CN116011306A - Velocity modeling method and device based on plane source seismic data and electronic equipment - Google Patents
Velocity modeling method and device based on plane source seismic data and electronic equipment Download PDFInfo
- Publication number
- CN116011306A CN116011306A CN202111217474.8A CN202111217474A CN116011306A CN 116011306 A CN116011306 A CN 116011306A CN 202111217474 A CN202111217474 A CN 202111217474A CN 116011306 A CN116011306 A CN 116011306A
- Authority
- CN
- China
- Prior art keywords
- seismic
- seismic data
- velocity
- data
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 33
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 11
- 238000003062 neural network model Methods 0.000 claims abstract description 5
- 238000010606 normalization Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 8
- 238000011176 pooling Methods 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 8
- 238000013135 deep learning Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004587 chromatography analysis Methods 0.000 description 3
- 230000005012 migration Effects 0.000 description 3
- 238000013508 migration Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- RTZKZFJDLAIYFH-UHFFFAOYSA-N Diethyl ether Chemical compound CCOCC RTZKZFJDLAIYFH-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 239000011148 porous material Substances 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
Images
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The application discloses a velocity modeling method, a velocity modeling device and electronic equipment based on plane source seismic data, which comprise the steps of establishing a depth domain seismic velocity model and forward modeling the depth domain seismic velocity model to generate multi-shot seismic data; superposing different multi-shot seismic data of the unified model by utilizing the multi-shot seismic data to generate super-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data; normalizing the seismic velocity model by using the seismic velocity model corresponding to the super gun seismic data to obtain input tag data; creating a speed modeling system and determining training parameters of the speed modeling system by using the convolutional neural network model; and obtaining the seismic data, predicting the seismic data through a speed modeling system, and outputting a predicted depth domain speed model. Through a neural network model, the generated training data and label data are used as input data to be input into a network, the seismic data are extracted, and a speed modeling system is created, so that the method is applicable to the ultra-large data volume of the current seismic exploration.
Description
Technical Field
The invention relates to the technical field of oil and gas geophysical prospecting engineering, in particular to a speed modeling method, device and electronic equipment based on plane source seismic data.
Background
In the field of seismic exploration, velocity modeling is always in a key link of seismic data processing, velocity parameters of seismic waves influence accuracy of seismic exploration deviation imaging processing and subsequent interpretation results, specific positions generated by waveforms such as reflection and diffraction on a seismic record can be determined only by determining the velocity parameters, inclination angles and stratum shapes of the specific positions are determined, a target reservoir position is determined, properties such as saturation, permeability and the like of rock and pore fluid can be studied according to the inverted velocity parameters, and oil and gas content, predicted reserves and the like are determined. Inaccurate velocity modeling can lead to offset results and interpretation results deviating from actual underground conditions, ultimately leading to failure of the seismic survey project.
In exploration earth physics, existing seismic velocity modeling techniques mainly include conventional velocity analysis based on CMP gathers for the time domain, residual curvature analysis based on prestack time migration gathers, and depth domain focus analysis techniques. In recent years, due to the rising of a prestack depth migration technology, a depth domain speed modeling technology gradually becomes a conventional flow in seismic exploration processing, and the depth domain speed modeling technology is mainly divided into a first arrival travel time tomographic inversion technology, a reflection wave travel time tomographic technology based on CRP gather pickup and a full waveform inversion technology, wherein the first arrival tomographic inversion can utilize received first arrival travel time information to carry out smooth near-surface speed modeling on a long-wavelength background speed field in the ground; the reflected wave travel time chromatography technology is used for picking up residual curvature based on CRP gathers to finish middle-deep velocity modeling, but the existing migration velocity analysis method cannot meet the inversion requirement of a high-precision velocity structure, and although the existing seismic exploration technology uses first-arrival chromatography and reflected wave chromatography as a whole set of basic frames for depth domain modeling, the obtained modeling result still hardly meets the requirement of seismic exploration fine processing due to the inherent defects of ray theory; the full waveform inversion has theoretical highest speed modeling precision, however, in practical application, the full waveform inversion is easy to fall into local extremum, and because the full waveform inversion is seriously dependent on an initial speed model and low-frequency information in seismic data, the full waveform inversion cannot be popularized and applied on a large scale in the seismic exploration practice of land data, and in addition, the full waveform inversion uses full waveform information, has long application period, and cannot meet the current processing demands of numerous short time and heavy tasks. With the development of computer hardware and mathematical algorithms, big data processing and deep learning applications have also been developed in recent years. Considering the geometric index increase of the seismic exploration data volume and the advance of the two-wide-one-high acquisition technology, the conventional seismic exploration processing flow starts to look out of the way.
