CN107783191B - The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise - Google Patents
The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise Download PDFInfo
- Publication number
- CN107783191B CN107783191B CN201710990336.0A CN201710990336A CN107783191B CN 107783191 B CN107783191 B CN 107783191B CN 201710990336 A CN201710990336 A CN 201710990336A CN 107783191 B CN107783191 B CN 107783191B
- Authority
- CN
- China
- Prior art keywords
- time
- signal
- frequency
- multidimensional
- data
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 87
- 238000001914 filtration Methods 0.000 claims abstract description 82
- 238000012952 Resampling Methods 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000009826 distribution Methods 0.000 claims description 16
- 239000004744 fabric Substances 0.000 claims description 8
- 230000010354 integration Effects 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 3
- 238000012986 modification Methods 0.000 claims description 3
- 230000004048 modification Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 230000015572 biosynthetic process Effects 0.000 abstract description 7
- 238000003786 synthesis reaction Methods 0.000 abstract description 7
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000003672 processing method Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 11
- 238000012545 processing Methods 0.000 description 10
- 238000004458 analytical method Methods 0.000 description 9
- 238000011084 recovery Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 239000002131 composite material Substances 0.000 description 4
- 238000005520 cutting process Methods 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 239000003208 petroleum Substances 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000012512 characterization method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000003825 pressing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 244000025254 Cannabis sativa Species 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 210000003739 neck Anatomy 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000010415 tropism Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/36—Effecting static or dynamic corrections on records, e.g. correcting spread; Correlating seismic signals; Eliminating effects of unwanted energy
- G01V1/364—Seismic filtering
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/30—Noise handling
- G01V2210/32—Noise reduction
- G01V2210/324—Filtering
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The present invention relates to a kind of methods of hyperspace space-time time-frequency method abatement seismic prospecting random noise, belong to multidimensional random noise method for reducing.Multidimensional space-time time-frequency peak filtering principle is elaborated first, devise a kind of hyperspace space-time path filtering method, it realizes and cuts down in the resampling in multidimensional space-time path space and time-frequency method random noise, and combine synthesis seismic prospecting record optimization actual seismic exploration record processing method.By generated data and physical record, the result shows that, this method can effectively cut down seismic prospecting random noise in the case of low signal-to-noise ratio, and the useful signal being able to maintain in multi-dimensional earthquake survey data.The present invention explores random noise abatement for multi-dimensional earthquake and provides a kind of effective tool, can be used for low signal-to-noise ratio seismic prospecting signal extraction and fidelity with complicated random noise and restore, the invention is practical, it is easy to accomplish.
Description
Technical field
The present invention relates to a kind of multidimensional random noise method for reducing, cut down problem especially for seismic prospecting random noise,
Using seismic prospecting signal multidimensional characteristic, propose that multidimensional time-frequency method technology cuts down seismic prospecting random noise, high-fidelity
Restore weak seismic prospecting useful signal.
Background technique
As the increasingly deficient of resource is visited to the continuous growth of petroleum resources and Yi Caiyi by country, seismic prospecting is gradually by big
It covers oil field to expand to craven fault, complicated structure area, this proposes seismic exploration technique bigger challenge, to seismic exploration technique
Requirement also gradually to develop to high s/n ratio, high-resolution, Hi-Fi " three high " by enhancing useful signal horizontal, and gradually
Develop to higher-dimension seismic prospecting.In " three high " requirement, high s/n ratio is basis.Reach high s/n ratio requirement to need while enhancing
Useful signal and abating noises.Improve the necessary item that multi-dimensional earthquake exploration data signal-to-noise ratio is high-precision exploring subsurface petroleum resources
Challenging problem in part and seismic prospecting.It just needs to break through this difficult point it is proposed that stability is high, practical
Noise reduction methods.
When seismic prospecting field data collection, geophone receives all vibrations at observation point, wherein just including needing
The noise information to be cut down.The generation of noise generally with focus, mode of excitation, propagation medium, receive environment, surface materials structure
And underground medium is related.The video frequency that do not fix with the simultaneous random noise, that is, random noise of organized noise, does not have
There is the specific direction of propagation, amplitude variations are also uneven.A kind of random noise is the disturbance unrelated with excitaton source, it is mainly
Move, wave, water flow, walked about by wind, grass, machine starts, the external force such as communications and transportation are randomly generated, its main feature is that frequency band variation
It is larger, intensity is different, depend on exciting received ambient environmental conditions, be that no earthquake stimulation is present in earthquake record
's.To seismic prospecting random noise have one accurately understanding be design random noise abatement algorithm starting point, only with
Based on the characteristic of machine noise, suitable noise reduction methods could be effectively designed, improve or selected.Different necks are come from present
A variety of methods in domain have been suggested and have been applied in the useful signal extraction and noise compacting of seismic data.According to these methods
Noise pressing principle, noise drawing method can be divided into following four classes: classical denoising method, based on correlation denoising side
Method is based on multiscale analysis denoising method, is based on time frequency analysis denoising method.These methods have been not limited to simply and directly
Using, corresponding innovatory algorithm or the algorithm combined with other methods are also put forward one after another, thus it is perfect its on ground
Deficiency present in the compacting of shake data noise reaches certain expected denoising effect.It is representational at present in these methods to be
Multiscale analysis denoising method and be based on time frequency analysis denoising method.Multiscale analysis denoising method is by carrying out more rulers to data
Degree decomposes, according to the feature differences such as the Energy distribution of different scale layer useful signal and noise and frequency distribution, respectively to difference
Scale carries out processing and achievees the purpose that noise is suppressed.Wavelet transformation (Wavelet Transform, WT) is a kind of typical more rulers
Analysis method is spent, it has preferable temporal frequency localization analysis ability, therefore it compares Fourier transform in the expression of signal
More effectively.Since last century the nineties, in terms of de-noising research, had been achieved in conjunction with other technologies in small echo both at home and abroad
Many research achievements (Johnston andSilvernan, 1997；Chang et al., 2000), wavelet field noise-removed technology at present
It is preferably applied in field of seismic exploration.But this method is largely dependent upon in terms of the de-noising of seismic data
The correctness of the selection of threshold value, threshold value selection directly influences the quality of noise pressing result, and the transformation does not have side
Tropism has certain limitation in the complicated multidimensional signal of processing.Denoising method based on time frequency analysis can effectively by
Time-domain and frequency domain combine, and the time varying characteristic of signal is obtained from different moments frequency spectrum, by the time-frequency distributions of signal come
Signal is estimated, to realize the compacting of seismic signal noise and signal enhancing.When time-frequency method (TFPF) method is typical
Frequency analysis method, based on seismic prospecting noise space-time characterisation mathematics characterization, development adapts to land and explores noise complexity
Time-frequency distributions, improve that signal is non-linear, improves the key technologies such as signal-to-noise ratio and time frequency resolution simultaneously.Time-frequency method is
A kind of Time-frequency Filter algorithm that can restore signal under Low SNR.Peak of the algorithm based on analytic signal time-frequency distributions
Value frequency is capable of the principle of unbiased esti-mator instantaneous frequency, by realizing to signals and associated noises modulating-coding and the filtering of time-frequency domain peak value
The recovery of useful signal under strong random noise.Time-frequency method requires less signal priori in filtering, and in low letter
It makes an uproar not by random noise disturbance than under, therefore achieves certain effect in terms of seismic prospecting random noise abatement.Time-frequency peak value
Filtering can be to the undistorted recovery of signal, however, time-frequency peak filtering is to variation when it is linear that signal, which changes over time,
The amplitude of faster high-frequency seism Exploration signals keeps effect not satisfactory.
Summary of the invention
The present invention provides a kind of method of hyperspace space-time time-frequency method abatement seismic prospecting random noise, purpose
It is clearly to restore seismic prospecting useful signal, enhances the continuity of lineups, it is random simultaneously effective cuts down seismic prospecting
Noise improves the signal-to-noise ratio of seismic prospecting data, is further processed and lays the foundation for multi-dimensional earthquake Exploration signals.
The technical solution adopted by the present invention is that: including the following steps:
(1) multidimensional spatio-temporal filtering path is constructed, conventional multi-dimensional earthquake exploration observation data are that the underground of wave detector record is anti-
The superposition of the random noises such as signal and ambient noise is penetrated, is indicated are as follows:
D (x)=S (x)+V (x)
Wherein multi-dimensional earthquake observation data are D (x), and x indicates that hyperspace coordinate, S (x) are effective seismic prospecting signal,
V (x) is random noise；
Multidimensional spatio-temporal filtering path is the filter paths of multidimensional time-frequency method, multidimensional spatio-temporal filtering path is defined as:
Tr (x)=φ (x)
Wherein φ () is M dimension filtering path, and as M=1, φ () is the straight line along time orientation；As M=2, φ
() is time-space direction two dimension path, can choose the radial alignment for having angle with time orientation, also can choose
Hyperbolic or parabolic curve；When M is more than or equal to 3, then the curve that changes for M dimension space with signal characteristic；
Multidimensional spatio-temporal filtering path is constructed using earthquake record multidimensional characteristic, using similarity function as cost function,
It predicts that the lineups of current road and neighboring track move towards, estimates multidimensional spatio-temporal filtering path function, for multi-dimensional earthquake data, i-th
Road seismic data is di, per pass data include m sampled point, lineups and neighboring track dkLocal delay is τi,k；It should to determine
Delay time, the setting parsing window centered on i-th any point, parsing window inclination angle is θ, calculates seismic data in parsing window
Similarity factor is denoted as PθD changes inclination angle and repeats the above steps, using maximum similarity factor direction as the i-th road Dao Yuk
Local delay:
Wherein PθFor the feature extraction operator for the parsing window that inclination angle is θ, similarity factor operator is used here, utilizes recurrence side
Method, according to the local delay information under the local delay information acquisition being calculated together, the then local delay of pending data
It is represented by τ=[τ1,2,τ2,3,…,τN-1,N], it includes seismic channel number that wherein N, which is seismic data, thus calculates multidimensional spatio-temporal filtering
Path parameter obtains the estimation of multidimensional spatio-temporal filtering pathIt indicates are as follows:
Wherein | | | | indicate Euclid norm, a indicates filtering path parameter；
(2) multidimensional data resampling
Data are sampled along multidimensional spatio-temporal filtering path, the sampling road Shi Yi is horizontal axis, using sampled point as the longitudinal axis, by ground
The point that shake road intersects with filtering path indulges path edge as sampled point, the corresponding sample sequence of such each path
Axis successively translates and extracts sample sequence, the resampling until completing entire seismic prospecting record, obtained resampling data Dr
(x), it indicates are as follows:
(3) resampling data modulating-coding
In multidimensional space-time path domain, counterweight sampled data Dr(x) frequency modulation(PFM) coding is carried out, analytic signal is converted into
Z (x), it may be assumed that
Exp is exponential function in formula, and μ is scale factor, and λ is integration variable, and j is imaginary unit；
(4) time-frequency distributions of analytic signal are calculated
The time-frequency distributions of analytic signal are calculated, select pseudo- Eugene Wigner prestige pulling cloth or its high-resolution modification version as needed
This, obtains the pseudo- Eugene Wigner prestige pulling cloth W of analytic signalz(x, f) is indicated are as follows:
Wherein f indicates frequency, and * indicates complex conjugate operation, and υ is integration variable, and the length of window function h (υ) is denoted as the long WL of window,
The determination of window function length is as follows:
Wherein fhFor the highest frequency of signal, σ2For random noise variance, C is characteristic parameter, according to the system of signal and noise
Characteristic is counted to describe；
(5) filtering signal is obtained, peak value filtering is carried out to the time-frequency distributions of analytic signal, calculates the when frequency division of analytic signal
Estimation of the cloth crest frequency as useful signal, it may be assumed that
In formulaIt for filtered signal, is filtered by time-frequency domain peak value, filtering signal obtained is actual signal
Unbiased esti-mator.
By above-mentioned steps, multidimensional time-frequency peak filtering, which realizes, disappears to the random noise in seismic prospecting data
Subtract, recovers useful signal from noisy exploration record.
In multi-dimensional earthquake exploration, seismic exploration data has multidimensional characteristic, makes full use of multidimensional special in filtering
Sign is played an important role for improving filtering algorithm noise removal capability.Therefore, in seismic prospecting signal processing, in addition to considering ground
The time domain waveform feature and noise profile of seismic exploration signal, are more conducive in conjunction with seismic prospecting data space characteristic and multidimensional characteristic
The high-fidelity for improving random noise abatement ability and signal restores precision.Based on this, the present invention proposes hyperspace space-time time-frequency
Peak value filtering algorithm has further expanded the ability of time-frequency method processing seismic prospecting signal.The invention proposes multidimensional
The algorithm of time-frequency method constructs multidimensional spatio-temporal filtering trajectory model and parameter setting criterion.
The multidimensional characteristic of present invention combination multi-dimensional earthquake exploration record, from the linear conditions of time-frequency method unbiased esti-mator
Incision proposes multidimensional spatio-temporal filtering path concept and determines principle, constructs multidimensional space-time time-frequency peak filtering, to greatest extent
The linearity for improving seismic prospecting analytic signal instantaneous frequency.The present invention is to solve multi-dimensional earthquake under strong random noise disturbance
The recovery of Exploration signals high-fidelity provides effective filters solutions, the strong random noise in effectively compacting multi-dimensional earthquake survey data
Meanwhile improving the continuity and clarity of lineups.
The invention has the advantages that with the raising required high s/n ratio, high-resolution, high-fidelity, multi-dimensional earthquake exploration
It has become and detects one of the important means of petroleum resources.The present invention is proposed according to the multidimensional information of multi-dimensional earthquake Exploration signals
Random noise method for reducing based on hyperspace time-frequency peak filtering, has not been reported at home and abroad.Advantages of the present invention
It is to construct multidimensional spatio-temporal filtering path first, signal characteristic and noise statistics property is not only utilized, also sufficiently combines
Multi-dimensional earthquake data association message, to can with high fidelity restore useful signal while cutting down seismic prospecting random noise.
In addition, the present invention is by multidimensional data resampling, to multidimensional signal dimension-reduction treatment, the validity of filtering algorithm and practical is enhanced
Property.In random noise abatement principle, the present invention is by counterweight sampled data modulating-coding, by random noise abatement problem conversion
For the instantaneous Frequency Estimation problem of time-frequency domain analytic signal, it can restore effective by filtering to analytic signal time-frequency distributions peak value
Signal simply and effectively solves the problems, such as that random noise abatement and useful signal restore under low signal-to-noise ratio.The present invention is effectively cutting down
Strong random noise simultaneously, improves the continuity and clarity of lineups in multi-dimensional earthquake survey data.The present invention is making an uproar by force at random
Sound abatement aspect effect is prominent, and shows the ability for preferably keeping signal amplitude.The present invention is multi-dimensional earthquake exploration letter
Number processing provides a kind of effective and practical random noise method for reducing.
The present invention can remove common seismic exploration denoising method and be difficult to the strong random noise removed, improve actual seismic and survey
The signal-to-noise ratio of record is visited, it is significant to multi-dimensional earthquake Exploration signals high-fidelity recovery effects.
Detailed description of the invention
Fig. 1 is 3-d seismic exploration time-frequency peak filtering Filtering Model figure；
Fig. 2 is the filtering path schematic diagram of multi-dimensional earthquake exploration time-frequency method；
Fig. 3 is noisy synthesis seismic prospecting record filtering path diagram, (a) noisy synthesis seismic prospecting record, (b) when multidimensional
The filtering path of empty time-frequency method；
Fig. 4 (a)~Fig. 4 (h) is multidimensional space-time time-frequency method denoising result figure, the original composite traces of Fig. 4 (a), Fig. 4
(b) noisy composite traces, Fig. 4 (c) Wavelet Denoising Method as a result, Fig. 4 (d) Wavelet Denoising Method difference, Fig. 4 (e) time-frequency method denoising
As a result, the difference of Fig. 4 (f) time-frequency method denoising, Fig. 4 (g) denoising result of the present invention, the difference that Fig. 4 (h) present invention denoises；
Fig. 5 is that actual seismic exploration record filter result compares figure, (a) actual seismic exploration record, (b) Wavelet Denoising Method knot
Fruit, (c) multidimensional time-frequency method denoising result.
Specific embodiment
Including the following steps:
(1) multidimensional spatio-temporal filtering path is constructed, conventional multi-dimensional earthquake exploration observation data are that the underground of wave detector record is anti-
The superposition of the random noises such as signal and ambient noise is penetrated, is indicated are as follows:
D (x)=S (x)+V (x)
Wherein multi-dimensional earthquake observation data are D (x), and x indicates that hyperspace coordinate, S (x) are effective seismic prospecting signal,
V (x) is random noise, and the present invention is based on making an uproar in multidimensional space-time time-frequency peak filtering abatement seismic prospecting record at random
Sound restores useful signal S (x)；
Multidimensional spatio-temporal filtering path is the filter paths of multidimensional time-frequency method, multidimensional spatio-temporal filtering path is defined as:
Tr (x)=φ (x)
Wherein φ () is M dimension filtering path, and as M=1, φ () is the straight line along time orientation；As M=2, φ
() is time-space direction two dimension path, can choose the radial alignment for having angle with time orientation, also can choose
Hyperbolic or parabolic curve；When M is more than or equal to 3, then the curve that changes for M dimension space with signal characteristic；
Multidimensional time-frequency method method constructs multidimensional spatio-temporal filtering path first with earthquake record multidimensional characteristic, uses
Similarity function predicts that the lineups of current road and neighboring track move towards, estimates multidimensional spatio-temporal filtering path letter as cost function
Number, for multi-dimensional earthquake data, the i-th seismic data is di, per pass data include m sampled point, lineups and neighboring track dk
Local delay is τi,k；To determine the delay time, the setting parsing window centered on i-th any point, parsing window inclination angle is θ,
The similarity factor for calculating seismic data in parsing window, is denoted as PθD changes inclination angle and repeats the above steps, by maximum similarity factor institute
In local delay of the direction as the i-th road Dao Yuk:
Wherein PθFor the feature extraction operator for the parsing window that inclination angle is θ, similarity factor operator is used here, utilizes recurrence side
Method, according to the local delay information under the local delay information acquisition being calculated together, the then local delay of pending data
It is represented by τ=[τ1,2,τ2,3,…,τN-1,N], it includes seismic channel number, air filter when thus can calculate multidimensional that wherein N, which is seismic data,
Wave path parameter obtains the estimation of multidimensional spatio-temporal filtering pathIt indicates are as follows:
Wherein | | | | indicate Euclid norm, a indicates filtering path parameter；
(2) multidimensional data resampling
It is sampled along multidimensional spatio-temporal filtering path to data, the sampling road Shi Yi is horizontal axis, will using sampled point as the longitudinal axis
Seismic channel and the point that intersect of filtering path are as sampled point, one sample sequence of such each path correspondence, and by path edge
The longitudinal axis successively translates and extracts sample sequence, the resampling until completing entire seismic prospecting record, obtained resampling data
Dr(x), it indicates are as follows:
(3) resampling data modulating-coding
In multidimensional space-time path domain, counterweight sampled data Dr(x) frequency modulation(PFM) coding is carried out, analytic signal is converted into
Z (x), it may be assumed that
Exp is exponential function in formula, and μ is scale factor, and λ is integration variable, and j is imaginary unit；
(4) time-frequency distributions of analytic signal are calculated
The time-frequency distributions of analytic signal are calculated, select pseudo- Eugene Wigner prestige pulling cloth or its high-resolution modification version as needed
This, obtains the pseudo- Eugene Wigner prestige pulling cloth W of analytic signalz(x, f) is indicated are as follows:
Wherein f indicates frequency, and * indicates complex conjugate operation, and υ is integration variable, and the length of window function h (υ) is denoted as the long WL of window,
The determination of window function length is as follows:
Wherein fhFor the highest frequency of signal, σ2For random noise variance, C is characteristic parameter, according to the system of signal and noise
Characteristic is counted to describe；
(5) filtering signal is obtained, peak value filtering is carried out to the time-frequency distributions of analytic signal, calculates the when frequency division of analytic signal
Estimation of the cloth crest frequency as useful signal, it may be assumed that
In formulaFor filtered signal.It is filtered by time-frequency domain peak value, filtering signal obtained is actual signal
Unbiased esti-mator.
By above-mentioned steps, multidimensional time-frequency peak filtering, which realizes, disappears to the random noise in seismic prospecting data
Subtract, recovers useful signal from noisy exploration record.
Applicating example:
To prove that random noise of the present invention suppresses the effect of algorithm, inventive algorithm is applied to synthesis seismic prospecting record
Processing.The synthesis seismic prospecting constructed is recorded as shown in part (a) of attached drawing 3, containing there are two the same of differently curved degree
Phase axis, road spacing are 30m, sample frequency 500HZ.Shallow earthquake wavelet dominant frequency is 45Hz, speed 1800m/s, starting
Arrival time is 0.1s；Deep seismic wavelet dominant frequency is 28Hz, and speed 1900m/s, starting arrival time is 0.8s.It is added
White Gaussian random noise makes the signal-to-noise ratio of noisy seismic prospecting record be -9dB, to verify random noise suppression of the present invention
Noise removal capability and signal recovery effects.Generated data is handled by the multi-C filtering path extraction algorithm first step, is obtained
The filtering path obtained is as shown in part (b) of attached drawing 3.
Interpretation of result: from part (b) of attached drawing 3 it can be seen that two differently curved degree lineups with extracted
Two filtering path tendencies matching, also just demonstrate inventive algorithm in this way in low signal-to-noise ratio, can for difference
The lineups of bending degree establish the validity of matching optimization filtering trajectory model.
Cut down effect further to verify random noise of the present invention, is handled using multidimensional space-time time-frequency peak filtering attached
Composite traces shown in Fig. 4.The synthesis seismic prospecting record constructed is made of Ricker wavelet, such as attached drawing 4 (a), wherein lineups
Dominant frequency is 45Hz and twice 25Hz, and intersecting axle is the lineups that dominant frequency is 45Hz and dominant frequency is 25Hz lineups.Road spacing is
30m, sample frequency 1000HZ.White Gaussian random noise, which is added, makes the signal-to-noise ratio of noisy seismic prospecting record be -5dB, such as
Shown in attached drawing 4 (b).Using Wavelet Algorithm, time-frequency peak filtering and multidimensional space-time time-frequency peak filtering are to containing
Composite traces of making an uproar is handled, to verify the noise removal capability and signal recovery effects of random noise abatement technology of the present invention.Fig. 4
(c) and Fig. 4 (d) is hard -threshold Wavelet Denoising Method and its difference with noisy record, and Fig. 4 (e) and Fig. 4 (f) are respectively a length of using window
The time-frequency peak filtering filter result of 9 sampled points and its difference with noisy record；Fig. 4 (g) and Fig. 4 (h) is respectively to press
The multidimensional space-time time-frequency method result and its difference with noisy record that inventive algorithm (1)-(5) step obtains.
Interpretation of result: as can be seen that three kinds of filtering methods take in terms of random noise abatement from Fig. 4 (a)~4 (h)
Obtained preferably effect, but Wavelet noise-eliminating method and time-frequency peak filtering while cutting down random noise to effective same phase
Axis has different degrees of abatement, and three lineups have from the difference image that can be seen that Wavelet noise-eliminating method in difference image
It remains, the biggish lineups residual in inclination angle is particularly evident in the difference image of time-frequency method method.And random noise of the present invention
Almost without effective lineups residual in the difference image of abatement, random noise method for reducing of the present invention is also just demonstrated in this way and is being pressed
Random noise processed is meanwhile, it is capable to high fidelity restore useful signal.
Actual seismic survey data processing verifying effectiveness of the invention.It is explored and is remembered using certain mining area part actual seismic
Record, sample frequency 1000Hz, as shown in part (a) of attached drawing 5, by Wavelet noise-eliminating method and algorithm application proposed by the present invention
In record processing, processing result is shown and the part (b)-(c) of attached drawing 5 respectively.
Interpretation of result: former actual seismic exploration record and filter result are compared, it has been found that two methods filtered noise
It is all fine to cut down effect, but method proposed by the present invention is better than other two kinds in terms of lineups continuity, clarity two
Method, it was demonstrated that random noise method for reducing of the present invention is protected on width in random noise abatement and useful signal better performance,
More thorough to noise reduction, lineups are also more continuous.
3-d seismic exploration data processing
For the validity for proving inventive algorithm, synthesis 3-d seismic exploration common-source point note is applied the inventive method to
Record, and compared with the filter result of curvlet algorithm, the signal-to-noise ratio and related coefficient recorded after filtering is recorded in table 1.From table
Data we can see that the corresponding signal-to-noise ratio of multidimensional space-time TFPF method and related coefficient in the case where any signal-to-noise ratio
It is all highest, therefore this method preferably can restore useful signal under noise background, improve the letter of noisy earthquake record
It makes an uproar ratio.By the data in observation table, it can be found that converting denoising method compared to Curvelet, random noise of the invention disappears
Subtract method and apparent advantage is maintained for the signal amplitude in low signal-to-noise ratio earthquake record (SNR < 0dB).
Result signal-to-noise ratio (dB) and related coefficient comparison after 1 three kinds of filtering method filtering of table
In conclusion exploring number for abatement actual seismic based on the hyperspace time-frequency method of multidimensional signal feature
According to random noise be effective.The invention takes full advantage of the space characteristics of multi-dimensional earthquake survey data, so that multidimensional space-time
Time-frequency method can be along effective lineups trend pass filtering, therefore is conducive to preferably keep weak useful signal amplitude, increases
Strong lineups continuity.The present invention cuts down under Low SNR compared with common seismic explores random noise method for reducing
It is with the obvious advantage that seismic prospecting random noise and high-fidelity restore useful signal aspect.For improving seismic prospecting data signal-to-noise ratio,
More accurately restoring useful signal has biggish contribution, helps more accurately to realize that subsurface geologic structures are imaged.
Claims (2)
1. a kind of method of hyperspace space-time time-frequency method abatement seismic prospecting random noise, it is characterised in that: including
Following steps:
(1) multidimensional spatio-temporal filtering path is constructed, conventional multi-dimensional earthquake exploration observation data are the subsurface reflective letter of wave detector record
Number and ambient noise superposition, indicate are as follows:
D (x)=S (x)+V (x)
Wherein multi-dimensional earthquake observation data are D (x), and x indicates that hyperspace coordinate, S (x) are effective seismic prospecting signal, V (x)
For random noise；
Multidimensional spatio-temporal filtering path is the filter paths of multidimensional time-frequency method, multidimensional spatio-temporal filtering path is defined as:
Tr (x)=φ (x)
Wherein φ () is M dimension filtering path；
By calculating multidimensional spatio-temporal filtering path parameter, the estimation of multidimensional spatio-temporal filtering path is obtainedIt indicates are as follows:
Wherein | | | | indicate Euclid norm, a indicates filtering path parameter；
Obtain the estimation of multidimensional spatio-temporal filtering pathThe step of it is as follows:
Multidimensional spatio-temporal filtering path is constructed using earthquake record multidimensional characteristic, using similarity function as cost function, prediction
Current road and the lineups of neighboring track move towards, and estimate multidimensional spatio-temporal filtering path function, genuine for multi-dimensional earthquake data, i-th
Shake data are di, per pass data include m sampled point, lineups and neighboring track dkLocal delay is τi,k；To determine the delay
Time, the setting parsing window centered on i-th any point, parsing window inclination angle are θ, and seismic data is similar in calculating parsing window
Coefficient is denoted as PθD changes inclination angle and repeats the above steps, using maximum similarity factor direction as the office in the i-th road Dao Yuk
Portion's delay:
Wherein PθFor the feature extraction operator for the parsing window that inclination angle is θ, similarity factor operator is used here, utilizes recursion method, root
According to the local delay information under the local delay information acquisition being calculated together, then the local delay of pending data can indicate
For τ=[τ1,2,τ2,3,…,τN-1,N], it includes seismic channel number that wherein N, which is seismic data, thus calculates multidimensional spatio-temporal filtering path ginseng
Number obtains the estimation of multidimensional spatio-temporal filtering path
(2) multidimensional data resampling
Data are sampled along multidimensional spatio-temporal filtering path, the sampling road Shi Yi is horizontal axis, using sampled point as the longitudinal axis, by seismic channel
With the point that intersect of filtering path as sampled point, one sample sequence of such each path correspondence, and by path along the longitudinal axis according to
Secondary translation simultaneously extracts sample sequence, the resampling until completing entire seismic prospecting record, obtained resampling data Dr(x),
It indicates are as follows:
(3) resampling data modulating-coding
In multidimensional space-time path domain, counterweight sampled data Dr(x) frequency modulation(PFM) coding is carried out, analytic signal z (x) is converted into,
That is:
Exp is exponential function in formula, and μ is scale factor, and λ is integration variable, and j is imaginary unit；
(4) time-frequency distributions of analytic signal are calculated
The time-frequency distributions of analytic signal are calculated, select pseudo- Eugene Wigner prestige pulling cloth or its high-resolution modification version as needed,
Obtain the pseudo- Eugene Wigner prestige pulling cloth W of analytic signalz(x, f) is indicated are as follows:
Wherein f indicates frequency, and * indicates complex conjugate operation, and υ is integration variable, and the length of window function h (υ) is denoted as the long WL of window, window letter
The determination of number length is as follows:
Wherein fhFor the highest frequency of signal, σ2For random noise variance, C is characteristic parameter, according to the statistics of signal and noise spy
Property describes；
(5) filtering signal is obtained, peak value filtering is carried out to the time-frequency distributions of analytic signal, calculates the time-frequency distributions peak of analytic signal
It is worth estimation of the frequency as useful signal, it may be assumed that
In formulaIt for filtered signal, is filtered by time-frequency domain peak value, filtering signal obtained is the nothing of actual signal
Estimation partially.
2. a kind of side of hyperspace space-time time-frequency method abatement seismic prospecting random noise according to claim 1
Method, it is characterised in that: in step (1), as M=1, φ () is the straight line along time orientation；As M=2, when φ () is
M- direction in space two dimension path can choose the radial alignment for having angle with time orientation, also can choose hyperbolic or parabolic
Curve；When M is more than or equal to 3, then the curve that changes for M dimension space with signal characteristic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710990336.0A CN107783191B (en) | 2018-01-26 | 2018-01-26 | The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710990336.0A CN107783191B (en) | 2018-01-26 | 2018-01-26 | The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107783191A CN107783191A (en) | 2018-03-09 |
| CN107783191B true CN107783191B (en) | 2019-02-15 |
Family
ID=61435194
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710990336.0A Expired - Fee Related CN107783191B (en) | 2018-01-26 | 2018-01-26 | The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107783191B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108646296A (en) * | 2018-05-16 | 2018-10-12 | 吉林大学 | Desert seismic signal noise reduction methods based on Adaptive spectra kurtosis filter |
| CN111366973B (en) * | 2018-12-26 | 2022-08-05 | 中国石油天然气集团有限公司 | Frequency domain noise generation and addition method and device of forward model |
| CN111308561B (en) * | 2020-03-11 | 2020-10-23 | 中国科学院地质与地球物理研究所 | Method for removing strong noise of electromagnetic signal |
| CN114035229B (en) * | 2021-10-26 | 2023-11-10 | 西安石油大学 | Pre-stack seismic data wavelet threshold denoising optimal wavelet base selection method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104980187B (en) * | 2014-04-08 | 2017-09-19 | 华北电力大学 | A kind of method and device of signal denoising processing |
| CN105929447B (en) * | 2016-05-31 | 2017-11-21 | 中国石油大学(华东) | Consider the sparse hyperbola Radon transform methods in change summit of seismic wavelet stretching effect |
| CN107015276B (en) * | 2017-04-18 | 2019-04-12 | 吉林大学 | A kind of adaptive time-frequency method noise-eliminating method based on improvement Hough transformation |
| CN107589454A (en) * | 2017-07-25 | 2018-01-16 | 西安交通大学 | One kind is based on VMD TFPF compacting seismic prospecting random noise methods |
-
2018
- 2018-01-26 CN CN201710990336.0A patent/CN107783191B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN107783191A (en) | 2018-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107783191B (en) | The method of hyperspace space-time time-frequency method abatement seismic prospecting random noise | |
| Wu et al. | Noise attenuation for 2-D seismic data by radial-trace time-frequency peak filtering | |
| Tian et al. | Variable-eccentricity hyperbolic-trace TFPF for seismic random noise attenuation | |
| CN109669212B (en) | Seismic data processing method, stratum quality factor estimation method and device | |
| Xiong et al. | Random-noise attenuation for seismic data by local parallel radial-trace TFPF | |
| CN105700020A (en) | Random noise suppression method and apparatus for seismic data | |
| CN102053273A (en) | Inverse Q filtering method for seismic wave signal | |
| AU2009229124A1 (en) | Method for performing constrained polarization filtering | |
| CN107132579A (en) | A kind of attenuation of seismic wave compensation method for protecting earth formation | |
| US11754744B2 (en) | Surface wave prospecting method for jointly extracting Rayleigh wave frequency dispersion characteristics by seismoelectric field | |
| CN103364832A (en) | Seismic attenuation qualitative estimation method based on self-adaptive optimal kernel time frequency distribution | |
| CN110208856B (en) | Desert complex noise suppression method based on manifold partition 2D-VMD | |
| CN103091714A (en) | Self-adaption surface wave attenuation method | |
| CN106950600B (en) | A kind of minimizing technology of near surface scattering surface wave | |
| CN106019376A (en) | Seismic wave compensation method constructed through frequency driving space-variant Q value model | |
| CN102841380B (en) | A kind of surface relief complex structure Earthquakes grouping of data is concerned with noise attenuation method | |
| CN105182417A (en) | Surface wave separation method and system based on morphological component analysis | |
| CN104570116A (en) | Geological marker bed-based time difference analyzing and correcting method | |
| CN111708087A (en) | Method for suppressing seismic data noise based on DnCNN neural network | |
| CN106019377B (en) | A kind of two-dimensional seismic survey noise remove method based on time-space domain frequency reducing model | |
| CN104635264B (en) | Pre-stack seismic data processing method and device | |
| CN103135133A (en) | Vector noise reduction method and device for multi-component seismic data | |
| CN104280776A (en) | Self-adaption wavelet threshold solving method | |
| CN112578448A (en) | High-precision stratum quality factor extraction method | |
| CN110109179A (en) | Bandwidth compensation processing method, device and equipment |
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: 20190215 Termination date: 20200126 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |