CN106772646A - A kind of ground nuclear magnetic resonance method for extracting signal - Google Patents
A kind of ground nuclear magnetic resonance method for extracting signal Download PDFInfo
- Publication number
- CN106772646A CN106772646A CN201611178162.XA CN201611178162A CN106772646A CN 106772646 A CN106772646 A CN 106772646A CN 201611178162 A CN201611178162 A CN 201611178162A CN 106772646 A CN106772646 A CN 106772646A
- Authority
- CN
- China
- Prior art keywords
- noise
- frequency
- magnetic resonance
- mrs
- result
- 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.)
- Granted
Links
- 238000001225 nuclear magnetic resonance method Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 claims abstract description 54
- 238000005259 measurement Methods 0.000 claims abstract description 25
- 238000000605 extraction Methods 0.000 claims abstract description 20
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 16
- 238000005481 NMR spectroscopy Methods 0.000 claims abstract description 12
- 238000001514 detection method Methods 0.000 claims abstract description 12
- 238000007619 statistical method Methods 0.000 claims abstract description 5
- 238000005311 autocorrelation function Methods 0.000 claims description 14
- 230000010354 integration Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000012512 characterization method Methods 0.000 abstract description 2
- 238000001228 spectrum Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000000354 decomposition reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 239000003673 groundwater Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000009514 concussion Effects 0.000 description 2
- 238000005314 correlation function Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000005311 nuclear magnetism Effects 0.000 description 2
- GNFTZDOKVXKIBK-UHFFFAOYSA-N 3-(2-methoxyethoxy)benzohydrazide Chemical compound COCCOC1=CC=CC(C(=O)NN)=C1 GNFTZDOKVXKIBK-UHFFFAOYSA-N 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 125000004435 hydrogen atom Chemical group [H]* 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/38—Processing data, e.g. for analysis, for interpretation, for correction
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The present invention is specifically using based on industrial frequency harmonic modeling and autocorrelative ground nuclear magnetic resonance method for extracting signal.The method is comprised the following steps：Step (1):One group of noisy data of MRS is collected using ground nuclear magnetic resonance water detection appratus device；Step (2):Spike noise is judged whether using statistical method, if it does, remove spike noise and replaced with interpolation result, if it does not exist, then measurement data keeps constant；Step (3):The data of spike noise will be removed, industrial frequency harmonic noise is removed using the method for Harmonic Modeling；Step (4):Auto-correlation is carried out to the result of step (3) and overlap-add procedure reduces random noise；Step (5):MRS signal parameter extractions are carried out to the result after step (4).The inventive method solves the problem effectively extracted due to the MRS signals that the interference of strong industrial frequency harmonic and random noise are caused during magnetic resonance depth measurement water detection works, and the error of fitting of the MRS signal key characterization parameters of acquisition is smaller.
Description
Technical field
The present invention relates to ground nuclear magnetic resonance underground water detectable signal noise filtering and parameter extraction technical field, specifically
Using based on industrial frequency harmonic modeling and autocorrelative ground nuclear magnetic resonance method for extracting signal.
Background technology
Ground nuclear magnetic resonance (Magnetic Resonance Sounding, MRS) as a kind of geophysics new technology,
It is widely applied in recent years in underground water detection.Compared with the indirect groundwater prospecting method of traditional geophysics, ground nuclear magnetic resonance
Method has a high-resolution, high efficiency, abundant information and the characteristics of have only explanation to hydrologic parameter.Ground nuclear magnetic resonance side
The general principle of method is the MRS signals produced by Hydrogen Proton resonant transition in Underground water, extracts its characteristic parameter, then lead to
Inversion interpretation is crossed to obtain the water quality information of underground water.MRS methods are can be seen that from some researchs both at home and abroad and application not only
Good application effect is achieved at the aspect such as searching ground water and Evaluation of Groundwater Resources, and is obtaining hydrogeological parameter
Also made progress with quantification explanation aspect.
Although MRS methods have a clear superiority than other conventional methods, some shortcomings are there is also.It is actually detected
MRS signals be it is very faint receive volt level signal, noise jamming is very big in signal, including spike noise, random noise and work
Frequency harmonic noise etc., wherein influenceing maximum to it by the industrial frequency harmonic noise that power line interference causes.At present, for filtering nuclear-magnetism
There is much the theoretical and method of noise in resonance signal.Slander a man of virtue and ability brave in works《Magnetic resonance detection signal noise based on EMD suppresses
Method》, propose a kind of new signal time frequency processing method, can in the case of any priori of no input signal,
Adaptively by signal decomposition into several intrinsic mode functions (Intrinsic Mode Functions, IMF), signal is realized
Effective extraction of trend, but there are problems that end effect and modal overlap；Tian Baofeng et al. is in works《Based on reference coil and change
The magnetic resonance signal noise drawing method of adaptive step》In directly update processing parameter using observing the continuous recurrence of data, it is real
The removal of MRS signal noises under existing different signal to noise ratios and signal intensity, but the method needs distal coil acquisition noise number simultaneously
According to making reference, cannot be realized for single channel instrument, and correlated noise can only be processed, it is impossible to ensured multiple industrial frequency harmonics
Noise is all eliminated.
