Adversarial network-based sedimentary microfacies modeling method, system and equipment and medium
By employing an adversarial network-based sedimentary microfacies modeling method, this approach utilizes the sand-to-soil ratio parameter at well points and seismic attributes to filter highly correlated seismic attributes. Combined with well facies maps and sub-layer sand-to-soil ratio maps as training conditions, an adversarial network model is generated. This solves the problem of inaccurate modeling in traditional methods, achieving precise modeling of sedimentary microfacies and characterization of complex geological patterns, thus providing better support for oil and gas field exploration and development.
Patent Information
- Application Number
- CN202410563398.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-08
- Publication Date
- 2025-11-11
AI Technical Summary
Existing sedimentary microfacies modeling methods are difficult to achieve accurate modeling. Traditional methods suffer from problems such as fractures, over-simulation, or randomness during the simulation process, resulting in insufficient capture of detailed geological features and failing to meet the needs of oil and gas field exploration and development.
An adversarial network-based sedimentary microfacies modeling method is adopted. By utilizing the sand-to-soil ratio parameter and seismic attributes at well points, the seismic attributes with strong correlation are selected through the grey relational algorithm to form a seismic attribute grid and visualize it. Combined with the well facies map and the sand-to-soil ratio map of the sub-layer, an adversarial network model is generated to achieve accurate modeling of sedimentary microfacies.
The generation of more accurate and controllable sedimentary microfacies maps improves the model's generalization ability and predictive performance, enabling it to characterize complex geological patterns and enhance the accuracy of sedimentary microfacies representation, thus providing better support for oil and gas field exploration and development.
Smart Images
Figure CN120928419A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of sedimentary microfacies modeling, and relates to a sedimentary microfacies modeling method, system, device and medium based on adversarial networks. Background Technology
[0002] Sedimentary microfacies research is a crucial aspect of development geology, holding significant importance for understanding reservoir properties, oil and gas reservoir formation, and distribution patterns. As oil and gas field exploration and development become increasingly challenging, finding reservoirs with high-quality reserves is becoming more difficult. Therefore, designing new technologies and methods for accurate modeling of sedimentary microfacies is attracting growing attention. Conditional generative adversarial networks (GANs) are among the most popular deep learning methods, capable of generating complex and realistic images based on given conditions. Learning the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies models using GANs is a key technology for achieving accurate modeling of complex sedimentary microfacies.
[0003] Sedimentary microfacies modeling plays a crucial role in oil and gas field exploration and development, as the distribution patterns of sedimentary microfacies directly influence the characteristics of underground oil and gas flows. Traditional sedimentary microfacies modeling methods mainly include sequential indicator simulation, multi-point geostatistical simulation, and goal-based stochastic simulation. While multi-point geostatistical simulation considers nonlinear geological features during the simulation process, the stationarity of training images can lead to model breaks or unnatural over-simulations. Sequential indicator simulation approximates the real geological structure by progressively simulating each pixel of the geological model, but the discrete pixel representation may result in insufficient capture of detailed geological features. Goal-based stochastic simulation can effectively characterize the geometry of complex geological bodies, but the randomness can lead to significant differences in simulation results between adjacent locations. Due to the complexity and strong heterogeneity of stratigraphic structures, and the limitations of data from well logging, core samples, and oil testing, the aforementioned sedimentary microfacies modeling methods struggle to achieve accurate modeling. Summary of the Invention
[0004] To overcome the shortcomings of the prior art, the present invention aims to provide a sedimentary microfacies modeling method, system, device and medium based on adversarial networks. The present invention can achieve accurate modeling of sedimentary microfacies and generate more accurate and controllable sedimentary microfacies maps, providing better support for the exploration and development of oil and gas fields.
[0005] To achieve the above objectives, the present invention employs the following technical solution:
[0006] In a first aspect, the present invention provides a method for modeling sedimentary microfacies based on adversarial networks, comprising the following steps:
[0007] Using the sand-to-soil ratio parameter and seismic attributes at the well point, the seismic attributes with strong correlation to the sand-to-soil ratio parameter were selected based on the grey relational algorithm.
[0008] Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey line are extracted from the target layer to form a 6×6 seismic attribute grid. The grid is then visualized to obtain a preferred seismic attribute image. Based on the preferred seismic attribute two-dimensional image and the sand-to-land ratio prediction model, the sub-layer sand-to-land ratio value is predicted. The sub-layer sand-to-land ratio value is then visualized to obtain a sub-layer sand-to-land ratio map.
[0009] Using well facies maps and small-layer sand-soil ratio maps obtained from single-well sedimentary microfacies interpretation as conditional information, and existing sedimentary microfacies images as learning data, a conditional generative adversarial network model is trained to achieve accurate modeling of sedimentary microfacies.
[0010] Secondly, this invention provides a sedimentary microfacies modeling system based on adversarial networks, comprising a seismic attribute screening module, a sand-to-land ratio map acquisition module, and a model building module connected in sequence, wherein:
[0011] Seismic attribute filtering module: This module uses the sand-soil ratio parameter and seismic attributes at the well point to filter seismic attributes that are strongly correlated with the sand-soil ratio parameter based on the grey relational algorithm.
[0012] Sand-to-land ratio acquisition module: Based on the seismic attributes with strong correlation, it extracts data from multiple profiles of the main survey line and connecting survey line in the target layer to form a 6×6 seismic attribute grid, visualizes it to obtain the preferred seismic attribute image, predicts the sub-layer sand-to-land ratio value based on the preferred seismic attribute 2D image and sand-to-land ratio prediction model, and visualizes the sub-layer sand-to-land ratio value to obtain the sub-layer sand-to-land ratio map.
[0013] Model building module: It is used to use the well facies map and the sand-soil ratio map obtained from the interpretation results of single-well sedimentary microfacies as conditional information, and the existing sedimentary microfacies images as learning data to train the conditional generative adversarial network model, so as to achieve accurate modeling of sedimentary microfacies.
[0014] Thirdly, the present invention provides an electronic device, comprising: a processor; a memory for storing computer program instructions; and steps for implementing a depositional microphase modeling method based on adversarial networks when executing the computer program.
[0015] Fourthly, the present invention provides a storage medium storing computer program instructions, which are loaded and executed by a processor, wherein the processor performs a depositional microfacies modeling method based on an adversarial network.
[0016] Compared with the prior art, the present invention has the following beneficial effects:
[0017] 1. The method of this invention utilizes the sand-soil ratio parameter and seismic attributes at the well point, and uses the grey relational algorithm to screen the seismic attributes that are strongly correlated with the sand-soil ratio parameter. By mining the seismic attributes that are strongly correlated with the sand-soil ratio parameter, the generalization ability and prediction performance of the model are improved, and the accurate prediction of the sand-soil ratio is achieved.
[0018] 2. The method of this invention, based on strongly correlated seismic attributes, extracts data from multiple profiles of the main survey line and connecting survey lines in the target layer to form a 6×6 seismic attribute grid, and visualizes it to obtain a preferred seismic attribute image. Based on the preferred seismic attribute two-dimensional image and the sand-to-soil ratio prediction model, the sub-layer sand-to-soil ratio value is predicted, and the sub-layer sand-to-soil ratio value is visualized to obtain a sub-layer sand-to-soil ratio map. This provides an intuitive understanding of the spatial distribution of the sub-layer sand-to-soil ratio, effectively constrains the construction of the sub-layer sedimentary microfacies model, and makes the generated sedimentary microfacies map more accurate. Using the well facies map and sub-layer sand-to-soil ratio map obtained from the interpretation results of single-well sedimentary microfacies as conditional information, and using existing sedimentary microfacies images as learning data, a conditional generative adversarial network model is trained to achieve accurate modeling of sedimentary microfacies and generate a more accurate and controllable sedimentary microfacies map. This invention generates more accurate and controllable sedimentary microfacies maps by learning the distribution patterns and constraints of sedimentary microfacies and their relationship with sedimentary microfacies. It can depict complex geological patterns, has a high degree of accuracy in characterizing sedimentary microfacies, and has also achieved good simulation results in phase ratio reproduction, providing better support for the exploration and development of oil and gas fields.
[0019] 3. The system of this invention includes a seismic attribute screening module, a sand-to-soil ratio map acquisition module, and a model building module connected in sequence. Specifically: the seismic attribute screening module uses the sand-to-soil ratio parameters and seismic attributes at well points to screen seismic attributes with strong correlation to the sand-to-soil ratio parameters using a grey relational algorithm; the sand-to-soil ratio map acquisition module extracts data from multiple profiles of the main survey line and connecting survey lines in the target layer based on the strongly correlated seismic attributes, forms a 6×6 seismic attribute grid, visualizes it to obtain a preferred seismic attribute image, predicts the sub-layer sand-to-soil ratio value based on the preferred seismic attribute 2D image and the sand-to-soil ratio prediction model, and visualizes the sub-layer sand-to-soil ratio value to obtain a sub-layer sand-to-soil ratio map; the model building module uses the well facies map obtained based on the single-well sedimentary microfacies interpretation results and the sub-layer sand-to-soil ratio map as conditional information, and uses existing sedimentary microfacies images as learning data to train a conditional generative adversarial network model. The modules work together to achieve accurate modeling of sedimentary microfacies, generating more accurate and controllable sedimentary microfacies maps, providing better support for the exploration and development of oil and gas fields.
[0020] 4. The equipment and medium of this invention can also achieve accurate modeling of sedimentary microfacies, generating more accurate and controllable sedimentary microfacies maps, providing better support for the exploration and development of oil and gas fields. Attached Figure Description
[0021] Figure 1 This is a flowchart of the method of the present invention;
[0022] Figure 2 This is a schematic diagram of the sedimentary microfacies modeling method based on conditional generative adversarial networks of the present invention;
[0023] Figure 3 This is a diagram of the sand-to-land ratio prediction model based on convolutional networks of the present invention.
[0024] Figure 4 This is a system module diagram of the present invention. Detailed Implementation
[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0027] The present invention will now be described in further detail with reference to the accompanying drawings:
[0028] See Figure 1 , one A sedimentary microfacies modeling method based on adversarial networks includes the following steps:
[0029] S1. Using the sand-soil ratio parameter and seismic attributes at the well point, the seismic attributes with strong correlation to the sand-soil ratio parameter are selected according to the grey relational algorithm, as follows:
[0030] The sand-to-soil ratio parameter at the well point is used as the reference sequence, and the seismic attributes at the well point are used as the comparison sequence. The grey relational degree between the reference sequence and the comparison sequence is calculated using the grey relational analysis algorithm. Based on the grey relational degree, the seismic attributes with strong correlation to the sand-to-soil ratio parameter are selected.
[0031] Preferably, the process for obtaining seismic attributes at the well point is as follows:
[0032] Based on the coordinate information of the seismic attribute data, the n nearest locations to the well point are calculated and selected. The seismic attributes of the n nearest neighbor locations are then weighted and summed to obtain the seismic attributes at the well point.
[0033] The preferred formula for calculating seismic attributes at the well point is as follows:
[0034]
[0035]
[0036] Where Z0 represents the seismic attribute at the well point, Z i Let ω be the seismic attribute from the well point to the i-th nearest neighbor location, i = 1, 2, ..., n. i d represents the weight of the i-th nearest neighbor position. i It represents the distance from the well point to the i-th nearest neighbor.
[0037] Preferably, seismic attributes with strong correlation to the sand-land ratio parameter are selected using the grey relational analysis algorithm, as follows:
[0038] Using the sand-to-land ratio parameter as the reference sequence and the seismic attributes as the comparison sequence, the grey relational degree of the reference sequence and the comparison sequence is calculated and sorted. Based on the sorted grey relational degree, m seismic attributes that are sensitive to the sand-to-land ratio parameter are selected. The selected m seismic attributes are the seismic attributes with strong correlation.
[0039] S2. Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey line are extracted from the target layer to form a 6×6 seismic attribute grid. The grid is then visualized to obtain a preferred seismic attribute image. Based on the preferred seismic attribute two-dimensional image and the sand-to-land ratio prediction model, the sub-layer sand-to-land ratio value is predicted. The sub-layer sand-to-land ratio value is then visualized to obtain a sub-layer sand-to-land ratio map.
[0040] The preferred sand-to-land ratio prediction model is trained as follows:
[0041] Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey line are extracted from the target layer to form a 6×6 seismic attribute grid. The grid is then visualized to obtain the preferred seismic attribute images. These preferred seismic attribute images are used as training data, and the sand-to-soil ratio values at well points are used as training labels to construct a sand-to-soil ratio prediction sample set. Each sand-to-soil ratio prediction sample set contains m preferred seismic attribute two-dimensional images and one sand-to-soil ratio value.
[0042] A convolutional neural network was used to train a sand-land ratio prediction model based on a sand-land ratio prediction sample set.
[0043] S3. Using the well facies map and the small-layer sand-soil ratio map obtained from the interpretation results of single-well sedimentary microfacies as conditional information, and using existing sedimentary microfacies images as learning data, a conditional generative adversarial network model is trained to achieve accurate modeling of sedimentary microfacies.
[0044] Preferably, the well facies map and the sand-soil ratio map obtained based on the interpretation results of single-well sedimentary microfacies are used as conditional information, and the existing sedimentary microfacies images are used as learning data to train the conditional generative adversarial network model, as follows:
[0045] The generator inputs are random noise added to the small-layer sand-land ratio map and the well facies map obtained based on the single-well sedimentary microfacies interpretation results. The sedimentary microfacies image generated by the generator, the real sedimentary microfacies image and the conditional data are used as the inputs of the discriminator. The generator attempts to generate more realistic sedimentary microfacies images that match the conditional data, while the discriminator attempts to more accurately distinguish between the real sedimentary microfacies image and the generated sedimentary microfacies image. The generator and the discriminator are trained against each other to finally obtain the desired generator.
[0046] This generator learns the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies models, enabling the generation of more accurate sedimentary microfacies models.
[0047] See Figure 1In another feasible embodiment of the present invention, the following modifications are made as appropriate. Using the sand-to-soil ratio parameter and seismic attributes at the well point, a grey relational algorithm is used to filter seismic attributes with strong correlation to the sand-to-soil ratio parameter. By mining these strongly correlated seismic attributes, the generalization ability and predictive performance of the model are improved, achieving accurate prediction of the sand-to-soil ratio. Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey lines are extracted from the target layer to form a 6×6 seismic attribute grid. This grid is then visualized to obtain a preferred seismic attribute image. The sand-to-soil ratio value of the sub-layer is predicted based on the preferred seismic attribute two-dimensional image and the sand-to-soil ratio prediction model. The sub-layer... The numerical visualization of sand-to-soil ratios yields sublayer sand-to-soil ratio maps, providing an intuitive understanding of their spatial distribution and effectively constraining the construction of sublayer sedimentary microfacies models, resulting in more accurate sedimentary microfacies maps. By using well facies maps and sublayer sand-to-soil ratio maps derived from single-well sedimentary microfacies interpretation as conditional information, and existing sedimentary microfacies images as learning data, a conditional generative adversarial network (GAN) model is trained. This enables precise modeling of sedimentary microfacies, generating more accurate and controllable sedimentary microfacies maps. This helps exploration personnel accurately understand subsurface structures and sedimentary characteristics, providing better support for oil and gas field exploration and development. This invention generates more accurate and controllable sedimentary microfacies maps by learning the distribution patterns and constraints of sedimentary microfacies, enabling the depiction of complex geological patterns. It achieves high accuracy in representing sedimentary microfacies and demonstrates good simulation results in facies ratio reproduction, providing better support for oil and gas field exploration and development.
[0048] Example 1:
[0049] See Figure 1 and Figure 2 This embodiment discloses a sedimentary microfacies modeling method based on adversarial networks. It employs a grey relational analysis algorithm to mine seismic attributes with strong correlation to the sand-to-soil ratio parameter. Seismic attribute images are selected as input to a convolutional neural network model to obtain sand-to-soil ratio prediction results and visualize them. Local and global features are used as joint constraints. Local features are well facies data, and global features are sub-layer sand-to-soil ratio maps. An adversarial network is trained to learn the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies, generating a more accurate and controllable sedimentary microfacies map. The specific steps are as follows:
[0050] S1. Based on the coordinate information of the seismic attribute data, calculate and filter the n nearest locations to the well point. Then, perform a weighted sum of the seismic attributes of the n nearest neighbor locations to obtain the seismic attribute at the well point. Let the well point coordinates be (X0, Y0), and the coordinates of the i-th location be (X... i ,Y i The seismic attribute from the well point to the i-th nearest neighbor is Z. iLet i = 1, 2, ..., n, and let d be the distance from the well point to the i-th nearest neighbor. i Where i = 1, 2, ..., n. The formula for calculating the seismic attribute Z0 at the well point is as follows:
[0051]
[0052] Where ω i This represents the weight of the i-th nearest neighbor position.
[0053]
[0054] For the sand-soil ratio parameter and seismic attribute data at well locations, a grey relational analysis algorithm is used to analyze sensitive parameters and filter parameters sensitive to the sand-soil ratio. The sand-soil ratio is set as the reference sequence, and the seismic attributes as the comparison sequence. The grey relational degree of the reference sequence and the comparison sequence is calculated and sorted to filter out m seismic attributes sensitive to the sand-soil ratio parameter.
[0055] S2. Based on the above sensitivity parameter analysis results, the preferred seismic attribute images within a 6×6 seismic trace area centered on the well point are used as training data. The sand-to-soil ratio values at the well point are used as training labels. Each sample contains m preferred seismic attribute two-dimensional images and one sand-to-soil ratio value. This forms the sand-to-soil ratio prediction sample set. A convolutional neural network is used to train the sand-to-soil ratio prediction model. (See [link to relevant documentation]). Figure 3 This is a diagram of the sand-to-land ratio prediction model based on convolutional networks.
[0056] Based on the sand-to-land ratio prediction model, a two-dimensional image of the preferred seismic attributes is input to realize the numerical prediction of the sand-to-land ratio of the sub-layer, and the prediction results of the sub-layer sand-to-land ratio are visualized to obtain the sand-to-land ratio map of the sub-layer.
[0057] S3. Using sand-to-land ratio maps and well facies maps obtained from single-well sedimentary microfacies interpretation as conditional information, and existing sedimentary microfacies images as training data, a conditional generative adversarial network (GAN) model is trained. During training, randomly generated noise is added to the sand-to-land ratio map and well facies map as input to the generator. The sedimentary microfacies images generated by the generator, real sedimentary microfacies images, and conditional data are used as input to the discriminator. The generator attempts to generate more realistic sedimentary microfacies images that conform to the conditional data, while the discriminator attempts to more accurately distinguish between real and generated samples. The generator and discriminator train against each other to ultimately obtain the desired generator. This generator learns the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies models, enabling the generation of more accurate sedimentary microfacies models.
[0058] This invention can learn the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies, generating more accurate and controllable sedimentary microfacies maps. It can depict complex geological models, has a high degree of accuracy in characterizing sedimentary microfacies, and has also achieved good simulation results in phase ratio reproduction, providing better support for the exploration and development of oil and gas fields.