Therefore, it is expected to provide a velocity modeling method based on planar source seismic data, so as to solve the technical problem that the conventional velocity modeling technology is difficult to be applied to huge data amount nowadays.
The information disclosed in the background section of the invention is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.
Disclosure of Invention
The invention provides a speed modeling method, a device and electronic equipment based on plane source seismic data, which at least solve the technical problem that the traditional speed modeling technology is difficult to be applied to huge data quantity and the like.
In a first aspect, embodiments of the present disclosure provide a velocity modeling method based on planar source seismic data, including:
establishing a depth domain seismic velocity model;
forward modeling is carried out on the depth domain seismic velocity model to generate multi-shot seismic data;
superposing different multi-shot seismic data of a unified model by utilizing the multi-shot seismic data to generate super-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
normalizing the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
creating a speed modeling system and determining training parameters of the speed modeling system by using a convolutional neural network model;
and obtaining seismic data, predicting the seismic data through the velocity modeling system, and outputting a predicted depth domain velocity model.
As a specific implementation manner of the embodiment of the present disclosure, the specific method for establishing a depth domain seismic velocity model includes: establishing depth domain seismic velocity models of various complex geological structures in a pseudo-random mode; the pseudo-random calculation formula is as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
wherein V is a generated depth domain seismic velocity model, F is a velocity model generating function, and is determined according to a velocity value, an inclination angle and a velocity model layer number; val is a random stratum velocity value, dip is a random stratum dip value, and layer is a random stratum layer value.
As a specific implementation manner of the embodiment of the disclosure, a specific formula for generating super gun seismic data by superposing different multi-gun seismic data of a unified model is as follows:
wherein U is obs，is Observed seismic record obtained by forward modeling of earth is cannon, U superg ather Super gun seismic data of the plane wave source are generated;
the specific formula for obtaining the input training data by carrying out standardized processing on the super gun seismic data is as follows:
wherein U represents plane wave source super gun seismic data, U mean Representing the mean value of the plane wave source super gun seismic data record, U std The standard deviation of the plane wave source super gun seismic data record is represented, and epsilon is a small value.
As a specific implementation manner of the embodiment of the present disclosure, the specific method for forward modeling the depth domain seismic velocity model to generate multi-shot seismic data includes: and forward modeling is performed by using the depth domain seismic velocity model and adopting a time domain finite difference algorithm of a time second order space tenth order to generate multi-shot seismic data.
As a specific implementation manner of the embodiment of the present disclosure, the specific method for obtaining input tag data by normalizing the seismic velocity model by using the seismic velocity model corresponding to the super gun seismic data includes: carrying out maximum and minimum normalization processing on the seismic velocity model by using the seismic velocity model of the plane wave source super gun seismic data to obtain input tag data; the normalized calculation formula is as follows:
V＝(V-V min )/(V max -V min )
wherein V represents a seismic velocity model, V min ，V max Indicating the set normalized speed range. As a specific implementation manner of the embodiment of the present disclosure, the formula of the convolutional neural network model is as follows:
Conv＝(relu)(x·ω+b)ω，b∈random(kernal_size)
wherein Conv represents a calculation result of a convolution layer, x represents input of the convolution layer, ω represents weight of the convolution layer, b represents offset of the convolution layer, and kernel_size is the size of a convolution kernel adopted;
MaxPool＝max(m 1 ，m 2 ，…，m n )
wherein MaxPool represents the pooling result, m 1 ，m 2 ，…，m n Values of each characteristic diagram of the pooling unit;
(m 1 ，m 2 ，…，m n )＝F(x)
wherein, (m) 1 ，m 2 ，…，m n ) For the up-sampled feature map values, F is the nearest interpolation up-sampling function and x is the input feature to be up-sampled.
As a specific implementation manner of the embodiment of the present disclosure, the creating a speed modeling system and determining training parameters of the speed modeling system specifically includes:
setting network input parameters;
setting super gun seismic data;
inputting tag data;
the network outputs the data.
In a second aspect, embodiments of the present disclosure further provide a velocity modeling apparatus based on planar source seismic data, including:
the building module is used for building a depth domain seismic velocity model;
the generation module is used for forward modeling the depth domain seismic velocity model to generate multi-shot seismic data;
the standardized processing module is used for superposing different multi-shot seismic data of the unified model to generate super-shot seismic data by utilizing the multi-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
the normalization processing module is used for carrying out normalization processing on the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
the building module is used for building a speed modeling system by utilizing the convolutional neural network model and determining training parameters of the speed modeling system;
and the output module is used for acquiring seismic data, predicting the seismic data through the speed modeling system and outputting a predicted depth domain speed model.
As a specific implementation manner of an embodiment of the present disclosure, the establishing module includes: establishing depth domain seismic velocity models of various complex geological structures in a pseudo-random mode; the pseudo-random calculation formula is as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
wherein V is a generated depth domain seismic velocity model, F is a velocity model generating function, and is determined according to a velocity value, an inclination angle and a velocity model layer number; val is a random stratum velocity value, dip is a random stratum dip value, and layer is a random stratum layer value.
In a third aspect, embodiments of the present disclosure further provide an electronic device, including:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method of velocity modeling based on planar source seismic data as described above.
In a fourth aspect, the presently disclosed embodiments also provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform a method of velocity modeling based on planar source seismic data as described above.
The invention has the beneficial effects that:
according to the invention, through the neural network model, the generated training data and label data are used as input data to be input into the network, the multidimensional characteristic information of the seismic data is extracted, and the velocity modeling system is created, so that the method can be suitable for the establishment of the velocity modeling model with high efficiency and accuracy due to the ultra-large data volume of the current seismic exploration.
The method and apparatus of the present invention have other features and advantages which will be apparent from or are set forth in detail in the accompanying drawings and the following detailed description, which are incorporated herein, and which together serve to explain certain principles of the present invention.
Drawings
The foregoing and other objects, features and advantages of the invention will be apparent from the following more particular descriptions of exemplary embodiments of the invention as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts throughout the exemplary embodiments of the invention.
FIG. 1 is a flow chart of a velocity modeling method based on planar source seismic data according to embodiment 1 of the invention;
FIGS. 2a, 2b and 2c are examples of pseudo-randomly generated depth domain seismic velocity models employed by the present invention;
FIG. 3 is the single shot data generated in FIG. 2 a;
FIG. 4 is a schematic diagram of the present invention for plane wave source super-cannon seismic data generation using single cannon data;
FIGS. 5a and 5b are, respectively, training data plane wave source super gun seismic data and its standardized schematic diagrams for use in the present invention;
FIGS. 6a and 6b are schematic diagrams of tag data and normalization thereof, respectively, used in the present invention;
FIG. 7 is a schematic diagram of a deep neural network model for use with the present invention;
FIGS. 8a and 8b are graphs of loss convergence curves and velocity model residuals for the training process of the present invention;
FIGS. 9a and 9b are schematic diagrams of inversion results versus true velocity field using the present invention; FIG. 9c is a schematic illustration of a single lane comparison of velocity at different lateral positions;
FIG. 10 is a block diagram of a velocity modeling apparatus based on planar source seismic data according to embodiment 2 of the invention.
The invention is further described below with reference to the drawings and the detailed description.
Detailed Description
Preferred embodiments of the present invention will be described in more detail below. While the preferred embodiments of the present invention are described below, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein.
Example 1
FIG. 1 shows a flowchart of the steps of a method for velocity modeling based on planar source seismic data, according to one embodiment of the invention, as shown in FIG. 1, comprising:
s01: establishing a depth domain seismic velocity model;
s02: forward modeling is carried out on the depth domain seismic velocity model to generate multi-shot seismic data;
s03: superposing different multi-shot seismic data of a unified model by utilizing the multi-shot seismic data to generate super-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
s04: normalizing the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
s05: creating a speed modeling system and determining training parameters of the speed modeling system by using a convolutional neural network model;
s06: and obtaining seismic data, predicting the seismic data through the velocity modeling system, and outputting a predicted depth domain velocity model.
Note that S0N does not represent the order.
Referring to fig. 2a, 2b and 2c, a depth domain seismic velocity model is built.
Specifically, depth domain seismic velocity models of various complex geologic structures are established in a pseudo-random manner. In an implementation, a function is generated for the depth domain seismic velocity model as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
the range of the velocity, the dip angle and the layer number can be adjusted according to the requirement, the generation result of the depth domain seismic velocity model is shown in fig. 2a, 2b and 2c, the size of the depth domain seismic velocity model generated in the example is 63 transverse channels, the depth sampling point is 50, the channel spacing is 10 meters, and the depth sampling interval is 10 meters.
Referring to fig. 3, the depth domain seismic velocity model performs forward modeling to generate multi-shot seismic data.
Specifically, a large number of depth domain seismic velocity models generated in the step S01 are utilized, and a time domain finite difference algorithm of a time second order space tenth order is adopted for forward modeling to generate multi-shot seismic data. In the implementation process, the depth domain seismic velocity model generated in the step S01 is used for forward modeling to generate multi-shot seismic data as shown in fig. 2a, and the result shown in fig. 3 is obtained. The total forward performance of the seismic records is 32 cannons, the cannon interval is 20 meters, the detection points are 63, the detection points are 10 meters apart, the sampling points of each cannon are 1001, and the sampling interval is 0.5ms.
Referring to fig. 4, 5a and 5b, by using the multi-shot seismic data, super-shot seismic data is generated by superposing different multi-shot seismic data of a unified model, and standardized processing is performed on the super-shot seismic data to obtain input training data.
Specifically, the multi-shot seismic data generated in the S02 are utilized to superimpose different multi-shot data of a unified model to generate plane wave source super-shot seismic data, and the plane wave source super-shot seismic data is subjected to standardized processing and used as input training data of a deep learning speed modeling system. And S02, the forward-modeling is carried out to obtain a large amount of multi-shot seismic data which are single-source observation data, and the plane wave source super-multi-shot seismic data are overlapped according to a plane wave source super-shot seismic data overlapping formula to obtain plane wave source super-shot seismic data as shown in figure 4. In the implementation process, the obtained plane wave source super gun seismic data are standardized line by line, and input training data shown in fig. 5b are obtained.
Referring to fig. 6a and 6b, the seismic velocity model corresponding to the super gun seismic data is utilized to perform normalization processing on the seismic velocity model to obtain input tag data.
Specifically, the tag data is selected as a seismic velocity model corresponding to the S03 plane wave source super gun seismic data, and the maximum and minimum normalization is carried out on the seismic velocity model to serve as input tag data of the deep learning velocity modeling system. In the implementation process, the maximum 6500 and minimum 1500 normalization operations are performed on the seismic velocity model, so as to obtain input tag data as shown in fig. 6 b. The normalized calculation formula is as follows:
V＝(V-V min )/(V max -V min )
wherein V represents a seismic velocity model, V min ，V max Indicating the set normalized speed range.
Referring to fig. 7, 8a and 8b, a velocity modeling system is created and training parameters of the velocity modeling system are determined using a convolutional neural network model.
Specifically, a deep learning speed modeling system is built by using a convolutional neural network model: the method comprises 11 layers of convolution layers, 5 layers of pooling layers, 5 layers of up-sampling layers and 10 layers of BN layers: the BN layer performs batch standardization operation on the input feature images according to the standardization mode in S03; the built deep learning speed modeling system is shown in fig. 7; the formula of the convolutional neural network model is as follows:
Conv＝(relu)(x·ω+b)ω，b∈random(kernal_size)
wherein Conv represents a convolution layer calculation result, x represents a convolution layer input, ω represents a convolution layer weight, b represents a convolution layer offset, and kernal_size is the adopted convolution kernel size.
MaxPool＝max(m 1 ，m 2 ，…，m n )
Wherein MaxPool represents the pooling result, m 1 ，m 2 ，…，m n Values of each characteristic diagram of the pooling unit are obtained.
(m 1 ，m 2 ，…，m n )＝F(x)
Wherein, (m) 1 ，m 2 ，…，m n ) Is thatThe up-sampled feature map values, F is the nearest interpolation up-sampling function, and x is the input feature to be up-sampled.
And determining training parameters of the deep learning speed modeling system, and performing network training until the network convergence is optimal. In the implementation process, the specific process of determining the training parameters of the speed modeling system comprises the following steps: setting network input parameters: the input format is [ Batchsize, height, width, channel ], set input plane wave source super gun seismic data shape= [ Batchsize,63, 1001,1], input tag data shape= [ Batchsize,63, 50,1], network output data shape= [ Batchsize,63, 50,1]. Where, batchsize represents the size specified for the small lot training. The optimizer selected in the network training process is an Adam optimizer, and the training function is RMSE (root mean square error). The training frequency of the deep neural network model is set to be 500, and the batch size used for training is set to be 8. The network training process is shown in fig. 8a and 8b, where 8a is a loss convergence curve and 8b is a velocity model convergence curve obtained by training.
9a, 9b and 9c, seismic data are acquired, predicted by the velocity modeling system, and a predicted depth domain velocity model is output.
Specifically, the created velocity modeling system is utilized to predict the seismic data, a predicted depth domain velocity model is output, the multi-shot seismic data obtained through forward modeling in fig. 9a is combined into plane wave source super-shot seismic data, the plane wave source super-shot seismic data is input after standardization, the super-shot seismic data is obtained, the maximum velocity value 6500 is used, and the minimum velocity value 1500 is subjected to inverse normalization to obtain a final prediction result, wherein the final prediction result is shown in fig. 9 b. The effectiveness of the present invention can be seen from a comparison of the single pass velocity profile of 9 c. From the prediction time, the result obtained by predicting by using the trained network is almost instantaneous, and the efficiency is far higher than that of the traditional depth domain speed modeling method.
Example 2
FIG. 10 illustrates a velocity modeling apparatus based on planar source seismic data according to one embodiment of the invention.
As shown in fig. 10, the velocity modeling apparatus based on planar source seismic data includes:
the building module is used for building a depth domain seismic velocity model;
the generation module is used for forward modeling the depth domain seismic velocity model to generate multi-shot seismic data;
the standardized processing module is used for superposing different multi-shot seismic data of the unified model to generate super-shot seismic data by utilizing the multi-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
the normalization processing module is used for carrying out normalization processing on the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
the building module is used for building a speed modeling system by utilizing the convolutional neural network model and determining training parameters of the speed modeling system;
and the output module is used for acquiring seismic data, predicting the seismic data through the speed modeling system and outputting a predicted depth domain speed model.
In one example, the setup module includes: establishing depth domain seismic velocity models of various complex geological structures in a pseudo-random mode; the pseudo-random calculation formula is as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
wherein V is a generated depth domain seismic velocity model, F is a velocity model generating function, and is determined according to a velocity value, an inclination angle and a velocity model layer number; val is a random stratum velocity value, dip is a random stratum dip value, and layer is a random stratum layer value.
Example 3
The present disclosure provides an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the planar source seismic data based velocity modeling method described above.
An electronic device according to an embodiment of the present disclosure includes a memory and a processor.
The memory is for storing non-transitory computer readable instructions. In particular, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device to perform the desired functions. In one embodiment of the present disclosure, the processor is configured to execute the computer readable instructions stored in the memory.
Example 4
Embodiments of the present disclosure provide a non-transitory computer readable storage medium storing computer instructions for causing a computer to perform a method of velocity modeling based on planar source seismic data as described above.
The non-transitory computer readable storage medium according to embodiments of the present disclosure, when executed by a processor, performs all or part of the steps of the methods of embodiments of the present disclosure described above.
The computer-readable storage medium described above includes, but is not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or removable hard disk), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).
It will be appreciated by persons skilled in the art that the above description of embodiments of the invention has been given for the purpose of illustrating the benefits of embodiments of the invention only and is not intended to limit embodiments of the invention to any examples given.
The foregoing description of embodiments of the invention has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described.
Claims (10)
1. A method of velocity modeling based on planar source seismic data, comprising:
establishing a depth domain seismic velocity model;
forward modeling is carried out on the depth domain seismic velocity model to generate multi-shot seismic data;
superposing different multi-shot seismic data of a unified model by utilizing the multi-shot seismic data to generate super-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
normalizing the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
creating a speed modeling system and determining training parameters of the speed modeling system by using a convolutional neural network model;
and obtaining seismic data, predicting the seismic data through the velocity modeling system, and outputting a predicted depth domain velocity model.
2. The method for modeling velocity based on planar source seismic data according to claim 1, wherein the specific method for establishing depth domain seismic velocity model comprises the following steps: establishing depth domain seismic velocity models of various complex geological structures in a pseudo-random mode; the pseudo-random calculation formula is as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
wherein V is a generated depth domain seismic velocity model, F is a velocity model generating function, and is determined according to a velocity value, an inclination angle and a velocity model layer number; val is a random stratum velocity value, dip is a random stratum dip value, and layer is a random stratum layer value.
3. The method for velocity modeling based on planar source seismic data according to claim 1, wherein the specific formula for generating super gun seismic data by superposing different multi gun seismic data of a unified model is as follows:
wherein U is obs，is Observed seismic record obtained by forward modeling of earth is cannon, U supergather Super gun seismic data of the plane wave source are generated;
the specific formula for obtaining the input training data by carrying out standardized processing on the super gun seismic data is as follows:
wherein U represents plane wave source super gun seismic data, U mean Representing the mean value of the plane wave source super gun seismic data record, U std The standard deviation of the plane wave source super gun seismic data record is represented, and epsilon is a small value.
4. The method for modeling velocity based on planar source seismic data according to claim 1, wherein the depth domain seismic velocity model performs forward modeling to generate multi-shot seismic data specifically: and forward modeling is performed by using the depth domain seismic velocity model and adopting a time domain finite difference algorithm of a time second order space tenth order to generate multi-shot seismic data.
5. The method for modeling velocity based on planar source seismic data according to claim 1, wherein the specific method for obtaining input tag data by normalizing the seismic velocity model by using the seismic velocity model corresponding to the super gun seismic data is as follows: carrying out maximum and minimum normalization processing on the seismic velocity model by using the seismic velocity model of the plane wave source super gun seismic data to obtain input tag data; the normalized calculation formula is as follows:
V＝(V-V min )/(V max -V min )
wherein V represents a seismic velocity model, V min ，V max Indicating the set normalized speed range.
6. The method of claim 1, wherein the convolution neural network model is formulated as follows:
Conv＝(relu)(x·ω+b)ω，b∈random(kernal_size)
wherein Conv represents a calculation result of a convolution layer, x represents input of the convolution layer, ω represents weight of the convolution layer, b represents offset of the convolution layer, and kernel_size is the size of a convolution kernel adopted;
MaxPool＝max(m 1 ，m 2 ，…，m n )
wherein MaxPool represents the pooling result, m 1 ，m 2 ，…，m n Values of each characteristic diagram of the pooling unit;
(m 1 ，m 2 ，…，m n )＝F(x)
wherein, (m) 1 ，m 2 ，…，m n ) For the up-sampled feature map values, F is the nearest interpolation up-sampling function and x is the input feature to be up-sampled.
7. The method of claim 1, wherein the creating a velocity modeling system and determining training parameters of the velocity modeling system specifically comprises:
setting network input parameters;
setting super gun seismic data;
inputting tag data;
the network outputs the data.
8. A velocity modeling apparatus based on planar source seismic data, comprising:
the building module is used for building a depth domain seismic velocity model;
the generation module is used for forward modeling the depth domain seismic velocity model to generate multi-shot seismic data;
the standardized processing module is used for superposing different multi-shot seismic data of the unified model to generate super-shot seismic data by utilizing the multi-shot seismic data, and carrying out standardized processing on the super-shot seismic data to obtain input training data;
the normalization processing module is used for carrying out normalization processing on the seismic velocity model by utilizing the seismic velocity model corresponding to the super gun seismic data to obtain input tag data;
the building module is used for building a speed modeling system by utilizing the convolutional neural network model and determining training parameters of the speed modeling system;
and the output module is used for acquiring seismic data, predicting the seismic data through the speed modeling system and outputting a predicted depth domain speed model.
9. The planar source seismic data-based velocity modeling apparatus of claim 8, wherein the building module comprises: establishing depth domain seismic velocity models of various complex geological structures in a pseudo-random mode; the pseudo-random calculation formula is as follows:
V＝F(val，dip，layer)val∈[1000，6000]，dip∈[-25°，25°]，layers∈[2，12]
wherein V is a generated depth domain seismic velocity model, F is a velocity model generating function, and is determined according to a velocity value, an inclination angle and a velocity model layer number; val is a random stratum velocity value, dip is a random stratum dip value, and layer is a random stratum layer value.
10. An electronic device, the electronic device comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of velocity modeling based on planar source seismic data of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111217474.8A CN116011306A (en) | 2021-10-19 | 2021-10-19 | Velocity modeling method and device based on plane source seismic data and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111217474.8A CN116011306A (en) | 2021-10-19 | 2021-10-19 | Velocity modeling method and device based on plane source seismic data and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116011306A true CN116011306A (en) | 2023-04-25 |
Family
ID=86017865
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111217474.8A Pending CN116011306A (en) | 2021-10-19 | 2021-10-19 | Velocity modeling method and device based on plane source seismic data and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116011306A (en) |
-
2021
- 2021-10-19 CN CN202111217474.8A patent/CN116011306A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106405651B (en) | Full waveform inversion initial velocity model construction method based on logging matching | |
| CN107831542B (en) | DDW high accuracy depth domain well shakes matching process | |
| CN102695970B (en) | An improved process for characterising the evolution of an oil or gas reservoir over time | |
| US11733413B2 (en) | Method and system for super resolution least-squares reverse time migration | |
| CN111123359B (en) | Logging while drilling and stratum grid constrained well periphery seismic imaging detection method and device | |
| US11668848B2 (en) | Method and system for seismic imaging using S-wave velocity models and machine learning | |
| US12032111B2 (en) | Method and system for faster seismic imaging using machine learning | |
| US9952341B2 (en) | Systems and methods for aligning a monitor seismic survey with a baseline seismic survey | |
| US11397273B2 (en) | Full waveform inversion in the midpoint-offset domain | |
| CN109073772B (en) | Determining displacement between seismic images using optical flow | |
| CN109154674A (en) | The displacement between seismic image is determined using light stream | |
| CN108508481B (en) | A kind of method, apparatus and system of longitudinal wave converted wave seismic data time match | |
| US11971513B2 (en) | System and method for forming a seismic velocity model and imaging a subterranean region | |
| CN106574980A (en) | System and method for rock property estimation of subsurface geologic volumes | |
| EP4133310A1 (en) | System and method for seismic inversion | |
| US12000971B2 (en) | Method and system for seismic processing using virtual trace bins based on offset attributes and azimuthal attributes | |
| CN107255832B (en) | A kind of inversion method of subsurface structure | |
| CN115170428A (en) | Noise reduction method for acoustic wave remote detection imaging graph | |
| CN112305615B (en) | Seismic data angle domain common imaging point gather extraction method and system | |
| CN116011306A (en) | Velocity modeling method and device based on plane source seismic data and electronic equipment | |
| CN110568488A (en) | Biological reef reservoir identification method based on nonlinear chaotic algorithm | |
| CN111596346B (en) | Elastic wave speed inversion method and device | |
| CN113219534B (en) | Pre-stack depth migration speed quality control method and device, medium and electronic equipment | |
| US20240069228A1 (en) | Learning hydrocarbon distribution from seismic image | |
| US20240272326A1 (en) | Identification and quantification of subsurface geobodies |
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 |