Sample labeling method and device using forward modeling and earthquake large model theory

Abstract

The embodiment of the invention provides a sample labeling method utilizing forward modeling. The method comprises the steps that an initial reflection coefficient model with the same volume as a preset reservoir needing to be simulated is constructed, wherein the initial reflection coefficient model comprises a plurality of data points; adding random fluctuating wrinkles to the reflection coefficient model; obtaining the offset of each data point in the reflection coefficient model in three directions of the reflection coefficient model according to a preset principle of seismic tectonic motion of a reservoir needing to be simulated; performing deformation and / or distortion transformation on the reflection coefficient model according to the offset of each data point in the reflection coefficient model in three directions, recording the position information of the changed data points in the reflection coefficient model in the stratum, and generating a mark corresponding to each data point according to the position information; and carrying out noise addition processing on the reflection coefficient model, converting the reflection coefficient model after noise addition into a seismic data volume, and obtaining the seismic data volume with position information marks.

Description

Technical Field

[0001] This invention relates to the field of oil and gas exploration and development, and in particular to a sample annotation method and apparatus utilizing forward modeling and large seismic model theory. Background Technology

[0002] Artificial intelligence sample set annotation is a crucial step in the training process of machine learning algorithms, and it is essential for improving model performance and accuracy. In the oil and gas industry, data annotation is developing towards specialization and segmentation, and the annotation of datasets faces challenges such as high specialization, large data volume, and high dimensionality.

[0003] However, with the advent of the era of large models, the sample labeling methods and technologies of general artificial intelligence can no longer meet the needs of intelligent development in the oil and gas industry. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed to provide a sample annotation method and apparatus that utilizes forward modeling and large-scale seismic model theory to overcome or at least partially solve the above problems.

[0005] In a first aspect, embodiments of the present invention provide a sample annotation method using forward modeling, comprising:

[0006] Construct an initial reflection coefficient model that is identical to the preset reservoir volume to be simulated; the reflection coefficient model contains several data points;

[0007] Add random undulating wrinkles to the reflection coefficient model;

[0008] Based on the principle of the pre-set reservoir seismic tectonic movement to be simulated, the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model is obtained.

[0009] Based on the offset of each data point in the reflection coefficient model in three directions, the reflection coefficient model is deformed and / or twisted, and the position information of the changed data points in the formation is recorded.

[0010] Based on the location information of the changed data points in the strata, a label corresponding to each data point is generated;

[0011] The reflection coefficient model is subjected to noise processing to obtain a noisy reflection coefficient model;

[0012] The noisy reflection coefficient model is converted into a seismic data volume, resulting in a seismic data volume with location information annotations.

[0013] In one embodiment, constructing an initial reflection coefficient model that is identical to a preset reservoir volume to be simulated includes:

[0014] Along the depth direction of the initial reflection coefficient model, uniformly distributed data points are set, and random numbers within the range of [0, 1] are used to assign values ​​to the uniformly distributed data points to generate a one-dimensional random sequence;

[0015] The one-dimensional random sequence is copied along line and channel directions that are perpendicular to the depth direction and mutually perpendicular to each other in the horizontal direction to obtain an initial reflection coefficient model that is the same as the preset reservoir volume to be simulated.

[0016] In one embodiment, adding random undulations and wrinkles to the reflection coefficient model includes:

[0017] The offset of each data point in the reflection coefficient model in the depth direction is randomly generated using Berlin noise;

[0018] The data points in the reflection coefficient model are mapped to new positions in the depth direction according to the offset.

[0019] In one embodiment, adding noise to the reflection coefficient model to obtain a noisy reflection coefficient model includes:

[0020] Construct a random Gaussian noise volume with the same volume as the reflection coefficient model;

[0021] The random Gaussian noise is subjected to low-pass filtering based on the preset frequency band and signal-to-noise ratio.

[0022] The low-pass filtered random Gaussian noise volume is superimposed on the reflection coefficient model to obtain the noisy reflection coefficient model.

[0023] In one embodiment, the seismic data volume is a seismic amplitude data volume;

[0024] Correspondingly, the noisy reflection coefficient model is converted into a seismic data volume, including:

[0025] Randomly sample frequencies within a preset frequency band;

[0026] And generate the Ricker wavelet corresponding to the frequency of the sampled data;

[0027] The earthquake amplitude data volume is obtained by convolving the Rick wavelet with the noisy reflection coefficient model along the depth direction.

[0028] Secondly, embodiments of the present invention provide a method for labeling theoretical samples of a large earthquake model, including:

[0029] Collect the preset location information of the reservoir to be simulated, the seismic calibration of the actual drilled well, the seismic attributes of oil and gas, and the stratigraphic parameters of the geological sedimentary deposits of the oil and gas reservoir;

[0030] Based on the wellbore seismic calibration of the actual drilled well, expert interpretation samples are labeled to obtain a seismic data volume with location information annotation; based on the oil and gas seismic attributes and the wellbore seismic calibration of the actual drilled well, mechanistic model samples are labeled to obtain a seismic data volume with location information annotation; based on the stratigraphic parameters of the oil and gas reservoir geological sedimentation and the location information, physical model annotation is performed to obtain a seismic data volume with location information annotation; based on the preset principle of seismic tectonic movement of the reservoir to be simulated, mathematical simulation samples are labeled using the forward modeling sample annotation method to obtain a seismic data volume with location information annotation.

[0031] In one embodiment, the step of labeling mechanistic model samples based on the oil and gas seismic attributes and actual well seismic calibration includes:

[0032] Based on the mechanism model formula, the seismic properties of the oil and gas are calculated to obtain the sample sensitive attribute volume;

[0033] Based on the actual drilling well vibration calibration, the reservoir threshold value of the sample sensitive attribute body is statistically analyzed.

[0034] The reservoir location is obtained based on the reservoir threshold value and the sensitive attributes of the sample sensitive attribute body;

[0035] The sensitive attribute volume is labeled with the reservoir location to obtain a seismic data volume with location information labeling.

[0036] In one embodiment, the physical model annotation based on the stratigraphic parameters of the oil and gas reservoir's geological sedimentation and the location information includes:

[0037] Based on the stratigraphic parameters of the oil and gas reservoir's geological sedimentation, design a physical model capable of inverting the preset simulated reservoir;

[0038] Determine the parameters of the physical model;

[0039] Construct the rock skeleton and fluid channels of the physical model;

[0040] Simulated data collection;

[0041] Based on the location information, the simulated data is labeled to obtain a seismic data volume with location information labels.

[0042] Thirdly, embodiments of the present invention provide a sample annotation acquisition device utilizing forward modeling, comprising:

[0043] An initial module is used to construct an initial reflection coefficient model that is identical to the preset reservoir volume to be simulated; the reflection coefficient model contains several data points;

[0044] Add a wrinkle module to add random undulating wrinkles to the reflection coefficient model;

[0045] The offset acquisition module is used to obtain the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model according to the preset principle of the seismic tectonic movement of the reservoir to be simulated.

[0046] The transformation module is used to deform and / or distort the reflection coefficient model according to the offset of each data point in the three directions, and record the position information of the transformed data points in the stratum.

[0047] The annotation generation module is used to generate annotations for each data point based on the location information of the changed data points in the strata.

[0048] A noise-adding module is used to add noise to the reflection coefficient model to obtain a noise-added reflection coefficient model.

[0049] The conversion module is used to convert the noise-added reflection coefficient model into a seismic data volume, resulting in a seismic data volume with location information annotations.

[0050] Fourthly, embodiments of the present invention provide a computing device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the program executed by the processor implements the sample annotation acquisition method using forward modeling.

[0051] Fifthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the sample annotation acquisition method using forward modeling.

[0052] In a sixth aspect, embodiments of the present invention provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the sample annotation acquisition method using forward modeling.

[0053] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:

[0054] This invention provides a sample annotation method using forward modeling. It automatically constructs a reflection coefficient model identical to the preset reservoir volume to be simulated; automatically adds random undulations, folds, deformations, and / or distortions to the reflection coefficient model; and automatically generates annotations based on the transformed reflection coefficient model to obtain seismic data volumes with location information annotations. This invention uses forward modeling to automatically simulate a large number of seismic data volumes with location information annotations, providing a large number of samples for large-scale seismic models, meeting the data volume requirements of large-scale seismic models; it also meets the intelligent development needs of the oil and gas industry and reduces the workload of geological exploration scientists. It has broad application prospects in the field of oil and gas seismic prediction and provides strong support for the sustainable development of the oil and gas industry.

