 2020-02-20 | 2021-01-05 | 吉林大学 | Ground magnetic resonance signal parameter extraction method based on MCMC |
CN112882111B (en) * | 2021-01-18 | 2022-05-03 | 吉林大学 | Magnetic resonance response signal parameter extraction method and system based on cyclic correlation |
CN115097533B (en) * | 2022-05-05 | 2023-06-30 | 吉林大学 | Magnetic resonance sounding signal extraction method based on TLS-ESPRIT algorithm |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6791323B2 (en) * | 2001-10-16 | 2004-09-14 | Cornell Research Foundation, Inc. | Method and apparatus for measuring and correcting motion effects using navigator echoes |
US7372263B2 (en) * | 2005-11-23 | 2008-05-13 | Baker Hughes Incorporated | Apparatus and method for measuring cased hole fluid flow with NMR |
CN102608663A (en) * | 2012-03-23 | 2012-07-25 | 西安电子科技大学 | Interference canceller applied to detecting core quadrupole moment resonance signal |
US9223055B2 (en) * | 2012-10-23 | 2015-12-29 | Schlumberger Technology Corporation | Identifying formation, matrix and fluid related characteristics from subsurface data using factor analysis |
CN203502602U (en) * | 2013-10-05 | 2014-03-26 | 吉林大学 | Nuclear magnetic resonance detection device eliminating power frequency harmonic interference |
CN104765072B (en) * | 2015-03-25 | 2017-11-24 | 吉林大学 | A kind of method for carrying out magnetic resonance forward probe with loop aerial rotation |
CN104777442B (en) * | 2015-04-07 | 2017-06-16 | 吉林大学 | A kind of nuclear magnetic resonance depth measurement FID signal noise suppressing method |
-
2017
- 2017-09-25 CN CN201710874077.5A patent/CN107607998B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN107607998A (en) | 2018-01-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107607998B (en) | A kind of nuclear magnetic resonance water detector Magnetic Resonance parameter extracting method and system | |
CN100385253C (en) | High-resolution radon transform for processing seismic data | |
Simons et al. | Spatiospectral localization of isostatic coherence anisotropy in Australia and its relation to seismic anisotropy: Implications for lithospheric deformation | |
Lipphardt Jr et al. | Blending HF radar and model velocities in Monterey Bay through normal mode analysis | |
Zhdanov et al. | Large-scale 3D inversion of marine magnetotelluric data: Case study from the Gemini prospect, Gulf of Mexico | |
CN105954802B (en) | A kind of conversion method and device of lithology data body | |
Duchkov et al. | Discrete almost-symmetric wave packets and multiscale geometrical representation of (seismic) waves | |
Zhao et al. | Validation of hydraulic tomography in an unconfined aquifer: A controlled sandbox study | |
Lekien et al. | Open‐boundary modal analysis: Interpolation, extrapolation, and filtering | |
Donelan et al. | A comparison of methods for estimating directional spectra of surface waves | |
Torbert et al. | A new method of 3‐D magnetic field reconstruction | |
CN108663653A (en) | Wave arrival direction estimating method based on L-shaped Electromagnetic Vector Sensor Array | |
CN101911099A (en) | Be used for analyzing the method and system of the singularity of digital signal | |
Adetunji et al. | Reexamination of magnetotelluric responses and electrical anisotropy of the lithospheric mantle in the Grenville Province, Canada | |
Irons et al. | Pulse and Fourier transform surface nuclear magnetic resonance: comprehensive modelling and inversion incorporating complex data and static dephasing dynamics | |
CN109143151A (en) | The uniform surface battle array tensor reconstructing method and signal source locating method of part array element damage | |
Müller-Petke | Non-remote reference noise cancellation-using reference data in the presence of surface-NMR signals | |
CN109613617B (en) | Underground water detection method and system based on magnetic resonance response signal parameter extraction | |
Dalgaard et al. | Enhancing SNMR model resolution by selecting an optimum combination of pulse moments, stacking, and gating | |
Bibby | Direct current resistivity modeling for axially symmetric bodies using the finite element method | |
Yan et al. | Multidirectional eigenvalue-based coherence attribute for discontinuity detection | |
You et al. | Multiscale ambient noise tomography of short-period Rayleigh waves across northern Taiwan | |
Wendel et al. | Source of the bursty bulk flow diffuse aurora: Electrostatic cyclotron harmonic and whistler waves in the coupling of bursty bulk flows to auroral precipitation | |
Brun et al. | Use of prior information in gradient-based history matching | |
Klose et al. | Toward subsurface magnetic permeability imaging with electromagnetic induction sensors: Sensitivity computation and reconstruction of measured data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |