CN114167500B - Method and device for separating adjacent gun interference based on convolutional neural network - Google Patents
Method and device for separating adjacent gun interference based on convolutional neural network Download PDFInfo
- Publication number
- CN114167500B CN114167500B CN202010944979.3A CN202010944979A CN114167500B CN 114167500 B CN114167500 B CN 114167500B CN 202010944979 A CN202010944979 A CN 202010944979A CN 114167500 B CN114167500 B CN 114167500B
- Authority
- CN
- China
- Prior art keywords
- data
- adjacent
- layer
- interference
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 74
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000000926 separation method Methods 0.000 claims abstract description 145
- 238000012549 training Methods 0.000 claims abstract description 81
- 238000012545 processing Methods 0.000 claims abstract description 52
- 238000000605 extraction Methods 0.000 claims abstract description 43
- 238000012360 testing method Methods 0.000 claims abstract description 36
- 230000006870 function Effects 0.000 claims description 38
- 210000002569 neuron Anatomy 0.000 claims description 25
- 230000004913 activation Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 13
- 238000003860 storage Methods 0.000 claims description 11
- 238000001514 detection method Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000011056 performance test Methods 0.000 claims description 6
- 230000001131 transforming effect Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 19
- 238000005516 engineering process Methods 0.000 description 11
- 238000004891 communication Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 239000000872 buffer Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005422 blasting Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000009466 transformation Effects 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Remote Sensing (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A method and a device for separating adjacent gun interference based on a convolutional neural network, wherein the method comprises the following steps: selecting any two pieces of data in the seismic data, and overlapping adjacent-gun interference data with main gun data to obtain first input data, wherein the adjacent-gun interference data is first tag data; obtaining actual data through a high-efficiency aliasing acquisition mode, performing adjacent gun interference separation on the actual data by using a sparse inversion method, taking the data before separation as second input data, and taking the data after separation as second tag data; according to the gather extraction method, the data are transformed into a detector gather and divided into a test gather and a training gather; initializing parameters of the convolutional neural network to obtain an initial model, inputting a training set into the initial model to obtain an adjacent gun interference separation model, and carrying out adjacent gun interference separation processing. The invention can effectively separate the adjacent gun noise, improve the separation efficiency and the precision and the signal-to-noise ratio of the seismic data, and provide high-quality basic data for the quantitative interpretation and the reservoir prediction of the seismic data.
Description
Technical Field
The invention relates to the technical field of seismic data processing, in particular to a method and a device for separating adjacent gun interference based on a convolutional neural network.
Background
The technology of seismic acquisition is an important ring in the seismic exploration process, and uses a blasting mode to excite seismic waves to cause vibration of underground medium, and then uses a receiving instrument called a detector to collect signals reflected from underground to the ground according to time sequence. In order to avoid the cross interference of the received signals, a long waiting time is set between adjacent cannons in the traditional earthquake acquisition process, but the on-site construction period is long, and the acquisition efficiency is low.
In order to meet the requirements of high-precision seismic exploration, high-density high-efficiency seismic data acquisition technology has been rapidly developed in recent years. The technology uses a plurality of groups of seismic sources, and is called an aliasing acquisition technology when the conditions of a timing rule are met. The technology shortens the acquisition time, greatly improves the seismic acquisition efficiency and reduces the cost. However, because the adjacent shot excitation time interval is shorter, the aliasing of the seismic waves from different shots occurs, and the signal-to-noise ratio and the imaging quality of the seismic data are seriously reduced. Therefore, adjacent gun interference separation is a necessary link for efficient aliasing acquisition data processing.
At present, the adjacent gun interference separation method is mainly divided into two types, one type is a filtering-based method: because the sources of the high-efficiency aliasing acquisition method are mostly excited by adopting a random time delay mode, the effective signals are continuous in other gathers (a common-detector gather, a common-center gather or a common-offset gather) except for the shot gather, and aliasing noise shows random discrete characteristics. Therefore, the filtering-based adjacent gun interference separation method mainly utilizes the random discrete features of adjacent gun interference on a non-gun set for separation. However, when the adjacent gun interference is serious, the method can damage effective signals, so that the adjacent gun interference separation effect is not ideal; another class is inversion-based methods: the aliasing process of the efficient aliasing acquisition technology can be regarded as a forward problem, and the adjacent gun interference separation can be regarded as an inversion problem. The separation method based on sparse inversion mainly utilizes the sparsity characteristic of the effective signals in the transformation domain, continuously contracts the threshold value in the iteration process, gradually extracts the effective signals, eliminates the adjacent gun interference noise, and improves the adjacent gun interference separation result. Compared with a separation method based on filtering, the sparse inversion method has better mixed data separation effect, but has higher calculation cost and is difficult to popularize in practical application. Particularly for high-efficiency mixed data, the separation method provides more serious challenges due to high aliasing degree, low signal-to-noise ratio and large data volume.
Therefore, a method for separating adjacent cannon interference with good separation effect and high efficiency is needed at present.
Disclosure of Invention
The embodiment of the invention mainly aims to provide a method and a device for separating adjacent gun interference based on a convolutional neural network, which realize the adjacent gun interference separation with good separation effect and high efficiency.
In order to achieve the above object, an embodiment of the present invention provides a method for separating adjacent gun interference based on a convolutional neural network, the method comprising:
selecting any two pieces of data in the seismic data as main shot data and adjacent shot interference data respectively, superposing the adjacent shot interference data with the main shot data according to random time delay, and taking the obtained aliasing data as first input data of network learning, wherein the adjacent shot interference data is taken as first tag data of the network learning;
Obtaining actual data containing adjacent gun interference in a high-efficiency aliasing acquisition mode, carrying out adjacent gun interference separation on the actual data by using a sparse inversion method, taking the actual data before the adjacent gun interference separation as second input data of network learning, and taking the actual data after the adjacent gun interference separation as second tag data of the network learning;
According to a gather extraction method, the first input data, the second input data, the first tag data and the second tag data are transformed into a detector gather, and the transformed data are divided into a test gather and a training gather according to a preset proportion;
initializing preset parameters of a convolutional neural network to obtain an initial model of the convolutional neural network, inputting the training set into the initial model of the convolutional neural network to perform model training to obtain a neighbor-cannon interference separation model, and performing neighbor-cannon interference separation processing by using the neighbor-cannon interference separation model.
Optionally, in an embodiment of the present invention, inputting the training set to an initial model of the convolutional neural network for model training, and obtaining the adjacent gun interference separation model includes: inputting the training set into an initial model of the convolutional neural network for model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition; and inputting the test set into the adjacent gun interference separation model for model test to obtain second output layer data, and comparing the second output layer data with the label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model.
Optionally, in an embodiment of the present invention, inputting the training set into an initial model of the convolutional neural network for model training, and obtaining the first output layer data includes: the training set is used as input layer data and is input to an initial model of the convolutional neural network for model training; performing convolution operation on the input layer data and n convolution kernels, and inputting an operation result into a ReLU activation function to obtain basic characteristics of the input layer data, wherein n is a positive integer; processing basic features of the input layer data by a plurality of feature extraction units to obtain essential features of adjacent gun interference data, wherein the feature extraction units comprise a convolution layer with n convolution kernels, a BN layer and a ReLu activation function layer, and the BN layer is used for carrying out normalization processing on the data; and carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
Optionally, in an embodiment of the present invention, the performing the adjacent-shot interference separation processing by using the adjacent-shot interference separation model includes: transforming the seismic data containing the adjacent gun interference to be processed into a detector point gather through a channel gather extraction method, and inputting the data transformed into the detector point gather into the adjacent gun interference separation model to obtain separated seismic data; and converting the separated seismic data into a shot set by a channel set extraction method, wherein the seismic data converted into the shot set are used for seismic data processing and interpretation.
The embodiment of the invention also provides a device for separating adjacent gun interference based on the convolutional neural network, which comprises:
The first data module is used for selecting any two pieces of data in the seismic data to be respectively used as main shot data and adjacent shot interference data, superposing the adjacent shot interference data with the main shot data according to random time delay, and taking the obtained aliasing data as first input data of network learning, and the adjacent shot interference data as first tag data of network learning;
The second data module is used for obtaining actual data containing adjacent gun interference in a high-efficiency aliasing acquisition mode, carrying out adjacent gun interference separation on the actual data by using a sparse inversion method, taking the actual data before the adjacent gun interference separation as second input data of network learning, and taking the actual data after the adjacent gun interference separation as second tag data of the network learning;
the data dividing module is used for converting the first input data, the second input data, the first tag data and the second tag data into a detector point gather according to a gather extraction method, and dividing the converted data into a test set and a training set according to a preset proportion;
The adjacent-gun interference separation module is used for initializing the parameters of a preset convolutional neural network to obtain an initial model of the convolutional neural network, inputting the training set into the initial model of the convolutional neural network to perform model training to obtain an adjacent-gun interference separation model, and performing adjacent-gun interference separation processing by using the adjacent-gun interference separation model.
Optionally, in an embodiment of the present invention, the adjacent gun interference separation module includes: the model training unit is used for inputting the training set into an initial model of the convolutional neural network to perform model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition; and the model test unit is used for inputting the test set into the adjacent gun interference separation model to perform model test to obtain second output layer data, and comparing the second output layer data with the label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model.
Optionally, in an embodiment of the present invention, the model training unit includes: an input layer data subunit, configured to input the training set as input layer data to an initial model of the convolutional neural network for model training; an input data feature subunit, configured to perform convolution operation on the input layer data and n convolution kernels, and input an operation result into a ReLU activation function to obtain a basic feature of the input layer data, where n is a positive integer; the device comprises an interference data characteristic subunit, a characteristic extraction unit and a data processing unit, wherein the interference data characteristic subunit is used for processing basic characteristics of input layer data through a plurality of characteristic extraction units to obtain essential characteristics of adjacent cannon interference data, the characteristic extraction unit comprises a convolution layer with n convolution kernels, a BN layer and a ReLu activation function layer, and the BN layer is used for carrying out normalization processing on the data; and the output layer data subunit is used for carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
Optionally, in an embodiment of the present invention, the adjacent gun interference separation module further includes: the seismic data processing unit is used for converting the seismic data which are to be processed and contain adjacent gun interference into a wave detection point gather through a channel gather extraction method, inputting the data converted into the wave detection point gather into the adjacent gun interference separation model, and obtaining separated seismic data; the seismic data application unit is used for converting the separated seismic data into a shot set through a shot set extraction method, and the seismic data converted into the shot set are used for seismic data processing and interpretation.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the above method when executing the program.
The present invention also provides a computer readable storage medium storing a computer program for executing the above method.
According to the invention, the adjacent gun interference separation is carried out through the convolutional neural network, and under the background of strong adjacent gun interference, the separation efficiency and accuracy are greatly improved while the adjacent gun noise can be effectively separated, the signal-to-noise ratio of the seismic data is improved, and high-quality basic data is provided for subsequent seismic data quantitative interpretation and high-accuracy reservoir prediction.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for separating adjacent gun interference based on a convolutional neural network in an embodiment of the invention;
FIG. 2 is a flowchart of training and testing an adjacent gun interference separation model in an embodiment of the invention;
FIG. 3 is a flowchart of a process for calculating an adjacent gun interference separation model in an embodiment of the invention;
FIG. 4 is a flow chart of the separation and application of the adjacent shot interference of the seismic data in practice in the embodiment of the invention;
FIG. 5 is a schematic diagram of seismic data including adjacent shot interference in an embodiment of the invention;
FIG. 6 is a schematic diagram of seismic data without adjacent shot interference in an embodiment of the invention;
FIG. 7 is a schematic diagram of the result of the separation process of adjacent cannon interference in the embodiment of the present invention;
FIG. 8 is a schematic diagram of the residual error between the result of the adjacent shot interference separation process and the seismic data in the embodiment of the invention;
FIG. 9 is a schematic representation of seismic data in accordance with an embodiment of the invention;
FIG. 10A is a schematic diagram of an oilfield seismic data after separation of adjacent shot interference using a sparse inversion method in accordance with an embodiment of the present invention;
FIG. 10B is a schematic diagram of adjacent shot interference for separating seismic data from an oilfield by using a sparse inversion method in accordance with one embodiment of the present invention;
FIG. 11A is a schematic diagram of seismic data from an oilfield after separation of adjacent shot interference using the present invention in accordance with one embodiment of the present invention;
FIG. 11B is a schematic diagram of adjacent shot interference for an oilfield seismic data separated using the present invention in accordance with an embodiment of the present invention;
FIG. 12 is a schematic diagram illustrating residual errors of data after adjacent shot interference separation performed by the sparse inversion method and seismic data of an oilfield after adjacent shot interference separation performed by the present invention in an embodiment of the present invention;
FIGS. 13A-13C are graphs showing data amplitude comparison of adjacent-shot interference separation processing performed by the adjacent-shot interference separation processing and the sparse inversion method according to an embodiment of the present invention;
Fig. 14 is a schematic structural diagram of a convolutional neural network-based adjacent gun interference separation device according to an embodiment of the present invention;
fig. 15 is a schematic structural diagram of an electronic device according to an embodiment of the invention.
Detailed Description
The embodiment of the invention provides a method and a device for separating adjacent gun interference based on a convolutional neural network.
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Fig. 1 is a flowchart of a method for separating adjacent gun interference based on a convolutional neural network, where the method includes:
step S1, selecting any two pieces of data in the seismic data as main shot data and adjacent shot interference data respectively, superposing the adjacent shot interference data with the main shot data according to random time delay, and taking the obtained aliasing data as first input data of network learning, wherein the adjacent shot interference data is taken as first tag data of the network learning.
The method comprises the steps of obtaining seismic data through a conventional seismic acquisition technology, selecting any two cannons, taking one cannon as a main cannon, taking the other cannon as adjacent cannon interference, overlapping the adjacent cannon with the main cannon according to a certain random time delay, and taking the obtained aliasing data as first input data of network learning and the adjacent cannon interference data as first tag data of network learning.
And S2, obtaining actual data containing adjacent gun interference in a high-efficiency aliasing acquisition mode, performing adjacent gun interference separation on the actual data by using a sparse inversion method, taking the actual data before the adjacent gun interference separation as second input data of network learning, and taking the actual data after the adjacent gun interference separation as second label data of the network learning.
The method comprises the steps of obtaining data actually containing adjacent gun interference through a high-efficiency aliasing acquisition technology, performing adjacent gun interference separation by using a sparse inversion-based method, taking the actual data containing the adjacent gun interference before separation as second input data of network learning, and taking the separated adjacent gun interference data as second tag data of the network learning.
And S3, according to a gather extraction method, the first input data, the second input data, the first tag data and the second tag data are converted into a detector point gather, and the converted data are divided into a test set and a training set according to a preset proportion.
The input data and the label data obtained in the step S1 and the step S2 are converted into the detector point gather according to a conventional gather extraction method, and the test gather and the training gather are divided according to a certain proportion, wherein the default is that the test gather accounts for 10% of the total gather.
And S4, initializing parameters of a preset convolutional neural network to obtain an initial model of the convolutional neural network, inputting the training set into the initial model of the convolutional neural network to perform model training to obtain a neighbor monitor interference separation model, and performing neighbor monitor interference separation processing by using the neighbor monitor interference separation model.
The method comprises the steps of initializing preset parameters of a convolutional neural network, wherein specific parameter setting comprises the steps of randomly initializing weight parameters and bias parameters of a convolutional kernel, and setting learning parameters of a BN layer to be 1,0 to obtain an initial model of the convolutional neural network.
And inputting the training set into an initial model of the convolutional neural network to perform model training to obtain the adjacent gun interference separation model. The seismic data containing adjacent shot interference is transformed into a wave detection point gather according to a conventional gather extraction method, the data of the wave detection point gather is input into an adjacent shot interference separation model to obtain separated data, and then the separated data is transformed into a shot gather according to the conventional gather extraction method, so that adjacent shot interference separation processing of the seismic data is completed.
As an embodiment of the present invention, as shown in fig. 2, inputting a training set into an initial model of the convolutional neural network to perform model training, and obtaining a neighbor monitor interference separation model includes:
And S21, inputting the training set into an initial model of the convolutional neural network to perform model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition.
Inputting a training set of the seismic data into an initial model of the convolutional neural network, acquiring data of an output layer, and judging whether the prediction accuracy of the current convolutional neural network meets a preset condition according to comparison between the data of the output layer and the label data in the training set; if the prediction accuracy meets the preset condition, the network parameters (including the weight parameters, the bias parameters and the learning parameters of the BN layer) of the current convolutional neural network are adjusted through a back propagation algorithm of the convolutional neural network until the prediction accuracy meets the preset condition, and the convolutional neural network with the prediction accuracy meeting the preset condition is determined to be a neighbor monitor interference separation model. The back propagation algorithm of convolutional neural networks is a process of obtaining optimal network parameters by minimizing the objective function (loss function) of the following equation.
Wherein Θ is the convolutional neural network, x is the input data in the training set of the seismic data, and y is the tag data in the training set of the seismic data. The minimization of the objective function (formula (1)) can be achieved by a gradient descent method, and the iterative process is as follows:
1) Let the output value of the ith neuron node of the first layer be From the output layer to the L layer, there areThe product of Hadamard is shown as follows.
2) Starting from layer l=l-1 to layer l=2 (L is the total number of layers), the back propagation algorithm calculation is performed according to the following three cases:
a) If the current layer is a convolutional layer, there are
B) If the current layer is ReLu active function layers, there are
C) If the current layer is BN layer, there areWherein the method comprises the steps ofΣ 2 is the variance of the current layer data.
3) From layer 2 to layer L, the network parameters are updated according to the following three cases:
a) If the current layer is a convolutional layer, there are
B) If the current layer is ReLu activation function layer, no network parameters can be updated;
c) If the current layer is BN layer, there are
And S22, inputting the test set into the adjacent gun interference separation model for model test to obtain second output layer data, and comparing the second output layer data with the label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model.
The method comprises the steps of inputting a test set of seismic data into a neighbor monitor interference separation model, comparing data of an output layer with label data of the test set, and analyzing denoising performance of the neighbor monitor interference separation model. If the requirements are not met, the seismic data are selected again to manufacture the data labels, and the model training and testing process is repeated.
In this embodiment, as shown in fig. 3, inputting the training set to the initial model of the convolutional neural network for model training, and obtaining the first output layer data includes:
And S31, taking the training set as input layer data, and inputting the input layer data into an initial model of the convolutional neural network to perform model training. As shown in fig. 5, the adjacent cannon interference separation network includes 50 layers in total. Firstly, input data of a training set of the seismic data is used as input layer data of a network.
And S32, carrying out convolution operation on the input layer data and n convolution kernels, and inputting an operation result into a ReLU activation function to obtain the basic characteristics of the input layer data, wherein n is a positive integer.
The input layer data and n convolution kernels are subjected to convolution operation (default n=64), and the operation result is input into a ReLU activation function to obtain basic characteristics of the input layer data. The convolution operation of the input layer data and the n convolution kernels can be realized through a convolution layer of a convolution neural network, and each output of the convolution layer is a value of a plurality of inputs of the combination convolution:
Where M l-1 is the set of neuron nodes of layer 1 (M l-1 represents the input layer when l-1 = 0), For the data of the j neuron nodes of the first layer,For the data of the ith neuron node of layer 1 (input layer when l-1 = 0),To connect the convolution kernel weight matrix of the ith neuron node of layer L-1 with the j neuron nodes of layer L,For the bias matrix of j neuron nodes of the first layer, the symbol "×" represents convolution operation. In general, the size of a neuronal receptive field is determined by the size of the convolution kernel, and when the convolution kernel is too small or too large, no effective local features can be extracted. Therefore, the size of the convolution kernel needs to be determined by combining the factors such as the seismic data, the adjacent gun interference separation principle, the operation speed and the like. A convolution kernel of 3*3 or 5*5 is typically used to convolve each time window of the input data with a convolution kernel, and finally each time window obtains a two-dimensional characteristic signal.
Inputting the operation result into the ReLU activation function can be realized through ReLu activation function layers of the convolutional neural network. ReLu the activation function layer refers to whenWhen the data is the data of the ith neuron node of the first layer-1, the data of the corresponding ith neuron node of the first layerCan be expressed as:
and step S33, processing the basic features of the input layer data by a plurality of feature extraction units to obtain the basic features of the adjacent gun interference data, wherein the feature extraction units comprise a convolution layer with n convolution kernels, a BN layer and a ReLu activation function layer, and the BN layer is used for carrying out normalization processing on the data.
Wherein a network structure comprising a convolution layer (comprising n convolution kernels), a BN layer and a ReLu activation function layer is defined as the feature extraction unit. The BN layer is used for normalizing data, and the basic process is as follows: for data x= { x 1,x2,…,xm }, m is the data length, and the data y i＝BNγ,β(xi is obtained after the BN layer processing, where i=1, 2, …, m, γ, β is the learning parameter, BN γ,β(xi) can be achieved by the formula (2):
Wherein epsilon is a non-zero minimum value and the default value is 1e-6. And processing the basic features of the input layer data by 15 feature extraction units to obtain the essential features of the adjacent cannon interference data.
And step S34, carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
And carrying out convolution operation on the essential characteristics of the adjacent cannon interference data and the n convolution kernels to obtain first output layer data of the convolution neural network.
As an embodiment of the present invention, as shown in fig. 4, performing the adjacent-shot interference separation process using the adjacent-shot interference separation model includes:
And S41, converting the seismic data containing the adjacent gun interference to be processed into a detector point gather through a channel set extraction method, and inputting the data converted into the detector point gather into the adjacent gun interference separation model to obtain separated seismic data.
Step S42, converting the separated seismic data into a shot gather through a shot gather extraction method, inputting the seismic data converted into the shot gather into industrial processing software, and processing and explaining the seismic data by a seismic data processor according to a conventional seismic data processing flow.
The seismic data (shown in fig. 5) containing adjacent gun interference is transformed into a detector point gather according to a conventional gather extraction method, the data of the detector point gather is input into an adjacent gun interference separation model to obtain separated data, and then the separated data is transformed into a gun gather (shown in fig. 7) according to the conventional gather extraction method. Fig. 6 is seismic model data without adjacent shot interference, and fig. 8 is the residual of the processing result and model data of the present invention, showing the effectiveness of the present invention. By effectively separating the adjacent gun noise, the separation efficiency and the separation precision are greatly improved, the signal to noise ratio of the seismic data is improved, and high-quality basic data can be provided for subsequent fine seismic data processing and interpretation. In the figure, trace represents the track number, and Time represents Time.
In a specific embodiment of the invention, a certain oilfield A block is taken as an application example, the processed data is marine seismic data, and synchronous source excitation is adopted, so that the adjacent shot interference is serious, and the further processing and interpretation of the seismic data are influenced. FIG. 9 is seismic data for the present survey area, with severe interference from adjacent shots. Fig. 10A is a result of performing adjacent gun interference separation by using a sparse inversion method, wherein adjacent gun interference is effectively suppressed, and fig. 10B is an adjacent gun interference separated by using a sparse inversion method, which is effective in separating the adjacent gun interference, but requires a large amount of computer resources, so that data which can be processed in each iteration are fewer, and the processing efficiency is lower; fig. 11A shows data after adjacent gun interference separation by using the present invention, and fig. 11B shows adjacent gun interference separated by the present invention. Fig. 12 shows a comparison of data obtained after suppressing the adjacent gun interference by two methods, which differ slightly. Fig. 13A-13C respectively select the Shallow (Shallow Layer), middle (MIDDLE LAYER), and deep (DEEP LAYER) three layers of the processing results of the two methods for Amplitude comparison, where the positions of the three layers are shown as black lines in the right small diagrams in fig. 13A-13C, and the ordinate Amplitude represents the Amplitude. In the middle and shallow layers, the method has higher similarity with the inversion method, but the method has better amplitude preservation at the position shown by an arrow. In the deep layer, compared with an inversion method, the method is more similar to original data, and the method can effectively protect effective signals and greatly improve the operation efficiency while effectively suppressing the adjacent gun interference. In the figure, raw represents an aliased seismic record, DL represents a separation result of the invention, and Inversion represents an Inversion method separation result.
According to the invention, the adjacent gun interference separation is carried out through the convolutional neural network, and under the background of strong adjacent gun interference, the separation efficiency and accuracy are greatly improved while the adjacent gun noise can be effectively separated, the signal-to-noise ratio of the seismic data is improved, and high-quality basic data is provided for subsequent seismic data quantitative interpretation and high-accuracy reservoir prediction.
Fig. 14 is a schematic structural diagram of an adjacent gun interference separation device based on a convolutional neural network, where the device includes:
The first data module 10 is configured to select any two pieces of data in the seismic data, and respectively use the two pieces of data as main shot data and adjacent shot interference data, superimpose the adjacent shot interference data with the main shot data according to random time delay, and obtain aliasing data as first input data of network learning, where the adjacent shot interference data is used as first tag data of network learning.
The method comprises the steps of obtaining seismic data through a conventional seismic acquisition technology, selecting any two cannons, taking one cannon as a main cannon, taking the other cannon as adjacent cannon interference, overlapping the adjacent cannon with the main cannon according to a certain random time delay, and taking the obtained aliasing data as first input data of network learning and the adjacent cannon interference data as first tag data of network learning.
The second data module 20 is configured to obtain actual data containing adjacent gun interference in an efficient aliasing acquisition manner, perform adjacent gun interference separation on the actual data by using a sparse inversion method, use actual data before the adjacent gun interference separation as second input data of network learning, and use the actual data after the adjacent gun interference separation as second tag data of network learning.
The method comprises the steps of obtaining data actually containing adjacent gun interference through a high-efficiency aliasing acquisition technology, performing adjacent gun interference separation by using a sparse inversion-based method, taking the actual data containing the adjacent gun interference before separation as second input data of network learning, and taking the separated adjacent gun interference data as second tag data of the network learning.
The data dividing module 30 is configured to transform the first input data, the second input data, the first tag data, and the second tag data into a detector gather according to a gather extraction method, and divide the transformed data into a test set and a training set according to a preset ratio.
The input data and the label data are converted into the wave-detecting point gather according to a conventional gather extraction method, and the test gather and the training gather are divided according to a certain proportion, wherein the default is that the test gather accounts for 10% of the total gather.
The adjacent-gun interference separation module 40 is configured to initialize parameters of a preset convolutional neural network to obtain an initial model of the convolutional neural network, input the training set to the initial model of the convolutional neural network to perform model training to obtain an adjacent-gun interference separation model, and perform adjacent-gun interference separation processing by using the adjacent-gun interference separation model.
The method comprises the steps of initializing preset parameters of a convolutional neural network, wherein specific parameter setting comprises the steps of randomly initializing weight parameters and bias parameters of a convolutional kernel, and setting learning parameters of a BN layer to be 1,0 to obtain an initial model of the convolutional neural network.
And inputting the training set into an initial model of the convolutional neural network to perform model training to obtain the adjacent gun interference separation model. The seismic data containing adjacent shot interference is transformed into a wave detection point gather according to a conventional gather extraction method, the data of the wave detection point gather is input into an adjacent shot interference separation model to obtain separated data, and then the separated data is transformed into a shot gather according to the conventional gather extraction method, so that adjacent shot interference separation processing of the seismic data is completed.
As an embodiment of the present invention, the adjacent gun interference separation module includes:
The model training unit is used for inputting the training set into an initial model of the convolutional neural network to perform model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition;
and the model test unit is used for inputting the test set into the adjacent gun interference separation model to perform model test to obtain second output layer data, and comparing the second output layer data with the label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model.
In this embodiment, the model training unit includes:
An input layer data subunit, configured to input the training set as input layer data to an initial model of the convolutional neural network for model training;
An input data feature subunit, configured to perform convolution operation on the input layer data and n convolution kernels, and input an operation result into a ReLU activation function to obtain a basic feature of the input layer data, where n is a positive integer;
The device comprises an interference data characteristic subunit, a characteristic extraction unit and a data processing unit, wherein the interference data characteristic subunit is used for processing basic characteristics of input layer data through a plurality of characteristic extraction units to obtain essential characteristics of adjacent cannon interference data, the characteristic extraction unit comprises a convolution layer with n convolution kernels, a BN layer and a ReLu activation function layer, and the BN layer is used for carrying out normalization processing on the data;
And the output layer data subunit is used for carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
As an embodiment of the present invention, the adjacent gun interference separation module further includes:
The seismic data processing unit is used for converting the seismic data which are to be processed and contain adjacent gun interference into a wave detection point gather through a channel gather extraction method, inputting the data converted into the wave detection point gather into the adjacent gun interference separation model, and obtaining separated seismic data;
The seismic data application unit is used for converting the separated seismic data into a shot set through a shot set extraction method, and the seismic data converted into the shot set are used for seismic data processing and interpretation.
Based on the same application conception as the adjacent gun interference separation method based on the convolutional neural network, the invention also provides the adjacent gun interference separation device based on the convolutional neural network. Because the principle of the convolution neural network-based adjacent gun interference separation device for solving the problem is similar to that of the convolution neural network-based adjacent gun interference separation method, the implementation of the convolution neural network-based adjacent gun interference separation device can be referred to the implementation of the convolution neural network-based adjacent gun interference separation method, and repeated parts are omitted.
According to the invention, the adjacent gun interference separation is carried out through the convolutional neural network, and under the background of strong adjacent gun interference, the separation efficiency and accuracy are greatly improved while the adjacent gun noise can be effectively separated, the signal-to-noise ratio of the seismic data is improved, and high-quality basic data is provided for subsequent seismic data quantitative interpretation and high-accuracy reservoir prediction.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the above method when executing the program.
The present invention also provides a computer readable storage medium storing a computer program for executing the above method.
As shown in fig. 15, the electronic device 600 may further include: a communication module 110, an input unit 120, an audio processing unit 130, a display 160, a power supply 170. It is noted that the electronic device 600 need not include all of the components shown in fig. 15; in addition, the electronic device 600 may further include components not shown in fig. 15, to which reference is made to the related art.
As shown in fig. 15, the central processor 100, sometimes also referred to as a controller or operational control, may include a microprocessor or other processor device and/or logic device, which central processor 100 receives inputs and controls the operation of the various components of the electronic device 600.
The memory 140 may be, for example, one or more of a buffer, a flash memory, a hard drive, a removable media, a volatile memory, a non-volatile memory, or other suitable device. The information about failure may be stored, and a program for executing the information may be stored. And the central processor 100 can execute the program stored in the memory 140 to realize information storage or processing, etc.
The input unit 120 provides an input to the central processor 100. The input unit 120 is, for example, a key or a touch input device. The power supply 170 is used to provide power to the electronic device 600. The display 160 is used for displaying display objects such as images and characters. The display may be, for example, but not limited to, an LCD display.
The memory 140 may be a solid state memory such as Read Only Memory (ROM), random Access Memory (RAM), SIM card, or the like. But also a memory which holds information even when powered down, can be selectively erased and provided with further data, an example of which is sometimes referred to as EPROM or the like. Memory 140 may also be some other type of device. Memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application/function storage 142, the application/function storage 142 for storing application programs and function programs or a flow for executing operations of the electronic device 600 by the central processor 100.
The memory 140 may also include a data store 143, the data store 143 for storing data, such as contacts, digital data, pictures, sounds, and/or any other data used by the electronic device. The driver storage 144 of the memory 140 may include various drivers of the electronic device for communication functions and/or for performing other functions of the electronic device (e.g., messaging applications, address book applications, etc.).
The communication module 110 is a transmitter/receiver 110 that transmits and receives signals via an antenna 111. A communication module (transmitter/receiver) 110 is coupled to the central processor 100 to provide an input signal and receive an output signal, which may be the same as in the case of a conventional mobile communication terminal.
Based on different communication technologies, a plurality of communication modules 110, such as a cellular network module, a bluetooth module, and/or a wireless local area network module, etc., may be provided in the same electronic device. The communication module (transmitter/receiver) 110 is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide audio output via the speaker 131 and to receive audio input from the microphone 132 to implement usual telecommunication functions. The audio processor 130 may include any suitable buffers, decoders, amplifiers and so forth. In addition, the audio processor 130 is also coupled to the central processor 100 so that sound can be recorded locally through the microphone 132 and so that sound stored locally can be played through the speaker 131.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The principles and embodiments of the present invention have been described in detail with reference to specific examples, which are provided to facilitate understanding of the method and core ideas of the present invention; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present invention, the present description should not be construed as limiting the present invention in view of the above.
Claims (6)
1. The method for separating adjacent gun interference based on the convolutional neural network is characterized by comprising the following steps of:
selecting any two pieces of data in the seismic data as main shot data and adjacent shot interference data respectively, superposing the adjacent shot interference data with the main shot data according to random time delay, and taking the obtained aliasing data as first input data of network learning, wherein the adjacent shot interference data is taken as first tag data of the network learning;
Obtaining actual data containing adjacent gun interference in a high-efficiency aliasing acquisition mode, carrying out adjacent gun interference separation on the actual data by using a sparse inversion method, taking the actual data before the adjacent gun interference separation as second input data of network learning, and taking the separated adjacent gun interference data as second label data of the network learning;
According to a gather extraction method, the first input data, the second input data, the first tag data and the second tag data are transformed into a detector gather, and the transformed data are divided into a test gather and a training gather according to a preset proportion;
initializing preset parameters of a convolutional neural network to obtain an initial model of the convolutional neural network, inputting the training set into the initial model of the convolutional neural network for model training to obtain an adjacent-cannon interference separation model, and carrying out adjacent-cannon interference separation processing by using the adjacent-cannon interference separation model;
the step of inputting the training set to the initial model of the convolutional neural network for model training, and the step of obtaining the adjacent gun interference separation model comprises the following steps:
Inputting the training set into an initial model of the convolutional neural network for model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition;
Inputting the test set into the adjacent gun interference separation model for model test to obtain second output layer data, and comparing the second output layer data with label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model;
the step of inputting the training set to the initial model of the convolutional neural network for model training, and the step of obtaining a first output layer data comprises the following steps:
the training set is used as input layer data and is input to an initial model of the convolutional neural network for model training;
performing convolution operation on the input layer data and n convolution kernels, inputting an operation result into a ReLU activation function, and obtaining basic characteristics of the input layer data by using the following formula, wherein n is a positive integer;
wherein M l-1 is the set of neuron nodes of layer 1, Data for the jth neuron node of the first layer,Is the data of the ith neuron node of the layer 1,To connect the convolution kernel weight matrix of the ith neuron node of the first-1 layer with the jth neuron node of the first layer,For the bias matrix of the jth neuron node of the first layer, the symbol "×" represents convolution operation;
Inputting the operation result into ReLu activation function, and realizing through ReLu activation function layer of convolutional neural network; wherein ReLu activates the function layer when When the data is the data of the ith neuron node of the first layer-1, the data of the corresponding ith neuron node of the first layerExpressed as:
Processing basic features of the input layer data by a plurality of feature extraction units to obtain essential features of adjacent gun interference data, wherein the feature extraction units comprise a convolution layer with n convolution kernels, a BN layer and a ReLu activation function layer, the BN layer is used for carrying out normalization processing on the data, and for the data x= { x 1,x2,…,xm }, m is the data length, and the data y i＝BNγ,β(xi) is obtained after the processing of the BN layer, wherein i=1, 2, the main numbers, m, gamma, beta are learning parameters, and BN γ,β(xi) is realized by using the following formula:
Wherein epsilon is a non-zero minimum value, and the default value is 1 multiplied by 10 -6;
and carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
2. The method for separating adjacent gun interference based on convolutional neural network as set forth in claim 1, wherein said performing the adjacent gun interference separation process by using the adjacent gun interference separation model comprises:
Transforming the seismic data containing the adjacent gun interference to be processed into a detector point gather through a channel gather extraction method, and inputting the data transformed into the detector point gather into the adjacent gun interference separation model to obtain separated seismic data;
And converting the separated seismic data into a shot set by a channel set extraction method, wherein the seismic data converted into the shot set are used for seismic data processing and interpretation.
3. An adjacent gun interference separation device based on a convolutional neural network, which is characterized by comprising:
The first data module is used for selecting any two pieces of data in the seismic data to be respectively used as main shot data and adjacent shot interference data, superposing the adjacent shot interference data with the main shot data according to random time delay, and taking the obtained aliasing data as first input data of network learning, and the adjacent shot interference data as first tag data of network learning;
the second data module is used for obtaining actual data containing adjacent gun interference in a high-efficiency aliasing acquisition mode, carrying out adjacent gun interference separation on the actual data by using a sparse inversion method, taking the actual data before the adjacent gun interference separation as second input data of network learning, and taking the separated adjacent gun interference data as second label data of the network learning;
the data dividing module is used for converting the first input data, the second input data, the first tag data and the second tag data into a detector point gather according to a gather extraction method, and dividing the converted data into a test set and a training set according to a preset proportion;
The adjacent-gun interference separation module is used for initializing the parameters of a preset convolutional neural network to obtain an initial model of the convolutional neural network, inputting the training set into the initial model of the convolutional neural network for model training to obtain an adjacent-gun interference separation model, and carrying out adjacent-gun interference separation processing by utilizing the adjacent-gun interference separation model;
Wherein, the adjacent cannon interference separation module includes:
The model training unit is used for inputting the training set into an initial model of the convolutional neural network to perform model training to obtain first output layer data, comparing the first output layer data with label data in the training set, judging whether the prediction accuracy meets a preset condition, and taking the initial model trained by the model as an adjacent gun interference separation model if the prediction accuracy meets the preset condition;
the model test unit is used for inputting the test set into the adjacent gun interference separation model to perform model test to obtain second output layer data, and comparing the second output layer data with the label data in the test set to obtain a denoising performance test result of the adjacent gun interference separation model;
wherein the model training unit comprises:
An input layer data subunit, configured to input the training set as input layer data to an initial model of the convolutional neural network for model training;
An input data feature subunit, configured to perform convolution operation on the input layer data and n convolution kernels, input an operation result into a ReLU activation function, and obtain a basic feature of the input layer data by using the following formula, where n is a positive integer;
wherein M l-1 is the set of neuron nodes of layer 1, Data for the jth neuron node of the first layer,Is the data of the ith neuron node of the layer 1,To connect the convolution kernel weight matrix of the ith neuron node of the first-1 layer with the jth neuron node of the first layer,For the bias matrix of the jth neuron node of the first layer, the symbol "×" represents convolution operation;
Inputting the operation result into ReLu activation function, and realizing through ReLu activation function layer of convolutional neural network; wherein ReLu activates the function layer when When the data is the data of the ith neuron node of the first layer-1, the data of the corresponding ith neuron node of the first layerExpressed as:
An interference data feature subunit, configured to process the basic features of the input layer data by using multiple feature extraction units to obtain the essential features of the adjacent cannon interference data, where the feature extraction unit includes a convolution layer with n convolution kernels, a BN layer, and a ReLu activation function layer, the BN layer is configured to perform normalization processing on data, and for data x= { x 1,x2,…,xm }, m is a data length, and data y i＝BNγ,β(xi) is obtained after processing by the BN layer, where i=1, 2..m, γ, β is a learning parameter, and BN γ,β(xi) is implemented using the following formula:
Wherein epsilon is a non-zero minimum value, and the default value is 1 multiplied by 10 -6;
And the output layer data subunit is used for carrying out convolution operation on the essential characteristics of the adjacent gun interference data and n convolution kernels to obtain the first output layer data.
4. The convolutional neural network-based neighbor-based shot-to-interference separation device of claim 3, wherein the neighbor-to-shot-to-interference separation module further comprises:
The seismic data processing unit is used for converting the seismic data which are to be processed and contain adjacent gun interference into a wave detection point gather through a channel gather extraction method, inputting the data converted into the wave detection point gather into the adjacent gun interference separation model, and obtaining separated seismic data;
The seismic data application unit is used for converting the separated seismic data into a shot set through a shot set extraction method, and the seismic data converted into the shot set are used for seismic data processing and interpretation.
5. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any one of claims 1 to 2 when executing the program.
6. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program for executing the method of any one of claims 1 to 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010944979.3A CN114167500B (en) | 2020-09-10 | 2020-09-10 | Method and device for separating adjacent gun interference based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010944979.3A CN114167500B (en) | 2020-09-10 | 2020-09-10 | Method and device for separating adjacent gun interference based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114167500A CN114167500A (en) | 2022-03-11 |
| CN114167500B true CN114167500B (en) | 2024-06-28 |
Family
ID=80475542
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010944979.3A Active CN114167500B (en) | 2020-09-10 | 2020-09-10 | Method and device for separating adjacent gun interference based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114167500B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114839673B (en) * | 2022-07-01 | 2022-09-23 | 中国海洋大学 | Separation method, separation system and computer equipment for multi-seismic-source efficient acquisition wave field |
| 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 |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108107474A (en) * | 2018-02-02 | 2018-06-01 | 中国石油天然气集团有限公司 | A kind of aliased data separation method and device based on sparse inversion |
| CN110058305A (en) * | 2019-05-24 | 2019-07-26 | 吉林大学 | A kind of DAS seismic data noise-reduction method based on convolutional neural networks |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104597499B (en) * | 2013-10-31 | 2017-02-08 | 中国石油天然气集团公司 | Adjacent shot interference suppressing method and device for independent simultaneous sweeping seismic data of seismic vibrators |
| CN105527649B (en) * | 2015-11-30 | 2018-06-22 | 中国科学院地质与地球物理研究所 | A kind of more focus blended data separation methods of the highly effective gathering of multiple domain separating for several times |
| US11169293B2 (en) * | 2016-12-13 | 2021-11-09 | Cgg Services Sas | Device and method for model-based deblending |
| CN107942389A (en) * | 2017-11-16 | 2018-04-20 | 中国科学院地质与地球物理研究所 | For suppressing method, system and the computer-readable medium of adjacent big gun interference |
| CN110806602B (en) * | 2018-08-06 | 2022-04-22 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
| US11175424B2 (en) * | 2019-02-25 | 2021-11-16 | Saudi Arabian Oil Company | Seismic data de-blending |
| CN110568483A (en) * | 2019-07-22 | 2019-12-13 | 中国石油化工股份有限公司 | Automatic evaluation method for seismic linear noise suppression effect based on convolutional neural network |
| CN110568485B (en) * | 2019-09-06 | 2021-02-26 | 广州海洋地质调查局 | Neural network-based multi-channel seismic continuous recording and separating method |
| CN111580162A (en) * | 2020-05-21 | 2020-08-25 | 长江大学 | Seismic data random noise suppression method based on residual convolutional neural network |
-
2020
- 2020-09-10 CN CN202010944979.3A patent/CN114167500B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108107474A (en) * | 2018-02-02 | 2018-06-01 | 中国石油天然气集团有限公司 | A kind of aliased data separation method and device based on sparse inversion |
| CN110058305A (en) * | 2019-05-24 | 2019-07-26 | 吉林大学 | A kind of DAS seismic data noise-reduction method based on convolutional neural networks |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114167500A (en) | 2022-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20180061439A1 (en) | Automatic audio captioning | |
| CN114167500B (en) | Method and device for separating adjacent gun interference based on convolutional neural network | |
| CN109670575B (en) | Method and apparatus for simultaneously performing activation and convolution operations, and learning method and learning apparatus therefor | |
| CN108596267B (en) | Image reconstruction method, terminal equipment and computer readable storage medium | |
| US10325352B1 (en) | Method and device for transforming CNN layers to optimize CNN parameter quantization to be used for mobile devices or compact networks with high precision via hardware optimization | |
| US20200293875A1 (en) | Generative Adversarial Network Based Audio Restoration | |
| CN111105375B (en) | Image generation method, model training method and device thereof, and electronic equipment | |
| CN111667444B (en) | Image compressed sensing reconstruction method based on multi-channel residual error network | |
| KR20190041921A (en) | Method and device for performing activation and convolution operation at the same time, learning method and learning device for the same | |
| CN110619264B (en) | Method and device for identifying microseism effective signals based on UNet++, and method and device for identifying microseism effective signals based on UNet++ | |
| US20220130407A1 (en) | Method for isolating sound, electronic equipment, and storage medium | |
| Novoselov et al. | SEDENOSS: SEparating and DENOising Seismic Signals with dual‐path recurrent neural network architecture | |
| EP3959650A1 (en) | Abdominal multi-organ segmentation with organ-attention networks | |
| Huai et al. | Zerobn: Learning compact neural networks for latency-critical edge systems | |
| EP3686809A1 (en) | Method and device for transforming cnn layers to optimize cnn parameter quantization to be used for mobile devices or compact networks with high precision via hardware optimization | |
| CN117076858B (en) | Deep learning-based low-frequency geomagnetic strong interference suppression method and system | |
| CN116106880B (en) | Underwater sound source ranging method and device based on attention mechanism and multi-scale fusion | |
| CN112464725A (en) | First arrival picking method and device based on deep learning network | |
| CN112397086A (en) | Voice keyword detection method and device, terminal equipment and storage medium | |
| CN111724410A (en) | Target tracking method based on residual attention | |
| CN113722893B (en) | Seismic record inversion method, device, equipment and storage medium | |
| Wang et al. | A physics-augmented deep learning method for seismic data deblending | |
| CN112305588B (en) | Automatic first arrival wave pickup method and device | |
| CN113314148B (en) | Light-weight neural network generated voice identification method and system based on original waveform | |
| CN111856558B (en) | Automatic gain method for seismic record |
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 |