[0055] This invention also provides a method for labeling theoretical samples for large-scale seismic models. This method utilizes expert interpretation sample labeling, mechanistic model sample labeling, physical model sample labeling, and forward simulation sample labeling. These three methods can automatically simulate a large number of samples required for large-scale seismic models, covering the labeling needs of downstream seismic tasks and meeting the sample and feature supply requirements for fine-tuning scenarios of large-scale seismic models. This reduces the workload of geological prospectors, allowing them to focus on higher-value-added scientific research. It can achieve economies of scale in the exploration and development of various oil and gas reservoirs and has broad application prospects in the field of oil and gas seismic prediction, providing strong support for the sustainable development of the oil and gas industry.

[0056] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0057] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0058] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0059] Figure 1 A flowchart of a sample annotation method using forward modeling provided in an embodiment of the present invention;

[0060] Figure 2 A flowchart of the earthquake large model theoretical sample annotation method provided in the embodiments of the present invention;

[0061] Figure 3 A cross-sectional view of the sample annotation method using forward modeling provided in the embodiments of the present invention; from left to right: random stratigraphic reflection coefficient profile (generated after step S11), stratigraphic reflection coefficient profile with added fractured-vuggy reservoir reflection coefficient (generated after step S11), seismic profile generated after wavelet convolution (generated after step S16), and final seismic profile with added tectonic trend (generated after step S17).

[0062] Figure 4 The seismic data volumes with location information annotations are obtained by four annotation methods of the earthquake large model theoretical sample annotation method provided in the embodiments of the present invention.

[0063] Figure 5 A cross-sectional view of a crack / cavity predicted using the large earthquake model theory sample annotation method provided in an embodiment of the present invention;

[0064] Figure 6 A cross-sectional view of a fault predicted by the earthquake large model theoretical sample annotation method provided in an embodiment of the present invention;

[0065] Figure 7 This is a structural block diagram of a sample annotation acquisition device using forward modeling provided in an embodiment of the present invention. Detailed Implementation

[0066] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0067] To address the aforementioned technical issues and enable the annotation of sample datasets in the application scenario of fine-tuning large-scale earthquake models, this invention provides a sample annotation method utilizing forward modeling, the flowchart of which is shown below. Figure 1 As shown, it includes:

[0068] S11. Construct an initial reflection coefficient model with the same preset reservoir volume as the reservoir to be simulated; the reflection coefficient model contains several data points;

[0069] S12. Add random undulating wrinkles to the reflection coefficient model;

[0070] S13. Based on the pre-set principle of the seismic tectonic movement of the reservoir to be simulated, obtain the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model.

[0071] S14. Based on the offset of each data point in the reflection coefficient model in three directions, perform deformation and / or distortion transformation on the reflection coefficient model, and record the position information of the changed data points in the formation.

[0072] S15. Based on the changed location information of the data points in the strata, generate a label corresponding to each data point;

[0073] S16. Add noise to the reflection coefficient model to obtain the noisy reflection coefficient model;

[0074] S17. The noise-added reflection coefficient model is converted into a seismic data volume to obtain a seismic data volume with location information annotations.

[0075] This invention provides a sample annotation method using forward modeling. It involves randomly constructing a reflection coefficient model with the same volume as a preset reservoir to be simulated; randomly adding random undulations, folds, deformations, and / or distortions to the reflection coefficient model; and automatically generating annotations based on the randomly transformed reflection coefficient model to obtain seismic data volumes with location information annotations. This invention randomly generates seismic data volumes with location information annotations. For example, a computer program can automatically simulate a large number of seismic data volumes with location information annotations, randomly simulating preset reservoirs to be simulated, thereby meeting the sample size requirements of large-scale seismic models.

[0076] This invention utilizes forward modeling, which can automatically simulate large volumes of seismic data with location information annotations. This provides a large number of samples for large-scale seismic models, meeting their data volume requirements and fulfilling the intelligent development needs of the oil and gas industry. It has broad application prospects in the field of oil and gas seismic prediction, providing strong support for the sustainable development of the oil and gas industry.

[0077] In one embodiment, constructing the reflection coefficient model in step S11 may include, for example, the following steps:

[0078] Along the depth direction of the initial reflection coefficient model, uniformly distributed data points are set, and random numbers within the range of [0, 1] are used to assign values ​​to the uniformly distributed data points to generate a one-dimensional random sequence;

[0079] The one-dimensional random sequence is copied along line and channel directions perpendicular to the depth direction and mutually perpendicular to each other in the horizontal direction to obtain an initial reflection coefficient model with the same preset reservoir volume to be simulated. At this time, the reflection coefficient model has the same data point values ​​in the horizontal direction, and the reflection coefficient model itself has a stacking characteristic. The cross-sectional view of the completed reflection coefficient model can be, for example, as shown in the figure. Figure 3 The leftmost image is shown.

[0080] Since the aforementioned data is only a standard model and lacks the undulations and folds characteristic of real geological environments, random undulations and folds can be added to the reflection coefficient model in step S12 to simulate real geological environments. The specific implementation of step S12 is as follows:

[0081] The offset in the depth direction of each data point in the reflection coefficient model is randomly generated using Berlin noise. A positive offset indicates that the data point is a simulated ground uplift, while a negative offset indicates that the data point is a simulated ground collapse. The aforementioned Berlin noise is a noise function based on fractal structure, which can simulate natural noise to a certain extent. Fractals are geometric shapes containing detailed structures at arbitrarily small scales, and usually have a fractal dimension that strictly exceeds the topological dimension.

[0082] The data points in the reflection coefficient model are mapped to new positions in the depth direction according to the offset.

[0083] In the aforementioned step S13, seismic tectonics refers to the reflection on the seismic profile of the morphology (including folds, faults, etc.) formed by the deformation or displacement of rock strata or rock masses under the action of internal and external stresses of the Earth.

[0084] In one embodiment, to eliminate components outside the simulated seismic frequency band, the reflection coefficient model is noise-added in step S16, which can be implemented as follows:

[0085] Construct a random Gaussian noise volume with the same volume as the reflection coefficient model; the aforementioned same volume means that the random Gaussian noise volume has the same number of data points as the reflection coefficient model in the depth direction, line number direction, and track number direction.

[0086] Based on the preset frequency band and signal-to-noise ratio, the random Gaussian noise volume is subjected to low-pass filtering;

[0087] The low-pass filtered random Gaussian noise volume is superimposed on the reflection coefficient model to obtain the noisy reflection coefficient model.

[0088] In this embodiment of the invention, in the scenario of fine-tuning the large earthquake model, it is also necessary to process the aforementioned noisy reflection coefficient model into the form of a sample dataset of the large earthquake model. Therefore, the noisy reflection coefficient model is converted into an earthquake data volume through step S17.

[0089] Furthermore, the aforementioned earthquake data volume can be, for example, an earthquake amplitude data volume;

[0090] Accordingly, step S17 can be performed, for example, in the following manner:

[0091] Randomly sample frequencies within a preset frequency band;

[0092] It generates the Ricker wavelet corresponding to the sampling frequency; the Ricker wavelet is a seismic wavelet with only one peak, short duration, fast convergence, and 0 phase; the seismic wavelet is a signal with a definite start time, finite energy, and a certain duration, and is a basic element of seismic records.

[0093] The earthquake amplitude data volume is obtained by convolving the Ricker wavelet with the noisy reflection coefficient model along the depth direction.

[0094] Based on the same inventive concept, this invention also provides a method for labeling theoretical samples of a large-scale earthquake model, the flowchart of which is shown below. Figure 2 As shown, it includes:

[0095] S21. Collect the preset location information of the reservoir to be simulated, the seismic calibration of the actual drilled well, the seismic attributes of oil and gas, and the stratigraphic parameters of the geological sedimentation of the oil and gas reservoir.

