CN114839673A - Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field - Google Patents
Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field Download PDFInfo
- Publication number
- CN114839673A CN114839673A CN202210762892.3A CN202210762892A CN114839673A CN 114839673 A CN114839673 A CN 114839673A CN 202210762892 A CN202210762892 A CN 202210762892A CN 114839673 A CN114839673 A CN 114839673A
- Authority
- CN
- China
- Prior art keywords
- seismic
- data
- matrix
- acquisition
- source
- 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
- 238000000926 separation method Methods 0.000 title claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 71
- 238000000034 method Methods 0.000 claims abstract description 66
- 238000005065 mining Methods 0.000 claims abstract description 51
- 238000013528 artificial neural network Methods 0.000 claims abstract description 31
- 238000003062 neural network model Methods 0.000 claims abstract description 27
- 238000012360 testing method Methods 0.000 claims abstract description 24
- 238000004519 manufacturing process Methods 0.000 claims abstract description 14
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 81
- 238000010606 normalization Methods 0.000 claims description 17
- 238000001514 detection method Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 11
- 238000005520 cutting process Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 9
- 238000004088 simulation Methods 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 239000011541 reaction mixture Substances 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 13
- 238000010586 diagram Methods 0.000 description 18
- 230000008569 process Effects 0.000 description 13
- 238000005422 blasting Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000012952 Resampling Methods 0.000 description 2
- 230000001427 coherent effect Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000010304 firing Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
Images
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
- G01V1/282—Application of seismic models, synthetic seismograms
-
- 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
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 invention belongs to the technical field of submarine seismic exploration, and discloses a separation method, a separation system and computer equipment for a multi-seismic-source high-efficiency acquisition wave field. The method for separating the ocean bottom seismic exploration multi-seismic source high-efficiency acquisition wave fields based on the denoising neural network comprises the following steps: acquiring mixed acquisition data, preprocessing seismic data, manufacturing small-size seismic data, normalizing a training set, building and training a neural network model and testing the separation of acquired wave fields of multiple seismic sources in submarine seismic exploration to obtain a common shot gather record after aliasing noises are removed. The invention uses the trained network model to carry out mixed mining separation in the submarine seismic exploration multi-seismic-source high-efficiency acquisition wave field separation, has high operation speed and less artificial intervention, and ensures that the submarine seismic exploration multi-seismic-source separation effect is good. The invention has small manpower resource investment and low cost in the separation of the multi-seismic-source high-efficiency acquisition wave field in the submarine seismic exploration.
Description
Technical Field
The invention belongs to the technical field of submarine seismic exploration, and particularly relates to a separation method, a separation system and computer equipment for a multi-seismic-source high-efficiency acquisition wave field.
Background
Marine seismic exploration is the fundamental data for marine oil and gas exploration, geological disaster early warning and marine geological science research. An air gun is generally used as a seismic source, a streamer comprising equally spaced detectors is used for receiving, a certain sampling rate and a recording length are designed according to the exploration depth and the exploration purpose, for example, a 1 millisecond sampling rate is used for recording 6 seconds, each detector records 6000 sampling points at 1 millisecond intervals after the seismic source is excited, and then underground imaging is carried out by using the acquired data in an indoor processing stage so as to support subsequent research. In a conventional operation mode, in order to enable wave fields generated by two adjacent guns not to interfere with each other, the firing interval of the air guns is usually larger than a recording period, and the data acquisition density of seismic exploration is very high, so that an exploration work area needs to spend months or even years for field construction, and the field data acquisition cost is huge. In order to save exploration cost and accelerate production cycle, the industry puts forward the requirement of multi-seismic source mixed acquisition, namely, two or more seismic sources are used for blasting at the same time to generate mutually-overlapped wave field records, and then a single-seismic source wave field is separated by using a certain data processing means. With the deep development of marine oil and gas exploration, the marine seismic exploration is gradually switched from the traditional streamer acquisition to the ocean bottom seismic acquisition mode, and the traditional streamer acquisition mode has certain advantages in terms of signal-to-noise ratio, azimuth angle and final imaging quality of signals. In addition, submarine seismic exploration is different from traditional streamer acquisition, and a seismic source and a detector are relatively independent, so that the construction mode has high flexibility, and convenience is provided for multi-seismic source hybrid acquisition.
The aliasing wave field separation method commonly used at present is a denoising method and a sparse inversion method. The denoising method takes the adjacent shot interference as noise, noise suppression is carried out by using methods such as median filtering, shearlet transformation and the like, the mixed data wave field separation is converted into an inversion problem by using a sparse inversion method, and solution is carried out through iterative inversion. The methods have the problems of complicated parameter tests, large human resources, low calculation speed, unclean wave field separation and the like.
In recent years, deep learning techniques have been developed in the field of seismic exploration, and important progress is made in various aspects such as first arrival pickup, noise suppression, multiple attenuation, velocity modeling, seismic data reconstruction and the like. For the problem of offshore high-efficiency acquisition wave field separation, seismic data are converted from a common shot gather to a common survey point gather, a common offset gather or a common central point gather, and the like, a wave field generated by an original main seismic source still has certain coherence, while a wave field generated by an auxiliary seismic source is not coherent, and shows certain randomness, so that the wave field separation can be carried out by using a denoising method in a deep learning network. For the trained network model, when the method is applied to wave field separation of other data, positive and negative transformation operations of different domains are not required, and parameters are not required to be filled manually, so that the computer computation amount and the manual workload are superior to those of the traditional method, and the method is more beneficial to industrial application.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) in the prior art, mixed mining separation is not carried out in the process of separating the multi-seismic-source high-efficiency acquisition wave field in the submarine seismic exploration, so that the operation speed is low, and the separation effect and the accuracy of the submarine seismic exploration are poor.
(2) In the prior art, a large number of parameter tests are needed in the process of ocean bottom seismic exploration multi-seismic source high-efficiency acquisition wave field separation, the human resource investment is large, and the cost is high.
Disclosure of Invention
In order to overcome the problems in the related art, the disclosed embodiment of the invention provides a multi-seismic-source efficient acquisition wave field separation method, a separation system and computer equipment, and particularly relates to a sea bottom seismic exploration multi-seismic-source efficient acquisition wave field separation method based on a denoising neural network.
The technical scheme is as follows: the method for separating the ocean bottom seismic exploration multi-seismic source high-efficiency acquisition wave field based on the denoising neural network comprises the following steps:
obtaining non-mixed mining records through simulation or field acquisition, wherein the non-mixed mining records comprise a main seismic source and an auxiliary seismic source, and the main seismic source and the auxiliary seismic source are added through random time delay to obtain mixed mining data;
preprocessing seismic data based on the obtained non-mixed mining record data and mixed mining data;
making small-size seismic data based on the preprocessed seismic data;
carrying out normalization processing on the manufactured small-size seismic data by using a training set;
building and training a neural network model by using the normalized training set;
and testing the separation of the acquisition wave fields of the multiple seismic sources for the submarine seismic exploration by using the trained neural network model to obtain a common shot gather record after aliasing noises are removed.
In one embodiment, in the acquisition of the commingled data, every two shots in the non-commingled recording are mutually used as a main seismic source and an auxiliary seismic source, and the main seismic source and the auxiliary seismic sources are added by adopting a random time delay of +/-1 s to obtain the commingled data.
In one embodiment, in the seismic data preprocessing, the seismic data are extracted from the common shot set to the common demodulation point set, then the top cutting is carried out on the same cutting curve for the mixed acquisition record and the unmixed acquisition record, and the recorded amplitude value above the first arrival wave is set to be 0.
In one embodiment, in the production of small-size seismic data, longitudinal and transverse thinning sampling is carried out on the seismic data, each sample contains recorded global information, and an m × n small matrix is trained through the samples containing the global information;
the method for rarefaction sampling comprises the following steps: firstly, judging the row number and the column number of a co-detection dot set data matrix, if the row number of an original matrix is less than 500, zero padding is carried out on the lower part to 500 rows, and if the column number of the original matrix is less than 100 columns, zero padding is carried out on the right side to 100 columns; if the number of rows is not an integer multiple of 500 or the number of columns is not an integer multiple of 100, the last row is filled with zero downwards to the integer multiple of 500 or the last column is filled with zero to the integer multiple of 100 rightwards, and then small matrixes of 500 multiplied by 100 are extracted at equal intervals; for a matrix with n times of 500 rows and m times of 100 columns, obtaining an m multiplied by n small matrix after thinning and sampling; for the small-size data, 50% of the small-size data are selected as a training set, and the other 50% of the small-size data are selected as a verification set.
In one embodiment, in performing the normalization process of the training set, a matrix of data pairs consisting of samples and labels is formed
Sum matrix
The normalized training set is made as follows:
for matrix
Sum matrix
The maximum of the absolute values of the elements therein is found, respectively, and then the larger one of the two maximum values is taken:
then the matrix is processed
Sum matrix
All elements in (a) are divided by the maximum value:
and (5) performing operations of equations (1) to (5) on all data pairs consisting of the samples and the labels to finish the normalization of the training set.
In one embodiment, in the building of the neural network model, the built neural network model comprises 17 layers in total, wherein the 1 st layer is a two-dimensional convolution layer plus an nn. The second to 16 th layers are a two-dimensional convolutional layer plus a BN layer and an nn. prime () active layer, where the parameters of the convolutional layer are image _ channels =1, n _ channels =64, kernel _ size =3, stride =1, padding =1, bias = False; the 17 th layer of the network is a convolutional layer without a BN layer.
In one embodiment, in the training of the neural network model, the training parameter through the network is batch _ size =64, 64 samples are used for each training, the number of training rounds epoch =500, the variation between lr =1e-3 to 1e-6 is learned, the height of the samples is 500, the width is 100, the loss function uses the mean square error, and the mean square error function is defined as:
in the formula (6), the reaction mixture is,
and
respectively the output of the neural network and the label,
is composed of
And
and the number of the medium elements is determined, the optimizer uses an Adam optimizer, and the model parameters are stored on a disk after training is completed.
In one embodiment, in the test, the trained neural network model is applied to the actual mixed mining or artificially synthesized mixed mining record; firstly, loading trained network models from a disk, and then loading data in a test set one by one; for each 500 x 100 matrix
Firstly, normalization processing is carried out:
firstly, the maximum value of the absolute values of elements in the matrix is obtained, and then each element in the matrix is divided by the maximum value; for matrix
Output through a neural network to obtain
Finally is moved to
Reverse normalization:
and (3) carrying out operations from the formula (7) to the formula (9) on each element in the test set, filling each small matrix subjected to aliasing noise back to the corresponding position of the large matrix, and finally extracting data from the common detection point set back to the common shot set to obtain a final common shot set record subjected to aliasing noise removal.
Another object of the present invention is to provide a denoising neural network-based ocean bottom seismic exploration multi-source efficient acquisition wave field separation system implementing the separation method, which includes:
the mixed mining data acquisition module is used for acquiring non-mixed mining records, namely single seismic source records, by means of simulation or field acquisition, wherein every two shots in the non-mixed mining records are a main seismic source and an auxiliary seismic source, and the recorded main seismic source and the auxiliary seismic sources are added to acquire mixed mining data by means of random time delay of +/-1 s;
the seismic data preprocessing module is used for extracting seismic data from the common shot set to the common demodulation point set, then carrying out top cutting on the same cutting curve for mixed acquisition recording and non-mixed acquisition recording, and setting the recording amplitude value above the first arrival wave as 0;
the small-size seismic data manufacturing module is used for manufacturing large-size seismic data into small-size data;
a normalized training set module for a data pair matrix composed of a sample and a label
Sum matrix
Making normalized training set and pairing matrix
Sum matrix
Respectively solving the maximum value of the absolute value of the element, and then taking the larger one of the two maximum values; then the matrix is processed
Sum matrix
All elements in (1) are divided by the maximum value; performing matrix operation on all data pairs composed of samples and labels
Sum matrix
Normalized training to matrix
Sum matrix
Dividing all the elements in the training set by the maximum value to finish the normalization of the training set;
building a network model module for building a neural network model;
the neural network training module is used for training and optimizing a neural network model and storing parameters of the neural network model on a disk;
and the test set test module is used for applying the trained neural network model to actual mixed mining or artificially synthesized mixed mining records.
It is a further object of the present invention to provide a computer apparatus comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to execute the method for efficiently acquiring wavefield separation for ocean bottom seismic exploration based on denoised neural network.
By combining all the technical schemes, the invention has the advantages and positive effects that:
first, aiming at the technical problems existing in the prior art and the difficulty in solving the problems, the technical problems to be solved by the technical scheme of the present invention are closely combined with results, data and the like in the research and development process, and some creative technical effects are brought after the problems are solved. The specific description is as follows: aiming at the requirements of submarine seismic exploration on high-efficiency mixed acquisition wave field separation, a neural network suitable for mixed acquisition and separation is designed, a data preprocessing method, a data set manufacturing strategy and a training strategy are designed, and the combination of the methods can enable a trained neural network model to be effectively converged, so that a satisfactory mixed acquisition and separation effect is obtained, and support is provided for separating a single seismic source wave field from a multi-seismic source mixed acquisition wave field.
Secondly, considering the technical solution as a whole or from the perspective of products, the technical effects and advantages of the technical solution to be protected by the present invention are specifically described as follows: the trained network model is used for mixed mining separation in the submarine seismic exploration multi-seismic-source efficient acquisition wave field separation, the operation speed is high, manual intervention is less, and the submarine seismic exploration multi-seismic-source separation effect is good. The invention has small manpower resource investment and low cost in the separation of the multi-seismic-source high-efficiency acquisition wave field in the submarine seismic exploration.
Third, as an inventive supplementary proof of the claims of the present invention, there are also presented several important aspects: (1) the expected income and commercial value after the technical scheme of the invention is converted are as follows: for an actual exploration work area, the field data acquisition time is usually several weeks to several months, and the exploration cost is generally in direct proportion to the exploration time. (2) The technical scheme of the invention is more beneficial to ecological protection: in the submarine seismic exploration process, the ships, air gun ships, logistics support ships and the like which are collected and released by the instrument can influence marine organisms in the area, for example, the energy released by blasting of an air gun seismic source can shake or damage surrounding fishes.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure;
FIG. 1 is a flow chart of a method for separating ocean bottom seismic exploration multi-seismic-source high-efficiency acquisition wave fields based on a denoising neural network, which is provided by the embodiment of the invention;
FIG. 2 is a schematic diagram of a denoising neural network-based ocean bottom seismic exploration multi-seismic-source efficient acquisition wave field separation system provided by an embodiment of the invention;
FIG. 3 is a velocity model and shot point geophone position distribution diagram for wave equation forward modeling according to an embodiment of the present invention;
FIG. 4(a) is a 1 st shot wavefield diagram of seismic data common shot gather wave simulated by forward modeling of a wave equation according to an embodiment of the present invention;
FIG. 4(b) is a 801 st shot field diagram of seismic data common shot gather wave simulated by wave equation forward modeling according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a hybrid acquisition approach provided by an embodiment of the present invention;
FIG. 6 is a cross-sectional view of an aliased record common shot gather provided by an embodiment of the invention;
fig. 7(a) is a diagram of a common detector point set unmixed acquisition wave field according to an embodiment of the present invention;
FIG. 7(b) is a diagram of a co-detection point set hybrid wavefield provided by an embodiment of the present invention;
FIG. 8(a) is a wave field diagram before top cut provided by an embodiment of the present invention;
FIG. 8(b) is a diagram of wavefields after top-cutting in FIG. 8(a) provided by an embodiment of the present invention;
FIG. 9 is a schematic structural diagram of a hybrid mining separation neural network according to an embodiment of the present invention;
FIG. 10 is a graph of the variation of the loss value of the training process provided by an embodiment of the present invention;
fig. 11(a) is a 60 th detection point non-mixed acquisition record chart provided by the embodiment of the present invention;
fig. 11(b) is a 60 th mixed sampling record chart of the detection point provided in the embodiment of the present invention;
FIG. 11(c) is a recorded graph of the mixed mining separation of the method of the present invention provided by the embodiment of the present invention;
fig. 12(a) is a diagram of separation effect of non-commingled mining records of the 800 th cannon provided by the embodiment of the invention;
fig. 12(b) is a mixed mining recording effect diagram provided by the embodiment of the present invention;
FIG. 12(c) is a diagram of the mixed mining separation recording effect of the method of the present invention provided by the embodiment of the present invention;
in the figure: 1. a mixed mining data acquisition module; 2. a seismic data preprocessing module; 3. a small-size seismic data production module; 4. a normalized training set module; 5. building a network model module; 6. a neural network training module; 7. and a test set test module.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather should be construed as broadly as the present invention is capable of modification in various respects, all without departing from the spirit and scope of the present invention.
First, illustrative embodiments:
the method for separating the multi-seismic-source efficient acquisition wave field in the submarine seismic exploration based on the denoising neural network, provided by the embodiment of the invention, is used for designing the neural network suitable for mixed acquisition and separation aiming at the requirement of the high-efficiency mixed acquisition wave field in the submarine seismic exploration, and designing the data preprocessing method, the data set manufacturing strategy and the training strategy.
Example 1
As shown in fig. 1, the method for separating the multi-seismic-source high-efficiency acquisition wavefield in ocean bottom seismic exploration based on the denoising neural network provided by the embodiment of the invention comprises the following steps:
s101: and (5) making mixed mining data. Firstly, non-mixed-mining records, namely single seismic source records, are obtained through simulation or field acquisition, every two shots in the non-mixed-mining records are mutually used as a main seismic source and an auxiliary seismic source, random time delay of +/-1 s is adopted, and the two records are added to obtain mixed-mining data.
S102: and (5) preprocessing the seismic data. Firstly, seismic data are extracted from a common shot set to a common demodulation point set, then the top cutting is carried out on the same cutting curve for mixed mining recording and non-mixed mining recording, and the recording amplitude value above the first arrival wave is set to be 0.
S103: and (5) manufacturing small size. In the process of training the sample deep learning neural network, the video memory is an extremely scarce resource and directly determines whether the training can be carried out or not. Generally speaking, the video memory required by training increases with the increase of the size of input data, the network depth, the number of convolution kernels and the number of training samples in each batch, and the network depth, the number of convolution kernels and the number of training samples in each batch are key parameters for feature extraction of a neural network, and the network performance is influenced by reducing the parameters. Therefore, it is necessary to make small-size data for large-size seismic data. The method for making small-size data by denoising problems generally divides the data into a plurality of small blocks, and the method of the invention longitudinally and transversely thins and samples the data, so that each sample can contain recorded global information but not local information, and a neural network trained by the sample containing the global information has better mixed-sampling and separation effects. The specific thinning method comprises the following steps: firstly, judging the row number and the column number of a co-detection dot set data matrix, if the row number of an original matrix is less than 500, zero is filled at the lower part to 500 rows, and if the column number of the original matrix is less than 100 columns, zero is filled at the right side to 100 columns. If the number of rows is not an integer multiple of 500 or the number of columns is not an integer multiple of 100, the last row is zero filled down to an integer multiple of 500 or the last column is zero filled right to an integer multiple of 100, and then the small matrix of 500 × 100 is extracted at equal intervals. For a matrix with n times the number of rows and m times the number of columns of 500, a small m × n matrix can be obtained by thinning out the samples. For the small-size data, 50% of the small-size data are selected as a training set, and the other 50% of the small-size data are selected as a verification set.
S104: and (5) normalizing the training set. Data pair matrix composed of sample and label
Sum matrix
The normalized training set is made as follows:
namely, the pair matrix
And
the maximum of the absolute values of the elements therein is found, respectively, and then the larger one of the two maximum values is taken:
then the matrix is processed
And
all elements in (a) are divided by the maximum value:
for all samples and labelsAnd (4) carrying out operations of the formulas (1) to (5) on the data pairs, so that the normalization of the training set can be completed. Wherein the matrix
Is a sample matrix, a matrix
Is a matrix of labels.
S105: and (5) building a network model. The built neural network model comprises 17 layers in total, wherein the 1 st layer is a two-dimensional convolution layer plus an nn. PReLU () activation layer, wherein the parameters of the convolution layer are image _ channels =1, n _ channels =64, kernel _ size =3, stride =1, padding =1, bias = True; the second to 16 th layers are a two-dimensional convolutional layer plus a BN layer and an nn. prime () active layer, where the parameters of the convolutional layer are image _ channels =1, n _ channels =64, kernel _ size =3, stride =1, padding =1, bias = False; the 17 th layer of the network is a convolutional layer without a BN layer.
S106: and (5) carrying out neural network training. The training parameter of the neural network is batch _ size =64, 64 samples are used in each training, the training round number epoch =500, the variation between lr =1e-3 to 1e-6 is learned, the height of the samples is 500, the width is 100, the loss function uses the mean square error, and the definition of the mean square error function is as follows:
in the formula (6), the reaction mixture is,
and
respectively the output of the neural network and the label,
is composed of
And
the number of the elements in (B). The optimizer uses an Adam optimizer. And after the training is finished, saving the model parameters to a disk.
S107: and testing on the test set, and applying the trained model to actual mixed mining or artificially synthesized mixed mining records. Firstly, loading the trained network model from a disk, and then loading the data in the test set one by one. For each 500 x 100 matrix
Firstly, normalization processing is carried out:
that is, the maximum value of the absolute values of the elements in the matrix is first obtained, and then the maximum value is divided by each element in the matrix. To pair
Output through a neural network to obtain
Finally is moved to
Reverse normalization:
and (3) carrying out operations from a formula (7) to a formula (9) on each element in the test set, filling each small matrix subjected to aliasing noise back to the corresponding position of the large matrix, and finally extracting data from the common detection point set back to the common shot set to obtain a final common shot set record subjected to aliasing noise removal.
Example 2
As shown in fig. 2, an embodiment of the present invention provides a denoising neural network-based ocean bottom seismic exploration multi-source efficient acquisition wavefield separation system, including:
and the mixed acquisition data acquisition module 1 is used for acquiring non-mixed acquisition records, namely single seismic source records, by means of simulation acquisition or field acquisition, wherein every two shots in the non-mixed acquisition records are a main seismic source and an auxiliary seismic source, random time delay of +/-1 s is adopted, and the two records are added to obtain mixed acquisition data.
And the seismic data preprocessing module 2 is used for extracting seismic data from the common shot set to the common demodulation point set, then performing top cutting on the same cutting curve for mixed acquisition recording and non-mixed acquisition recording, and setting the recording amplitude value above the first arrival wave as 0.
A small-size seismic data production module 3. The method is used for manufacturing small-size data for large-size seismic data.
The training set module 4 is normalized. Matrix for data pairs composed for a sample and label
Sum matrix
Making normalized training set, pair matrix
And
respectively solving the maximum value of the absolute value of the element, and then taking the larger one of the two maximum values; then the matrix is processed
And
all elements in (1) are divided by the maximum value; performing matrix operation on all data pairs composed of samples and labels
And
normalized training to matrix
And
the normalization of the training set can be completed by the operation step of dividing all the elements in the training set by the maximum value.
And building a network model module 5. The neural network model is constructed;
the neural network training module 6 is used for training and optimizing the neural network model and storing parameters of the neural network model on a disk;
and the test set test module 7 is used for applying the trained neural network model to actual mixed mining or artificially synthesized mixed mining records.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
For the information interaction, execution process and other contents between the above-mentioned devices/units, because the embodiments of the method of the present invention are based on the same concept, the specific functions and technical effects thereof can be referred to the method embodiments specifically, and are not described herein again.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
II, application embodiment:
an embodiment of the present invention further provides a computer device, where the computer device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, the processor implementing the steps of any of the various method embodiments described above when executing the computer program.
Embodiments of the present invention further provide a computer-readable storage medium, where a computer program is stored, and when the computer program is executed by a processor, the steps in the above method embodiments may be implemented.
The embodiment of the present invention further provides an information data processing terminal, where the information data processing terminal is configured to provide a user input interface to implement the steps in the above method embodiments when implemented on an electronic device, and the information data processing terminal is not limited to a mobile phone, a computer, or a switch.
The embodiment of the present invention further provides a server, where the server is configured to provide a user input interface to implement the steps in the above method embodiments when implemented on an electronic device.
Embodiments of the present invention provide a computer program product, which, when running on an electronic device, enables the electronic device to implement the steps in the above method embodiments when executed.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the method according to the embodiments of the present invention may be implemented by a computer program, which may be stored in a computer-readable storage medium and used for instructing related hardware to implement the steps of the embodiments of the method according to the embodiments of the present invention. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to a photographing apparatus/terminal apparatus, a recording medium, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signal, telecommunication signal, and software distribution medium. Such as a usb-disk, a removable hard disk, a magnetic or optical disk, etc.
Third, evidence of the relevant effects of the examples:
in order to verify the beneficial effect of the method, the embodiment of the invention designs a model for receiving seismic signals on the sea surface blasting seabed. FIG. 3 shows a velocity model for forward wave equation modeling, which is 8000m wide and 3000m deep. Blasting is carried out on the sea surface at intervals of 5m, the total number of the blasts is 1600, and the detection points are distributed on the sea bottom at intervals of 50m, and the total number of the detection points is 160. For each shot firing, all of the demodulator sites on the seafloor are receiving. The wavelet used for simulation is a Rake wavelet, the dominant frequency is 30Hz, the sampling interval is set to be 0.5ms, and the recording length is 4 s. The forward modeling obtains shot gather records, which are shown in fig. 4(a) and 4(b) as the records for the 1 st and 801 th shots. Wherein, FIG. 4(a) is the 1 st shot wave field diagram; FIG. 4(b) is a 801 th shot wave field diagram; then, the aliasing records are made according to the mode of fig. 5, two seismic source ships travel to the same direction at a distance of 4000m, the seismic source 1 and the seismic source 2 have random excitation time delay of +/-1 s, so that the shot gather records of the 1 st shot and the 801 th shot, the 2 nd shot and the 802 nd shot …, the 800 th shot and the 1600 th shot which are mutually interfered in pairs can be obtained, and the construction time is shortened to be half of the original construction time.
FIG. 6 is an aliased record of shot 1, which contains the interference of the 801 th shot. The records of the common shot gather are extracted into a common geophone point gather, which is shown in fig. 7(a) and 7(b), and are unmixed acquisition wave field and mixed acquisition wave field of the common geophone point gather, respectively, so that for the main seismic source record, the signals in the common geophone point gather are still continuous and coherent, while the auxiliary seismic source record shows irregular interference. Wherein, fig. 7(a) is a common detector point set non-mixed acquisition wave field diagram; FIG. 7(b) is a diagram of a co-detector point set mixed-acquisition wave field; for seismic records, the part above the first arrival wave is not involved in imaging, and if the part of signals is reserved in the neural network training process, the final model parameters are easily influenced, so that the mixed mining separation effect is influenced. Therefore, for the unmixed data and mixed data of the co-detection point set, the top-cut is performed according to the method described in step two of the present invention, as shown in fig. 8(a) and fig. 8(b), which are schematic diagrams of the top-cut of the co-detection point set data, where fig. 8(a) is the wave field before the top-cut, and fig. 8(b) is the wave field after the top-cut of fig. 8 (a). For the common-detector-point wave-set field after top-cut, the data size of each common-detector-point wave-set is 8000 × 1600, according to the method described in the third step of the present invention, 16 times of resampling is performed in the row number direction and 16 times of resampling is performed in the column number direction, so that 2596 small-size data with the size of 500 × 100 are obtained, 40960 data pairs are generated in total by 160 detector-points, and 50% of the data pairs are selected as a training set, and the other 50% are selected as a test set. For each data pair consisting of a sample and a label of 500 × 100 size, the normalization process is performed according to the method described in step four of the present invention.
The neural network shown in fig. 9 is constructed according to the method described in the fifth step of the invention, then the neural network training is carried out according to the sixth step of the invention, and a loss curve in the training process is shown in fig. 10. After the training is finished, the verification set is verified by the method described in the seventh step of the invention. Fig. 11(a), fig. 11(b) and fig. 11(c) are respectively the 60 th detection point non-mixed collection record, mixed collection record and mixed collection separation record of the method of the present invention. As can be seen, after the mixed mining separation, most aliasing noises are suppressed, and the mixed mining separation record is very similar to the original non-mixed mining record. And after the mixed mining of the common geophone point set is separated, the common shot set is extracted again, and the 800 th shot non-mixed mining record, the mixed mining record and the mixed mining separation record of the method are respectively shown in fig. 12(a), fig. 12(b) and fig. 12 (c). Similarly, it can be seen that the method of the present invention suppresses most of the neighboring source records, and the present source records are effectively retained.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, and any modification, equivalent replacement, and improvement made by those skilled in the art within the technical scope of the present invention disclosed herein, which is within the spirit and principle of the present invention, should be covered by the present invention.
Claims (10)
1. The separation method for the multi-seismic source high-efficiency acquisition wave field is characterized by comprising the following steps of:
obtaining non-mixed mining records through simulation or field acquisition, wherein the non-mixed mining records comprise a main seismic source and an auxiliary seismic source, and the main seismic source and the auxiliary seismic source are added through random time delay to obtain mixed mining data;
preprocessing seismic data based on the obtained non-mixed mining record data and mixed mining data;
making small-size seismic data based on the preprocessed seismic data;
carrying out normalization processing on the manufactured small-size seismic data by using a training set;
building and training a neural network model by using the normalized training set;
and testing the separation of the acquisition wave fields of the multiple seismic sources for the submarine seismic exploration by using the trained neural network model to obtain a common shot gather record after aliasing noises are removed.
2. The method for separating the multi-source high-efficiency acquisition wave fields according to claim 1, wherein in the acquisition of the mixed acquisition data, every two shots in the non-mixed acquisition record are a main source and an auxiliary source, and the main source and the auxiliary sources are added by adopting random time delay of +/-1 s to obtain the mixed acquisition data.
3. The method for multi-source wavefield separation with high efficiency of acquisition as claimed in claim 1, wherein in the seismic data preprocessing, the seismic data are extracted from the common shot set to the common wave detection point set, and then the cut-off curve is performed for the mixed acquisition record and the unmixed acquisition record, and the recorded amplitude value above the first arrival wave is set to 0.
4. The method for separating the multi-source efficient acquisition wavefield of claim 1, wherein in the production of the small-size seismic data, the seismic data are longitudinally and transversely sampled, each sample contains the recorded global information, and an m × n small matrix is trained by the samples containing the global information;
the method for rarefaction sampling comprises the following steps: firstly, judging the row number and the column number of a co-detection dot set data matrix, if the row number of an original matrix is less than 500, zero padding is carried out on the lower part to 500 rows, and if the column number of the original matrix is less than 100 columns, zero padding is carried out on the right side to 100 columns; if the number of rows is not an integer multiple of 500 or the number of columns is not an integer multiple of 100, the last row is filled with zero downwards to the integer multiple of 500 or the last column is filled with zero to the integer multiple of 100 rightwards, and then small matrixes of 500 multiplied by 100 are extracted at equal intervals; for a matrix with n times of 500 rows and m times of 100 columns, obtaining an m multiplied by n small matrix after thinning and sampling; for the small-size data, 50% of the small-size data are selected as a training set, and the other 50% of the small-size data are selected as a verification set.
5. The method for multi-source efficient acquisition wavefield separation of claim 1, wherein the normalization of the training set is performed using a matrix of data pairs consisting of samples and labels
Sum matrix
The normalized training set is made as follows:
for matrix
Sum matrix
The maximum of the absolute values of the elements therein is found, respectively, and then the larger one of the two maximum values is taken:
then the matrix is processed
Sum matrix
All elements in (1) are divided by the maximum value:
and (5) performing operations of equations (1) to (5) on all data pairs consisting of the samples and the labels to finish the normalization of the training set.
6. The multi-source efficient acquisition wavefield separation method of claim 1, wherein in the building of the neural network model, the built neural network model comprises 17 layers, wherein the 1 st layer is a two-dimensional convolution layer plus an nn. The second to 16 th layers are a two-dimensional convolutional layer plus a BN layer and an nn. prime () active layer, where the parameters of the convolutional layer are image _ channels =1, n _ channels =64, kernel _ size =3, stride =1, padding =1, bias = False; the 17 th layer of the network is a convolutional layer without a BN layer.
7. The method for separating the multi-source efficient acquisition wavefield of claim 1, wherein in the training of the neural network model, the training parameter of the network is batch _ size =64, each training uses 64 samples, the training round number epoch =500 learns the variation between lr =1e-3 and 1e-6, the height of the samples is 500, the width is 100, the loss function uses the mean square error, and the mean square error function is defined as:
in the formula (6), the reaction mixture is,
and
respectively the output of the neural network and the label,
is composed of
And
and the number of the medium elements is determined, the optimizer uses an Adam optimizer, and the model parameters are stored on a disk after training is completed.
8. The multi-source efficient acquisition wavefield separation method of claim 7, wherein in the test, the trained neural network model is applied to actual commingled acquisition or artificially synthesized commingled acquisition records; firstly, loading trained network models from a disk, and then loading data in a test set one by one; for each 500 x 100 matrix
Firstly, normalization processing is carried out:
firstly, the maximum value of the absolute values of elements in the matrix is obtained, and then each element in the matrix is divided by the maximum value; for matrix
Output through a neural network to obtain
Finally is moved to
Reverse normalization:
and (3) carrying out operations from the formula (7) to the formula (9) on each element in the test set, filling each small matrix subjected to aliasing noise back to the corresponding position of the large matrix, and finally extracting data from the common detection point set back to the common shot set to obtain a final common shot set record subjected to aliasing noise removal.
9. A separation system for implementing the multi-source high-efficiency acquisition wavefield separation method of any one of claims 1-8, wherein the separation system comprises:
the mixed mining data acquisition module (1) is used for acquiring non-mixed mining records, namely single seismic source records, by means of simulation or field acquisition, wherein every two shots in the non-mixed mining records are a main seismic source and an auxiliary seismic source, and the recorded main seismic source and the auxiliary seismic sources are added to acquire mixed mining data by means of random time delay of +/-1 s;
the seismic data preprocessing module (2) is used for extracting seismic data from the common shot set to the common demodulation point set, then performing top cutting on the same cutting curve for mixed acquisition recording and non-mixed acquisition recording, and setting the recording amplitude value above the first arrival wave as 0;
the small-size seismic data manufacturing module (3) is used for manufacturing large-size seismic data into small-size data;
a normalized training set module (4) for a matrix of data pairs consisting of a sample and a label
Sum matrix
Making normalized training set, pair matrix
Sum matrix
Respectively maximizing the absolute value of the element thereinThe value, then the larger of these two maximum values; then the matrix is processed
Sum matrix
All elements in (1) are divided by the maximum value; performing matrix operation on all data pairs composed of samples and labels
Sum matrix
Normalized training to matrix
Sum matrix
Dividing all the elements in the training set by the maximum value to finish the normalization of the training set;
a network model building module (5) is used for building a neural network model;
the neural network training module (6) is used for training and optimizing the neural network model and storing parameters of the neural network model on a disk;
and the test set test module (7) is used for applying the trained neural network model to actual mixed mining or artificially synthesized mixed mining records.
10. A computer arrangement, characterized in that the computer arrangement comprises a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to carry out the method of multi-source efficient acquisition wavefield separation according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210762892.3A CN114839673B (en) | 2022-07-01 | 2022-07-01 | Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210762892.3A CN114839673B (en) | 2022-07-01 | 2022-07-01 | Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114839673A true CN114839673A (en) | 2022-08-02 |
| CN114839673B CN114839673B (en) | 2022-09-23 |
Family
ID=82574152
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210762892.3A Active CN114839673B (en) | 2022-07-01 | 2022-07-01 | Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114839673B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115267899A (en) * | 2022-08-15 | 2022-11-01 | 河北地质大学 | DnCNN mixed seismic source seismic data separation method and system based on boundary preservation |
| CN115620113A (en) * | 2022-12-20 | 2023-01-17 | 成都理工大学 | Elastic wave vector separation method for generating countermeasure network based on deep convolution |
| CN115829001A (en) * | 2022-11-08 | 2023-03-21 | 中国科学院地质与地球物理研究所 | Transient electromagnetic-excitation field separation and multi-parameter information extraction method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110568485A (en) * | 2019-09-06 | 2019-12-13 | 广州海洋地质调查局 | neural network-based multi-channel seismic continuous recording and separating method |
| CN112327356A (en) * | 2020-10-30 | 2021-02-05 | 中国海洋大学 | Aliasing record separation method based on inphase axis iterative tracking extraction |
| CN112987099A (en) * | 2021-04-19 | 2021-06-18 | 吉林大学 | Low-frequency seismic data reconstruction method based on multi-seismic-source convolutional neural network |
| CN113570041A (en) * | 2021-07-29 | 2021-10-29 | 中国地质大学（北京） | Neural network and method for suppressing seismic data noise of marine optical fiber streamer by using neural network |
| CN113885077A (en) * | 2021-09-30 | 2022-01-04 | 太原理工大学 | Multi-seismic-source seismic data separation method based on deep learning |
| CN114167500A (en) * | 2020-09-10 | 2022-03-11 | 中国石油天然气股份有限公司 | Method and device for separating adjacent shot interference based on convolutional neural network |
| CN114296134A (en) * | 2021-12-24 | 2022-04-08 | 西安交通大学 | Deep convolutional network seismic data unmixing method and system |
-
2022
- 2022-07-01 CN CN202210762892.3A patent/CN114839673B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110568485A (en) * | 2019-09-06 | 2019-12-13 | 广州海洋地质调查局 | neural network-based multi-channel seismic continuous recording and separating method |
| CN114167500A (en) * | 2020-09-10 | 2022-03-11 | 中国石油天然气股份有限公司 | Method and device for separating adjacent shot interference based on convolutional neural network |
| CN112327356A (en) * | 2020-10-30 | 2021-02-05 | 中国海洋大学 | Aliasing record separation method based on inphase axis iterative tracking extraction |
| CN112987099A (en) * | 2021-04-19 | 2021-06-18 | 吉林大学 | Low-frequency seismic data reconstruction method based on multi-seismic-source convolutional neural network |
| CN113570041A (en) * | 2021-07-29 | 2021-10-29 | 中国地质大学（北京） | Neural network and method for suppressing seismic data noise of marine optical fiber streamer by using neural network |
| CN113885077A (en) * | 2021-09-30 | 2022-01-04 | 太原理工大学 | Multi-seismic-source seismic data separation method based on deep learning |
| CN114296134A (en) * | 2021-12-24 | 2022-04-08 | 西安交通大学 | Deep convolutional network seismic data unmixing method and system |
Non-Patent Citations (4)
| Title |
|---|
| 宋家文 等: "基于稀疏反演的高效混采数据分离方法", 《石油地球物理勘探》 * |
| 王汉闯 等: "多震源混合地震数据分离方法(英文)", 《APPLIED GEOPHYSICS》 * |
| 石太昆 等: "混采地震数据高效高精度分离处理方法研究进展", 《石油物探》 * |
| 童思友 等: "基于卷积神经网络的海洋地震勘探拖缆混采数据分离", 《中国海洋大学学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115267899A (en) * | 2022-08-15 | 2022-11-01 | 河北地质大学 | DnCNN mixed seismic source seismic data separation method and system based on boundary preservation |
| CN115267899B (en) * | 2022-08-15 | 2024-01-12 | 河北地质大学 | DnCNN mixed source seismic data separation method and system based on boundary preservation |
| CN115829001A (en) * | 2022-11-08 | 2023-03-21 | 中国科学院地质与地球物理研究所 | Transient electromagnetic-excitation field separation and multi-parameter information extraction method and system |
| US11892588B1 (en) | 2022-11-08 | 2024-02-06 | Institute Of Geology And Geophysics, Chinese Academy Of Sciences | Method and system for transient electromagnetic-induced polarization field separation and multi-parameter information extraction |
| CN115620113A (en) * | 2022-12-20 | 2023-01-17 | 成都理工大学 | Elastic wave vector separation method for generating countermeasure network based on deep convolution |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114839673B (en) | 2022-09-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114839673B (en) | Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field | |
| CN106842320B (en) | The parallel 3-D seismics wave field generation method of GPU and system | |
| CN102395902B (en) | Seismic imaging systems and methods employing a fast target-oriented illumination calculation | |
| US8559270B2 (en) | Method for separating independent simultaneous sources | |
| CN107505654B (en) | Full waveform inversion method based on earthquake record integral | |
| US20100054082A1 (en) | Reverse-time depth migration with reduced memory requirements | |
| US9541659B2 (en) | Noise removal from 3D seismic representation | |
| US11221426B2 (en) | Method for improved processing of data with time overlapping recordings of energy sources | |
| US10215869B2 (en) | System and method of estimating anisotropy properties of geological formations using a self-adjoint pseudoacoustic wave propagator | |
| CN113885077B (en) | Multi-source seismic data separation method based on deep learning | |
| CN107340540B (en) | Direction wave decomposition method, device and the computer storage medium of elastic wave field | |
| CN111781635B (en) | Seabed four-component elastic wave Gaussian beam depth migration method and device | |
| CN109738952A (en) | The direct offset imaging method in passive source based on full waveform inversion driving | |
| WO2021127382A1 (en) | Full waveform inversion in the midpoint-offset domain | |
| EP4031910B1 (en) | Noise attenuation methods applied during simultaneous source deblending and separation | |
| CN113391351A (en) | Method for analyzing and extracting ore-concentrated area structure based on passive source seismic wave field | |
| CN113885079A (en) | Elastic wave field decoupling-based high-precision multi-azimuth reverse-time seismic source imaging method | |
| US11422277B2 (en) | Seismic data filtering based on distances between seismic sources | |
| CN110907989A (en) | Method and system for reconstructing quasi-ground seismic reflection wave imaging | |
| US20230129626A1 (en) | Separation of Seismic Sources by Joint Interpolation and Deblending | |
| CN114296134B (en) | Deep convolutional network seismic data unmixing method and system | |
| CN112882101B (en) | Random noise attenuation method and device for pre-stack seismic data | |
| CN110618459B (en) | Seismic data processing method and device | |
| US20220413175A1 (en) | Enhanced projection on convex sets for interpolation and deblending | |
| CN112558181B (en) | Sensitivity calibration method and device for near-field detector of marine air gun |
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 |