CN104459809A discloses " the all-wave NMR signal noise filtering method based on independent component analysis ", should
Algorithm can effectively realize the SNR estimation and compensation of all-wave MRS signals, and after data fitting initial amplitude and relaxation time it is relative
Error is less than ± 5.00%, compared to other classic algorithms algorithms with more de-noising performance superiority, but the method pin
To the industrial frequency harmonic noise in removal magnetic resonance signal.CN104898172A discloses a kind of " nuclear magnetic resonance based on cross-correlation
All-view signal noise filtering method ", the invention is uncorrelated to the sinusoidal signal of Larmor frequency using noise, and MRS amplitudes decline
The characteristics of subtracting sinusoidal signal there is correlation with the sinusoidal signal of Larmor frequency, noise is filtered by computing cross-correlation, then
The envelope of cross-correlation waveform is fitted, and reconstructs the cross-correlation waveform without noise, finally extracting nuclear-magnetism using uncoiling integration method is total to
MRS signals in all-wave of shaking data.The method operational data amount of calculation is small, can suppress simultaneously power frequency and its harmonic noise, with
Machine noise and spike noise, hence it is evident that improve nuclear magnetic resonance all-wave data SNR.But when earth's magnetic field is unstable or signal to noise ratio
When relatively low, Larmor frequency can not be obtained accurately, after carrying out cross-correlation calculation using the signal that there is frequency departure, signal parameter
Extract error big.A kind of " combined signal denoising based on empirical mode decomposition Yu wavelet analysis is disclosed in CN104636609A
Method ", according to the autocorrelation of signal, the signal to being mixed with white Gaussian noise carries out EMD decomposition, due to the property that EMD is decomposed,
White Gaussian noise has no longer been real white noise, but the statistical property of white noise is approximately present, i.e., described to be mixed with Gauss white noise
The auto-correlation function of the signal of sound obtains maximum, amplitude change poor over time and changes in zero point, but it is over time
Decay quickly.Can select the IMF components that noise plays a leading role using this species diversity effectively reduces noise to signal
Influence.But the method be mainly EMD is decomposed according to autocorrelation performance after the attribute of each mode judge, be mainly used in low
White Gaussian noise is removed under signal to noise ratio, so its application has limitation.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of ground nuclear magnetic resonance method for extracting signal, solve magnetic
Because strong industrial frequency harmonic disturbs the effective problem extracted of MRS signals caused with random noise in resonance depth measurement water detection work.
The present invention is achieved in that a kind of ground nuclear magnetic resonance method for extracting signal, and the method is comprised the following steps：
Step (1):One group of noisy data of observation MRS is collected using ground nuclear magnetic resonance water detection appratus device；
Step (2):Spike noise is judged whether using statistical method, if it does, removal spike noise and with slotting
Value result replaces, if it does not exist, then measurement data keeps constant；
Step (3):The data of spike noise will be removed, industrial frequency harmonic noise is removed using the method for Harmonic Modeling；
Step (4):Auto-correlation is carried out to the result of step (3) and overlap-add procedure reduces random noise；
Step (5):MRS signal parameter extractions are carried out to the result after step (4).
Further, modeling method in step (3) is concretely comprised the following steps：
Step 3a：The hunting zone and step-length for setting industrial frequency harmonic fundamental frequency carry out coarse scan；
Step 3b：It is the form of Ax=b by industrial frequency harmonic model conversation, system of linear equations of the solution comprising measurement data is obtained
To the coefficient matrix of each harmonic wave frequency of power frequency；
Step 3c：Calculate the model estimate value of different scanning value in the range of industrial frequency harmonic fundamental frequency search；
Step 3d：2 norms of computation model estimate and measurement data difference；
Step 3e：It is determined that the norm of error 2 is obtained the frequency of minimum value, so as to obtain the adjacent frequency in both sides；
Step 3f：The both sides adjacent frequency that will be obtained in step 3e chooses appropriate step-length again as sweep limits next time
Secondary scanning；
Step 3g：Judge whether the number of times of scanning setting, the frequency according to the Norm minimum of error 2 if reaching carries out work
Frequency Harmonic Modeling；Otherwise repeat step 3c~3g；
Step 3h:Industrial frequency harmonic model is subtracted from the result of step (2), industrial frequency harmonic noise is removed.
Further, step 3f：Step-length is (fm2-fm1)/M, wherein fm1, fm2The both sides of respectively minimum value frequency are adjacent
Frequency, wherein M=3~7.
Further, the scanning times that step 3g scannings set are as 3-7 times.
Further, carry out to the result that step (3) is obtained auto-correlation in described step (4) and overlap-add procedure reduce with
Machine noise is concretely comprised the following steps：
Step 4a：MRS signals are substituted into auto-correlation function formula, range of integration is changed, the auto-correlation of MRS signals is tried to achieve
Expression formula；
Step 4b：Whether acquisition time is judged more than 1s, and the front half section if greater than 1s interception autocorrelation results is usedApproximately, wherein e0For initial amplitude,For relaxation time, f are that Larmor frequency and τ are from phase
Close time interval；Result is closed if less than 1s from phase 0 to useApproximately, wherein tmaxFor
Time maximum；
Step 4c：The autocorrelation result of multiple measurement data is overlapped after judgement based on step 4b, is further subtracted
Small random noise.
Further, concretely comprising the following steps for parameter extraction is carried out to the result that step (4) is obtained in described step (5)：
Step 5a:Data to superposition in step (4) carry out Hilbert transform, then by LPF, are converted into two
Individual quadrature component；
Step 5b:The characteristic parameter A of MRS signals is solved using non-linear fitting method0And the relaxation timeWhereinInitial amplitude e is calculated again0As a result, parameter extraction result is initial amplitude e0And the relaxation time
Compared with prior art, beneficial effect is the present invention：The present invention is proposed and modeled based on industrial frequency harmonic and from phase
The ground nuclear magnetic resonance method for extracting signal of pass, for the all-wave MRS data of single channel collection, not only can disposably remove institute
There is harmonic wave to disturb, and have also combined autocorrelation method Attenuating Random Noise, effective MRS is realized finally by nonlinear fitting
The extraction of signal characteristic parameter.The inventive method solve magnetic resonance depth measurement water detection work in due to strong industrial frequency harmonic disturb and with
The problem that the MRS signals that machine noise is caused effectively are extracted, the error of fitting of the MRS signal key characterization parameters of acquisition is smaller, together
When the present invention breach classical noise-eliminating method and need the limitation of the other conditions such as multi-channel detection, save substantial amounts of manpower and materials.
Brief description of the drawings
Fig. 1 is ground nuclear magnetic resonance method for extracting signal FB(flow block) of the present invention；
Fig. 2 is that industrial frequency harmonic models denoising principle and design sketch, and Fig. 2 (a) is its time-domain signal, and MRS signals are submerged in makes an uproar
Among sound；Fig. 2 (b) is its power spectrum, there is the harmonic noise of different amplitudes on the frequency of harmonic wave；Fig. 2 (c) Harmonic Modelings
Afterwards, V is subtracted in measurement data bharmonicCancellation process is completed, after the de-noising of 2 (d) Harmonic Modeling, the profile of MRS signals is presented and declined
Subtract trend, the frequency of MRS signals is only existed in power spectrum；
Fig. 3 is auto-correlation de-noising effect figure；Black line is preferable MRS signals in Fig. 3 (a), and parameter is e0=100nV,F=2326Hz andGrey lines are the measurement letters after the white Gaussian noise for adding 100nV mean square deviations
Number, now the signal to noise ratio of data is 0dB, and MRS signals are submerged in noise substantially；Fig. 3 (b) is measurement signal and MRS signals
Frequency spectrum, it is seen that the spectrum peak of MRS signals is 10 times of noise spectrum average.3 (c) by after auto-correlation function, measurement signal
The result of (grey lines) and MRS signals (black line)；Spectrum peak and noise spectrum after 3 (d) MRS signal autocorrelations；
Fig. 4 is nonlinear fitting design sketch after auto-correlation, and Fig. 4 (a) is that, without the curve of auto-correlation function, Fig. 4 (b) is
By the curve of auto-correlation function；
Fig. 5 is emulation data and data processed result figure, the noisy MRS signals of Fig. 5 (a)；Fig. 5 (b) is from noisy MRS signals
Industrial frequency harmonic model is subtracted, power frequency component figure is obtained；Signal graph after the de-noising that 5 (c) auto-correlation and overlap-add procedure are obtained, 5 (d)
The real part of MRS signals, the imaginary part of 5 (e) MRS signals, the superimposed curves and imaginary part of 5 (f) MRS signal real parts are shown in；5 (g) MRS signals
The superimposed curves of imaginary part；
Fig. 6 is measured data result figure；Fig. 6 (a-1) is the MRS signals of q=1.4As；6 (a-2) are q=4.2As
MRS signals；6 (a-3) are the MRS signals of q=8.6As；
Fig. 6 (b-1) is the signal-noise ratio signal of q=1.4As；6 (b-2) are the signal-noise ratio signal of q=4.2As；6(b-3)
It is the signal-noise ratio signal of q=8.6As.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to
Limit the present invention.
As shown in figure 1, it is a kind of based on industrial frequency harmonic modeling and autocorrelative ground nuclear magnetic resonance method for extracting signal, including
Following steps：
Step (1):One group of noisy data of observation MRS is collected using ground nuclear magnetic resonance water detection appratus device；
Step (2):Judged whether using statistical method (such as Vladimir Romanovskiy criterion and energy calculation scheduling algorithm)
Spike noise, if it does, remove spike noise and replaced with interpolation result, if it does not exist, then measurement data keeps constant；
Step (3):Industrial frequency harmonic noise is removed using the method for Harmonic Modeling；
Step (4):Auto-correlation is carried out to the result of step (3) and overlap-add procedure reduces random noise；
Step (5):MRS signal parameter extractions are carried out to the result after step (4)；
As shown in Fig. 2 the specific steps of the power frequency modeling method in step (3)：
Step 3a:Setting industrial frequency harmonic fundamental frequency f0Hunting zone 49.9Hz~50.1Hz, f0Carried out by 0.03Hz of step-length
Coarse scan；
Step 3b:Industrial frequency harmonic noise VharmonicIt is represented by the form of formula (1)
Wherein, AnWithIt is respectively n-th amplitude and phase of harmonic wave, and hasWithT is the time.Expression formula (1) is organized into linear matrix form Ax=b,
Wherein, b=[V1,V2,…,VP]TIn Vp(p=1,2 ... P) it is tpThe reception data at moment；X=[α1,…,αN,
β1,…,βN]TIt is harmonic constant, N is harmonic wave number, N=100 is taken according to the bandwidth (1~3kHz) for receiving signal.Solve linear
Equation group (2) obtains harmonic constant matrix x.
Step 3c:Difference f in frequency search range is calculated respectively0Model estimate value Vharmonic(f0)；
Step 3d:Calculate Vharmonic(f0) 2 norms with the difference (error) of measurement data b：||b-Vharmonic(f0)|
|2；
Step 3e:It is determined that corresponding frequency f when making the norm of error 2 obtain minimum valuemin, so as to obtain the adjacent frequency in both sides
Point (fm1,fm2)；
Step 3f:It is determined that sweep limits (f next timem1,fm2), f0Selection step-length is (fm2-fm1)/M, wherein M=3~7；
Step g:Judge whether that scanning 5 times (could be arranged to 3-7 times, the power frequency fundamental frequency of search is relative every time changes
0.1%) amount is less than, if reached, corresponding frequency f during according to 2 Norm minimum of errorminCarry out industrial frequency harmonic modeling；
Otherwise repeat step 3c~3g；
Step h:Industrial frequency harmonic model V is subtracted from the result of step (2)harmonic(fmin), so as to remove industrial frequency harmonic make an uproar
Sound；
One group of reception signal data as shown in Figure 2, Fig. 2 (a) is its time-domain signal, and MRS signals are submerged among noise；
Fig. 2 (b) is its power spectrum, there is the harmonic noise of different amplitudes on the frequency of harmonic wave.After Harmonic Modeling, in measurement data
V is subtracted in bharmonicCancellation process is completed, shown in such as Fig. 2 (c) and 2 (d).After Harmonic Modeling de-noising, the profile of MRS signals
Attenuation trend is presented, the frequency of MRS signals is only existed in power spectrum.
The auto-correlation of present invention offer and concretely comprising the following steps for overlap-add procedure：
Step 4a：MRS signals VMRSRepresent：
Wherein, e0Represent initial amplitude；Mean time of relaxation is represented, f represents that Larmor frequency (has with local earth's magnetic field
Close),Represent initial phase.
According to autocorrelation principle, by VMRSIn the expression formula of substitution auto-correlation function：
Wherein, τ is time interval.Because auto-correlation function is the real even function on independent variable τ, so τ is only taking it just
Semiaxis.Range of integration in expression formula (4) takes τ and t respectivelymax, tmaxIt is time of measuring length, by being derived by VMRS's
Auto-correlation expression formula is：
R (τ)=RC1 (τ)+RC2 (τ)+RC3 (τ)+RC4 (τ)
Wherein,
From expression formula (5)~(7), MRS signal auto-correlation functions include four components R C1~RC4, and MRS signals are certainly
First half after correlation only can be replaced with RC1；Latter half can be replaced with RC1+RC2；The result very little of RC3+RC4,
Can ignore.
Step 4b：Whether acquisition time is judged more than 1s, and the front half section if greater than 1s interception autocorrelation results is near with RC1
Seemingly；It is approximate with RC1+RC2 if less than 1s autocorrelation results.
Step 4c：The autocorrelation result R of multiple measurement data b is overlapped after judgement based on step 4b, i.e.,Wherein N is pendulous frequency.Signal to noise ratio is improved after n times superpositionTimes, further reduce random noise.
Fig. 3 gives noisy MRS signals through auto-correlation and the result of overlap-add procedure.Black line is preferable in Fig. 3 (a)
MRS signals, parameter is e0=100nV,F=2326Hz andGrey lines are that to add 100nV square
Measurement signal after poor white Gaussian noise, now the signal to noise ratio of data is 0dB, and MRS signals are submerged in noise substantially.Fig. 3
B () is the frequency spectrum of measurement signal and MRS signals, it is seen that the spectrum peak of MRS signals is 10 times of noise spectrum average.By certainly
After correlation function, the result of measurement signal (grey lines) and MRS signals (black line) is shown in shown in Fig. 3 (c).Now, grey lines with
Black line is similar to, and obvious attenuation trend is also presented, and only occurs in that a small amount of random noise, and signal to noise ratio is 23.75dB, and from phase
Compared before pass and greatly improved.Can also be drawn from the spectrum analysis in Fig. 3 (d), the spectrum peak after MRS signal autocorrelations is about
An order of magnitude is improve before being 100 times of noise spectrum average, with auto-correlation.Therefore, auto-correlation function can be pressed effectively
Random noise processed, improves signal to noise ratio.
As shown in figure 4, the parameter extracting method that the present invention is provided is concretely comprised the following steps：
Step 5a:To the data being superimposedHilbert transform is carried out, Hilbert transform expression formula is：
Real signal R (t) can be converted into complex signalIts envelope signal isAgain by LPF, two are obtained
Individual quadrature component VRC=Vreal+i·Vimag；
Step 5b:Two quadrature components obtained to step a using non-linear fitting method carry out parameter extraction such as formula
(8),
Obtain the characteristic parameter A of MRS signals0WithWhereinE is calculated again0As a result, parameter extraction result is
Initial amplitude e0And the relaxation time
By two components after Hilbert conversion as shown in Figure 4, wherein Fig. 4 (a) is without certainly to data in Fig. 3
The curve of correlation function, Fig. 4 (b) is by the curve of auto-correlation function.It is utilized respectively non-linear fitting method and solves what is obtained
Two matched curves of component are as shown in solid black lines in Fig. 4 and black dotted lines.Without the data SNR of auto-correlation function
Low, parameter extraction result is respectively e0=108.3nV, F=2326.1Hz；And by auto-correlation function
Data SNR is obviously improved, and parameter extraction result is e0=100.5nV,F=2326.0Hz, error difference
Reduce 7.8%, 10.0% and 0.01%.Therefore, auto-correlation function can improve the signal to noise ratio of data, so as to improve parameter
The degree of accuracy of extraction.
Embodiment 1
Based on industrial frequency harmonic modeling and autocorrelative ground nuclear magnetic resonance method for extracting signal, invention emulates one group of bag
Containing 16 measurement data sets of independent acquisition, reference picture 1 is comprised the following steps：
Step (1)：Constructed shown in noisy MRS signals such as Fig. 5 (a) using formula (3), its parameter is e0=100nV,F=2326Hz andThe harmonic noise fundamental frequency of collection is randomly generated in 49.9~50.1Hz every time, humorous
Ripple number is 100, harmonic amplitude random distribution in 200nV；The random noise of collection is the white Gaussian noise of 200nV every time；
Spike noise is not added in gathered data, it is assumed that all of spike noise is eliminated by existing method.
Step (2)：Industrial frequency harmonic model is obtained using the method for Harmonic ModelingWherein
f0=50.018125Hz, industrial frequency harmonic model is subtracted from noisy MRS signals, obtains power frequency component, shown in such as Fig. 5 (b),
It can be seen that after industrial frequency harmonic de-noising, the signal to noise ratio of data is substantially improved, 17.03dB is averagely improve, data amplitude is reduced to ±
In the range of 500nV；
Step (3):Result to step (2) carries out auto-correlation and overlap-add procedure, reduces random noise, believes after obtaining de-noising
Number, such as shown in Fig. 5 (c), it is seen that after auto-correlation processing, data are presented obvious attenuation trend, except the signal initial stage has pulse
(autocorrelation performance of random noise), the data amplitude at other moment is significantly improved than the data before auto-correlationAnd the influence of random noise is smaller, therefore signal to noise ratio has and is largely lifted, be 15.64~
16.43dB, signal to noise ratio averagely improves 16.10dB；
Step (4):Result to step (3) carries out MRS signal parameter extractions, using Hilbert transform (reference frequency
2325Hz) with LPF (by frequency 200Hz), the real part and imaginary part for obtaining MRS signals are shown in shown in Fig. 5 (d) and 5 (e).Build
After mould denoising Processing, due to also there is substantial amounts of random noise, the real part and imaginary part curve of data are distinguished not substantially, can only observed
Overall trend to curve slightly rises and falls (concussion of 1Hz).After auto-correlation processing, the real part and imaginary part of data can be clear
Distinguish, both show the concussion attenuation process of 1Hz, then the data of 16 collections and treatment are overlapped, obtain
The superimposed curves of MRS signals real part and imaginary part are shown in shown in Fig. 5 (f) and (g).Overlap-add procedure can further reduce random noise,
16 times superposition signal to noise ratio improves 4 times (12.04dB), carries out MRS signal parameters to two groups of data using non-linear fitting method and carries
Take, the parameter extraction result after modeling de-noising and auto-correlation processing is：e0=100.45nV,F=
2326.0Hz, relative error is respectively：0.45%, -2.14% and 0, control within ± 3%, meet application requirement.
Embodiment 2
(the ground earth's magnetic field is corresponding as 54720nT with the peaceful water reservoir tank side in Changchun Nongan County enamelware pot town for the present embodiment
Larmor frequencies are less than 2Hz in 2330Hz, excursion) entered using the ground nuclear magnetic resonance water detection appratus device of independent development
Field data collection of having gone is tested with data processing.In order to from shallow to being detected to water-bearing layer deeply, 0.2~8.5As's
In the range of it is ascending be provided with 20 groups of transmitting pulse squares by log series model, superposition 16 times is repeated per group pulse square, sample rate is
50kHz, acquisition time is 1s.The treatment of the 10th, 16 and 20 MRS signals as the inventive method of the collections of pulse square is right
As.As shown in figure 1, based on industrial frequency harmonic modeling and autocorrelative ground nuclear magnetic resonance method for extracting signal, comprising the following steps:
Step (1):One group of noisy data of observation MRS is collected using ground nuclear magnetic resonance water detection appratus device；
Step (2):Spike noise is judged whether using statistical method, if it does, removal spike noise and with slotting
Value result replaces, if it does not exist, then measurement data keeps constant；
Step (3)：Model is obtained using the method for Harmonic ModelingWherein f0
=50.018125Hz, subtracts industrial frequency harmonic model from the result of step (2), obtains power frequency component, and such as Fig. 6 (a-1) is q
The MRS signals of=1.4As；6 (a-2) are the MRS signals of q=4.2As；6 (a-3) are shown in the MRS signals of q=8.6As, it is seen that
There is substantial amounts of harmonic noise in measured data, noise level in the range of 2500~3000nV, and MRS signal amplitudes 80~
375nV, therefore signal to noise ratio is relatively low.By after industrial frequency harmonic denoising Processing, most of harmonic noises are significantly reduced, and harmonic fundamental frequency exists
Between 49.975~50.041Hz, the peak value of MRS signals (2330Hz) is obvious, and the wherein signal of q=1.4As is larger, and q
The signal of=4.2As and 8.6As is smaller, and signal to noise ratio improves nearly 20dB；
Step (4):Result to step (3) carries out auto-correlation and overlap-add procedure, reduces random noise, believes after obtaining de-noising
Number；
Step (5):MRS signal parameter extractions, such as Fig. 6 (b-1), 6 (b-2), 6 (b- are carried out to the result after step (4)
3) shown in), signal to noise ratio increment is respectively 12.05dB, 2.98dB and 7.71dB under 3 pulse squares, now by nonlinear fitting
Afterwards, parameter extraction result is：
During q=4.2As, e0Fitting result is 80.7nV, and inaccuracy is 15.9nV, andFitting result is 235.9ms,
Inaccuracy is 38.2ms；During q=1.4As, e0Fitting result is 396.8nV, and inaccuracy is 17.4nV, andFitting result
It is 248.6ms, inaccuracy is 10.8ms；During q=8.6As, e0Fitting result is 175.6nV, and inaccuracy is 13.4nV, andFitting result is 263.1ms, and inaccuracy is 33.1ms, illustrates the increase of SNR, it is possible to increase the standard of nonlinear fitting result
Exactness.
Presently preferred embodiments of the present invention is the foregoing is only, is not intended to limit the invention, it is all in essence of the invention
Any modification, equivalent and improvement made within god and principle etc., should be included within the scope of the present invention.
Claims (7)
1. a kind of ground nuclear magnetic resonance method for extracting signal, it is characterised in that the method is comprised the following steps：
Step (1):One group of noisy data of MRS is collected using ground nuclear magnetic resonance water detection appratus device；
Step (2):Spike noise is judged whether using statistical method, if it does, removing spike noise and using interpolation knot
Fruit replaces, if it does not exist, then measurement data keeps constant；
Step (3):The data of spike noise will be removed, industrial frequency harmonic noise is removed using the method for Harmonic Modeling；
Step (4):Auto-correlation is carried out to the result of step (3) and overlap-add procedure reduces random noise；
Step (5):MRS signal parameter extractions are carried out to the result after step (4).
2. according to the ground nuclear magnetic resonance method for extracting signal described in claim 1, it is characterised in that the modeling in step (3)
Method is concretely comprised the following steps：
Step 3a：The hunting zone and step-length for setting industrial frequency harmonic fundamental frequency carry out coarse scan；
Step 3b：It is the form of Ax=b by industrial frequency harmonic model conversation, system of linear equations of the solution comprising measurement data obtains work
Frequently the coefficient matrix of each harmonic wave frequency；
Step 3c：Calculate the model estimate value of different scanning value in the range of industrial frequency harmonic fundamental frequency search；
Step 3d：2 norms of computation model estimate and measurement data difference；
Step 3e：It is determined that the norm of error 2 is obtained the frequency of minimum value, so as to obtain the adjacent frequency in both sides；
Step 3f：The both sides adjacent frequency that will be obtained in step 3e is chosen appropriate step-length and is swept again as sweep limits next time
Retouch；
Step 3g：Judge whether the number of times of scanning setting, it is humorous that the frequency according to the Norm minimum of error 2 if reaching carries out power frequency
Ripple is modeled；Otherwise repeat step 3c~3g；
Step 3h:Industrial frequency harmonic model is subtracted from the result of step (2), industrial frequency harmonic noise is removed.
3. according to the ground nuclear magnetic resonance method for extracting signal described in claim 2, it is characterised in that step 3a settings power frequency is frequently
Hunting zone 49.9Hz~the 50.1Hz of point, coarse scan is carried out by 0.03Hz of step-length.
4. according to the ground nuclear magnetic resonance method for extracting signal described in claim 2, it is characterised in that step 3f：Step-length is
(fm2-fm1)/M, wherein fm1, fm2Respectively the both sides adjacent frequency of minimum value frequency, wherein M=3~7.
5. according to the ground nuclear magnetic resonance method for extracting signal described in claim 2, it is characterised in that step 3g scanning settings
Scanning times are 3-7 times.
6. according to the ground nuclear magnetic resonance method for extracting signal described in claim 1, it is characterised in that in described step (4)
Auto-correlation is carried out to the result that step (3) is obtained and overlap-add procedure reduces concretely comprising the following steps for random noise：
Step 4a：MRS signals are substituted into auto-correlation function formula, range of integration is changed, the auto-correlation expression of MRS signals is tried to achieve
Formula；
Step 4b：Whether acquisition time is judged more than 1s, and the front half section if greater than 1s interception autocorrelation results is usedApproximately, wherein e0For initial amplitude,For relaxation time, f are that Larmor frequency and τ are from phase
Close time interval；Used if less than 1s autocorrelation resultsApproximately, wherein tmaxFor when
Between maximum；
Step 4c：The autocorrelation result of multiple measurement data is overlapped after judgement based on step 4b, further reduce with
Machine noise.
7. according to the ground nuclear magnetic resonance method for extracting signal described in claim 1, it is characterised in that in described step (5)
Concretely comprising the following steps for parameter extraction is carried out to the result that step (4) is obtained：
Step 5a:Data to superposition in step (4) carry out Hilbert transform, then by LPF, are being converted into two just
Hand over component；
Step 5b:The characteristic parameter A of MRS signals is solved using non-linear fitting method0And the relaxation timeWhereinInitial amplitude e is calculated again0As a result, parameter extraction result is initial amplitude e0And the relaxation time
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611178162.XA CN106772646B (en) | 2016-12-19 | 2016-12-19 | A kind of ground nuclear magnetic resonance method for extracting signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611178162.XA CN106772646B (en) | 2016-12-19 | 2016-12-19 | A kind of ground nuclear magnetic resonance method for extracting signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106772646A true CN106772646A (en) | 2017-05-31 |
| CN106772646B CN106772646B (en) | 2018-11-06 |
Family
ID=58890460
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611178162.XA Expired - Fee Related CN106772646B (en) | 2016-12-19 | 2016-12-19 | A kind of ground nuclear magnetic resonance method for extracting signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106772646B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345039A (en) * | 2018-01-12 | 2018-07-31 | 吉林大学 | A method of eliminating adjacent frequency harmonic wave interference in ground nuclear magnetic resonance data |
| CN108363109A (en) * | 2018-01-12 | 2018-08-03 | 吉林大学 | The harmonic wave interference fundamental frequency method for fast searching of ground nuclear magnetic resonance data |
| CN109031287A (en) * | 2018-09-21 | 2018-12-18 | 西安交通大学 | ULTRA-WIDEBAND RADAR human body respiration signal detecting method through walls based on Faster-RCNN network |
| CN109100813A (en) * | 2018-08-14 | 2018-12-28 | 吉林大学 | A method of it is filtered based on collaboration and eliminates spike noise in ground nuclear magnetic resonance data |
| CN109765629A (en) * | 2019-01-30 | 2019-05-17 | 吉林大学 | A kind of ground magnetic resonance signal extracting method that can suppress same frequency noise jamming |
| CN109885906A (en) * | 2019-01-30 | 2019-06-14 | 吉林大学 | Magnetic resonance sounding signal sparse noise elimination method based on particle swarm optimization |
| CN111024398A (en) * | 2019-12-25 | 2020-04-17 | 西安交通大学 | Deconvolution method for maximum correlation kurtosis without period |
| CN111239837A (en) * | 2020-02-20 | 2020-06-05 | 吉林大学 | Ground nuclear magnetic resonance signal parameter extraction method based on MCMC |
| CN112834844A (en) * | 2020-12-31 | 2021-05-25 | 瑞斯康微电子（深圳）有限公司 | Method for eliminating peak singular signals in leakage current signals of lightning arrester |
| CN113343388A (en) * | 2021-06-22 | 2021-09-03 | 安徽容知日新科技股份有限公司 | Method for acquiring steady-state vibration data and computing equipment |
| CN117708717A (en) * | 2024-02-05 | 2024-03-15 | 吉林大学 | Magnetic resonance underground water detection signal high-precision extraction method based on random forest |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060255249A1 (en) * | 2004-06-23 | 2006-11-16 | Canwen Liu | Automatic background removal for input data |
| CN104614778A (en) * | 2015-01-27 | 2015-05-13 | 吉林大学 | Nuclear magnetic resonance underground water detection signal noise eliminating method based on independent component analysis (ICA) |
| CN104898172A (en) * | 2015-05-19 | 2015-09-09 | 吉林大学 | Cross-correlation-based nuclear magnetic resonance full wave signal noise filtering method |
-
2016
- 2016-12-19 CN CN201611178162.XA patent/CN106772646B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060255249A1 (en) * | 2004-06-23 | 2006-11-16 | Canwen Liu | Automatic background removal for input data |
| CN104614778A (en) * | 2015-01-27 | 2015-05-13 | 吉林大学 | Nuclear magnetic resonance underground water detection signal noise eliminating method based on independent component analysis (ICA) |
| CN104898172A (en) * | 2015-05-19 | 2015-09-09 | 吉林大学 | Cross-correlation-based nuclear magnetic resonance full wave signal noise filtering method |
Non-Patent Citations (2)
| Title |
|---|
| 田宝凤 等: "基于独立成分分析的全波核磁共振信号噪声滤除方法研究", 《物理学报》 * |
| 田宝凤 等: "核磁共振信号工频谐波的自适应滤除方法", 《吉林大学学报（信息科学版）》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345039A (en) * | 2018-01-12 | 2018-07-31 | 吉林大学 | A method of eliminating adjacent frequency harmonic wave interference in ground nuclear magnetic resonance data |
| CN108363109A (en) * | 2018-01-12 | 2018-08-03 | 吉林大学 | The harmonic wave interference fundamental frequency method for fast searching of ground nuclear magnetic resonance data |
| CN108363109B (en) * | 2018-01-12 | 2019-06-21 | 吉林大学 | The harmonic wave interference fundamental frequency method for fast searching of ground nuclear magnetic resonance data |
| CN108345039B (en) * | 2018-01-12 | 2019-07-23 | 吉林大学 | A method of eliminating adjacent frequency harmonic wave interference in ground nuclear magnetic resonance data |
| CN109100813A (en) * | 2018-08-14 | 2018-12-28 | 吉林大学 | A method of it is filtered based on collaboration and eliminates spike noise in ground nuclear magnetic resonance data |
| CN109031287A (en) * | 2018-09-21 | 2018-12-18 | 西安交通大学 | ULTRA-WIDEBAND RADAR human body respiration signal detecting method through walls based on Faster-RCNN network |
| CN109765629A (en) * | 2019-01-30 | 2019-05-17 | 吉林大学 | A kind of ground magnetic resonance signal extracting method that can suppress same frequency noise jamming |
| CN109885906A (en) * | 2019-01-30 | 2019-06-14 | 吉林大学 | Magnetic resonance sounding signal sparse noise elimination method based on particle swarm optimization |
| CN111024398A (en) * | 2019-12-25 | 2020-04-17 | 西安交通大学 | Deconvolution method for maximum correlation kurtosis without period |
| CN111024398B (en) * | 2019-12-25 | 2021-01-15 | 西安交通大学 | Deconvolution method for maximum correlation kurtosis without period |
| CN111239837A (en) * | 2020-02-20 | 2020-06-05 | 吉林大学 | Ground nuclear magnetic resonance signal parameter extraction method based on MCMC |
| CN111239837B (en) * | 2020-02-20 | 2021-01-05 | 吉林大学 | Ground magnetic resonance signal parameter extraction method based on MCMC |
| CN112834844A (en) * | 2020-12-31 | 2021-05-25 | 瑞斯康微电子（深圳）有限公司 | Method for eliminating peak singular signals in leakage current signals of lightning arrester |
| CN112834844B (en) * | 2020-12-31 | 2023-10-20 | 瑞斯康微电子（深圳）有限公司 | Method for eliminating peak singular signals in lightning arrester leakage current signals |
| CN113343388A (en) * | 2021-06-22 | 2021-09-03 | 安徽容知日新科技股份有限公司 | Method for acquiring steady-state vibration data and computing equipment |
| CN113343388B (en) * | 2021-06-22 | 2022-07-08 | 安徽容知日新科技股份有限公司 | Method for acquiring steady-state vibration data and computing equipment |
| CN117708717A (en) * | 2024-02-05 | 2024-03-15 | 吉林大学 | Magnetic resonance underground water detection signal high-precision extraction method based on random forest |
| CN117708717B (en) * | 2024-02-05 | 2024-04-12 | 吉林大学 | Magnetic resonance underground water detection signal high-precision extraction method based on random forest |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106772646B (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106772646B (en) | A kind of ground nuclear magnetic resonance method for extracting signal | |
| CN105116442B (en) | The reconstructing method of the weak seismic reflection signals of lithologic deposit | |
| CN106646637A (en) | Method for removing peak noise in nuclear magnetism signal | |
| CN109885903B (en) | Model-based ground nuclear magnetic resonance signal spike noise removing method | |
| CN107272062B (en) | A kind of Q estimation methods of underground medium of data-driven | |
| CN109828318A (en) | A kind of magnetic resonance depth measurement signal noise filtering method based on variation mode decomposition | |
| CN107045149A (en) | A kind of all-wave NMR signal noise filtering method based on double singular value decompositions | |
| CN110320555B (en) | Seismic data reconstruction method | |
| CN110221256A (en) | SAR disturbance restraining method based on depth residual error network | |
| CN104950326B (en) | The method for designing of controlled source nonlinear scanning signal based on target zone frequency spectrum | |
| CN104614769B (en) | A kind of Beamforming for suppressing seismic surface wave | |
| CN106680874A (en) | Harmonic noise suppression method based on waveform morphology sparse modeling | |
| CN106500671A (en) | A kind of based on LM algorithms decompose laser radar waveform determine sea water advanced method | |
| CN102081168A (en) | Method for improving surface wave attenuation performance in seismic data processing | |
| CN113887398A (en) | GPR signal denoising method based on variational modal decomposition and singular spectrum analysis | |
| CN104133248A (en) | High-fidelity sound wave interference suppression method | |
| CN108828658A (en) | A kind of ocean bottom seismic data reconstructing method | |
| CN104977602B (en) | A kind of control method and device of earthquake data acquisition construction | |
| CN108594301B (en) | A kind of method and processing terminal of the seismic data fusion with difference characteristic | |
| CN110031891A (en) | The Spatial Coupling exciting method of Surface Wave Elimination and shallow-layer multiple reflections refracted wave simultaneously | |
| CN105204068A (en) | Actual data spectrum characteristic-based nonlinear scanning signal design method | |
| TIAN et al. | Denoising and extraction method of magnetic resonance sounding signal based on adaptive harmonic modeling and autocorrelation | |
| LIN et al. | Segmented time-frequency peak filtering for random noise reduction of MRS oscillating signal | |
| CN106950597B (en) | Mixing source data separation method based on the filtering of three sides | |
| CN109857979A (en) | A kind of log approximating method based on wavelet analysis and BP neural network |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181106 |