[0096] S22. Based on the wellbore seismic calibration of actual drilled wells, perform expert interpretation sample annotation to obtain seismic data volumes with location information annotations; based on the seismic attributes of oil and gas and the wellbore seismic calibration of actual drilled wells, perform mechanism model sample annotation to obtain seismic data volumes with location information annotations; based on the stratigraphic parameters and location information of oil and gas reservoir geology and sedimentation, perform physical model annotation to obtain seismic data volumes with location information annotations; based on the pre-set principle of seismic tectonic movement of the reservoir to be simulated, use the forward modeling sample annotation method to perform mathematical simulation sample annotation to obtain seismic data volumes with location information annotations.

[0097] Reference Figure 5 As shown, Figure 5 This is a seismic profile when the reservoir to be simulated is a fractured-vuggy reservoir, where fractures and vuggy reservoirs are shown in red and the rest in blue. Figure 6 This is a seismic profile when the reservoir to be simulated is a fracture.

[0098] This invention also provides a method for labeling theoretical samples for large-scale seismic models. This method utilizes expert interpretation sample labeling, mechanistic model sample labeling, physical model sample labeling, and forward simulation sample labeling. These three methods can automatically simulate a large number of samples required for large-scale seismic models, covering the labeling needs of downstream seismic tasks and meeting the sample and feature supply requirements for fine-tuning scenarios of large-scale seismic models. This reduces the workload of geological prospectors, allowing them to focus on higher-value-added research and development. It can achieve economies of scale in the exploration and development of various oil and gas reservoirs and has broad application prospects in the field of oil and gas seismic prediction, providing strong support for the sustainable development of the oil and gas industry.

[0099] In step S22 above, the mechanism model samples are labeled based on the seismic properties of oil and gas and the seismic calibration of actual drilled wells. This can be achieved, for example, through the following steps, or by any means in the prior art. The embodiments of the present invention do not limit this:

[0100] Based on the mechanism model formula, the seismic properties of oil and gas are calculated to obtain the sample sensitive attribute volume;

[0101] Based on the vibration calibration of actual drilled wells, the reservoir threshold values ​​of the sensitive attribute bodies in the statistical samples were determined.

[0102] The reservoir location is obtained based on the reservoir threshold value and the sensitive attributes of the sample sensitive attribute body;

[0103] Using reservoir location to label sensitive attribute volumes, we obtain seismic data volumes with location information labels;

[0104] The seismic data volume with location information annotation obtained by using mechanistic model sample annotation can be, for example, used as... Figure 4 The seismic data volume below the mechanistic model annotations is shown. Since there are at least one hundred formulas for seismic attributes in the mechanistic model, therefore... Figure 4 The image shows a volume of seismic data with location information annotations, obtained by calculating seismic attributes using four different mechanistic model formulas. Figure 4 The red lines represent randomly generated faults.

[0105] In one embodiment, physical model annotation is performed based on stratigraphic parameters and location information of oil and gas reservoir geological sedimentation. This can be achieved, for example, through the following steps, or by any means in the prior art. The embodiments of the present invention do not limit this:

[0106] Based on the stratigraphic parameters of the oil and gas reservoir's geological sedimentation, design a physical model that can invert the pre-simulated reservoir.

[0107] Determine the parameters of the physical model;

[0108] Constructing the rock skeleton and fluid channels of the physical model;

[0109] Simulated data collection;

[0110] Based on the location information, the simulated data is labeled to obtain seismic data volumes with location information annotations. For example, the obtained seismic data volumes with location information annotations can be like... Figure 4 The seismic data volume below the annotation of the physical model is shown.

[0111] Since the embodiments of the present invention are aimed at the fine-tuning scenario of large earthquake models, after obtaining the seismic data volume with location information annotations, the following steps can be performed, for example, to make the obtained seismic data volume with location information annotations better applicable to the fine-tuning scenario of large earthquake models:

[0112] Analyze seismic data volumes with location information annotations; standardize the seismic data values ​​using the mean or standard deviation to ensure the values ​​are within ±5 standard deviations. Extract seismic and annotated profiles along the line and track numbers, respectively.

[0113] Construct a dataset; according to a certain ratio (e.g., a training set: validation set ratio of 8:2), take multiple seismic data volumes with location information annotations as sample datasets, and divide the sample datasets into training sets and validation sets.

[0114] The sample dataset obtained using the above method can be directly input into the large-scale seismic model to train the large-scale seismic model under fine-tuning scenarios, further improving the large-scale seismic model and thus enhancing the accuracy of the large-scale seismic model in predicting reservoirs.

[0115] The embodiments of this invention improve the sample annotation theory of large-scale oil and gas seismic models, enhance construction efficiency, and improve the coverage of oil and gas seismic sample label features. Through the four methods provided in the embodiments of this invention, the sample sets required for all scenarios related to earthquake prediction can be constructed quickly and efficiently, demonstrating strong universality.

[0116] Based on the same inventive concept, embodiments of the present invention also provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the sample annotation acquisition method using forward modeling.

[0117] Based on the same inventive concept, embodiments of the present invention also provide a sample annotation acquisition device utilizing forward simulation, the structural block diagram of which is shown below. Figure 7 As shown, it includes:

[0118] Initial module 71 is used to construct an initial reflection coefficient model that is identical to the preset reservoir volume to be simulated; the reflection coefficient model contains several data points;

[0119] Add a wrinkle module 72 to add random undulating wrinkles to the reflection coefficient model;

[0120] The offset acquisition module 73 is used to obtain the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model according to the preset principle of the seismic tectonic movement of the reservoir to be simulated.

[0121] The transformation module 74 is used to deform and / or twist the reflection coefficient model according to the offset of each data point in the three directions, and record the position information of the changed data points in the formation.

[0122] The annotation generation module 75 is used to generate annotations for each data point based on the location information of the changed data points in the strata.

[0123] The noise-adding module 76 is used to add noise to the reflection coefficient model to obtain the noise-added reflection coefficient model;

[0124] The conversion module 77 is used to convert the noisy reflection coefficient model into a seismic data volume, resulting in a seismic data volume with location information annotations.

[0125] Based on the same inventive concept, embodiments of the present invention also provide a computing device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the program executed by the processor is a method for obtaining sample annotations using forward modeling.

[0126] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a sample annotation acquisition method using forward modeling.

[0127] Since the principle behind the problem solved by these devices is similar to the aforementioned sample labeling acquisition method using forward modeling, the implementation of these devices can be found in the implementation of the aforementioned methods, and the repetitions will not be repeated.

[0128] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0129] This 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 block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can 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, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0130] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0131] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the functions specified in one or more boxes. Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of the invention. Therefore, if these modifications and variations of the invention fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A sample annotation method using forward modeling, characterized in that, include: Construct an initial reflection coefficient model that is identical to the preset reservoir volume to be simulated; the reflection coefficient model contains several data points; Add random undulating wrinkles to the reflection coefficient model; Based on the principle of the pre-set reservoir seismic tectonic movement to be simulated, the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model is obtained. Based on the offset of each data point in the reflection coefficient model in three directions, the reflection coefficient model is deformed and / or twisted, and the position information of the changed data points in the formation is recorded. Based on the location information of the changed data points in the strata, a label corresponding to each data point is generated; The reflection coefficient model is subjected to noise processing to obtain a noisy reflection coefficient model; The noisy reflection coefficient model is converted into a seismic data volume, resulting in a seismic data volume with location information annotations.

2. The method as described in claim 1, characterized in that, The construction of an initial reflection coefficient model with the same preset reservoir volume as the required simulation includes: Along the depth direction of the initial reflection coefficient model, uniformly distributed data points are set, and random numbers within the range of [0, 1] are used to assign values ​​to the uniformly distributed data points to generate a one-dimensional random sequence; The one-dimensional random sequence is copied along line and channel directions that are perpendicular to the depth direction and mutually perpendicular to each other in the horizontal direction to obtain an initial reflection coefficient model that is the same as the preset reservoir volume to be simulated.

3. The method as described in claim 2, characterized in that, Adding random undulations and wrinkles to the reflection coefficient model includes: The offset of each data point in the reflection coefficient model in the depth direction is randomly generated using Berlin noise; The data points in the reflection coefficient model are mapped to new positions in the depth direction according to the offset.

4. The method as described in claim 1, characterized in that, The step of adding noise to the reflection coefficient model to obtain a noisy reflection coefficient model includes: Construct a random Gaussian noise volume with the same volume as the reflection coefficient model; The random Gaussian noise is subjected to low-pass filtering based on the preset frequency band and signal-to-noise ratio. The low-pass filtered random Gaussian noise volume is superimposed on the reflection coefficient model to obtain the noisy reflection coefficient model.

5. The method as described in claim 1, characterized in that, The earthquake data volume is the earthquake amplitude data volume; Correspondingly, the noisy reflection coefficient model is converted into a seismic data volume, including: Randomly sample frequencies within a preset frequency band; And generate the Ricker wavelet corresponding to the frequency of the sampled data; The earthquake amplitude data volume is obtained by convolving the Rick wavelet with the noisy reflection coefficient model along the depth direction.

6. A method for labeling theoretical samples in a large-scale earthquake model, characterized in that, include: Collect the preset location information of the reservoir to be simulated, the seismic calibration of the actual drilled well, the seismic attributes of oil and gas, and the stratigraphic parameters of the geological sedimentary deposits of the oil and gas reservoir; Based on the actual well seismic calibration, expert interpretation samples were labeled to obtain seismic data volumes with location information annotations; Based on the aforementioned oil and gas seismic attributes and actual well seismic calibration, the mechanism model samples are labeled to obtain a seismic data volume with location information annotations; Based on the stratigraphic parameters of the oil and gas reservoir's geological sedimentation and the location information, a physical model is labeled to obtain a seismic data volume with location information annotations; Based on the principle of the pre-defined seismic tectonic movement of the reservoir to be simulated, the sample annotation method of forward modeling as described in any one of claims 1-5 is used to perform mathematical simulation sample annotation, thereby obtaining a seismic data volume with location information annotation.

7. The method as described in claim 6, characterized in that, The annotation of mechanistic model samples based on the oil and gas seismic attributes and actual well seismic calibration includes: Based on the mechanism model formula, the seismic properties of the oil and gas are calculated to obtain the sample sensitive attribute volume; Based on the actual drilling well vibration calibration, the reservoir threshold value of the sample sensitive attribute body is statistically analyzed. The reservoir location is obtained based on the reservoir threshold value and the sensitive attributes of the sample sensitive attribute body; The sensitive attribute volume is labeled with the reservoir location to obtain a seismic data volume with location information labeling.

8. The method as described in claim 6, characterized in that, The physical model annotation based on the stratigraphic parameters and location information of the oil and gas reservoir geological sediments includes: Based on the stratigraphic parameters of the oil and gas reservoir's geological sedimentation, design a physical model capable of inverting the preset simulated reservoir; Determine the parameters of the physical model; Construct the rock skeleton and fluid channels of the physical model; Simulated data collection; Based on the location information, the simulated data is labeled to obtain a seismic data volume with location information labels.

9. A sample annotation acquisition device using forward modeling, characterized in that, include: An initial module is used to construct an initial reflection coefficient model that is identical to the preset reservoir volume to be simulated; the reflection coefficient model contains several data points; Add a wrinkle module to add random undulating wrinkles to the reflection coefficient model; The offset acquisition module is used to obtain the offset of each data point in the reflection coefficient model in the depth direction, line number direction and trace number direction of the reflection coefficient model according to the preset principle of the seismic tectonic movement of the reservoir to be simulated. The transformation module is used to deform and / or distort the reflection coefficient model according to the offset of each data point in the three directions, and record the position information of the transformed data points in the stratum. The annotation generation module is used to generate annotations for each data point based on the location information of the changed data points in the strata. A noise-adding module is used to add noise to the reflection coefficient model to obtain a noise-added reflection coefficient model. The conversion module is used to convert the noise-added reflection coefficient model into a seismic data volume, resulting in a seismic data volume with location information annotations.

10. A computing device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the program executed by the processor implements the sample annotation acquisition method using forward modeling as described in any one of claims 1-5.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the sample annotation acquisition method using forward modeling as described in any one of claims 1-5.

12. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the sample annotation acquisition method using forward modeling as described in any one of claims 1-5.

Images