[0059] Example 2:
[0060] See Figure 1 and Figure 2 This embodiment selects seismic attribute data from seven sub-layers in an oilfield. Using the inverse distance weighting method, the four locations closest to the well point are calculated and selected based on the coordinate information of the seismic attribute data. The seismic attributes of these four nearest neighbor locations are then weighted and summed to obtain the seismic attributes at the well point. For the sand-soil ratio parameter and corresponding seismic attribute data at the well point location, a grey relational algorithm is used. The sand-soil ratio is set as the reference sequence, and each seismic attribute is set as a comparison sequence. The grey relational degree between the reference and comparison sequences is calculated, and the seismic attributes are sorted based on the grey relational degree results. According to the sorting results, seismic attributes with a correlation degree greater than 0.6 are selected as parameters sensitive to the sand-soil ratio. Analysis shows that the parameters sensitive to the sand-soil ratio include: arc length, average instantaneous frequency, minimum value, root mean square amplitude, total amplitude, and amplitude variation variance.
[0061] Based on the results of sensitive parameter analysis, a two-dimensional seismic attribute image centered on the well point is visualized. Sand-soil ratio parameters at the same well point location are matched with the sensitive seismic attribute image as a sample. Each sample contains six two-dimensional seismic attribute images and one sand-soil ratio value, thus constructing a sand-soil ratio prediction sample set. A convolutional neural network is used to train the sand-soil ratio prediction model. Based on the trained sand-soil ratio prediction model, sand-soil ratio values at non-well point locations are predicted, and the results are visualized to obtain the sand-soil ratio map of the region, i.e., the sub-layer sand-soil ratio map.
[0062] Using sand-to-land ratio maps and well facies maps obtained from single-well sedimentary microfacies interpretation as conditional information, and existing sedimentary microfacies images as training data, a conditional generative adversarial network (GAN) model is trained. Randomly generated noise is added to the sand-to-land ratio maps and well facies maps as input to the generator. The generated sedimentary microfacies images, real sedimentary microfacies images, and conditional data are used as input to the discriminator. The generator attempts to generate more realistic sedimentary microfacies images that conform to the conditional data, while the discriminator attempts to more accurately distinguish between real and generated samples. The generator and discriminator train against each other to ultimately obtain the desired generator. This generator learns the distribution patterns of sedimentary microfacies and the relationship between constraints and the sedimentary microfacies model, constructing a more accurate sedimentary microfacies model.
[0063] This invention employs a grey relational analysis algorithm, using the sand-land ratio parameter as a reference sequence and the seismic attribute data as a comparison sequence, to calculate the grey relational degree and identify seismic attributes that are strongly correlated with the sand-land ratio parameter.
[0064] Using two-dimensional seismic attribute images with strong correlation centered on well points as input, a convolutional neural network model is constructed to achieve accurate prediction of sand-land ratio, resulting in a predicted sand-land ratio image.
[0065] Using sand-land ratio images and sedimentary microfacies scatter plots drawn based on single-well sedimentary microfacies interpretation results as conditional information, and existing sedimentary microfacies maps based on expert experience as learning samples, a conditional adversarial network model is trained to achieve accurate modeling of sedimentary microfacies, so that the sedimentary microfacies model can both depict complex geological patterns and fit the conditional data well.
[0066] See Figure 4 Based on the above method, this invention discloses a sedimentary microfacies modeling system based on adversarial networks, comprising a seismic attribute screening module, a sand-to-land ratio map acquisition module, and a model building module connected in sequence, wherein:
[0067] Seismic attribute filtering module: This module uses the sand-soil ratio parameter and seismic attributes at the well point to filter seismic attributes that are strongly correlated with the sand-soil ratio parameter based on the grey relational algorithm.
[0068] Sand-to-land ratio acquisition module: Based on the earthquake attributes with strong correlation, it extracts data from multiple profiles of the main survey line and connecting survey line in the target layer to form a 6×6 earthquake attribute grid, visualizes it to obtain the preferred earthquake attribute image, predicts the sub-layer sand-to-land ratio value based on the preferred earthquake attribute 2D image and sand-to-land ratio prediction model, and visualizes the sub-layer sand-to-land ratio value to obtain the sub-layer sand-to-land ratio map.
[0069] Model building module: It is used to use the well facies map and the sand-soil ratio map obtained from the interpretation results of single-well sedimentary microfacies as conditional information, and the existing sedimentary microfacies images as learning data to train the conditional generative adversarial network model, so as to achieve accurate modeling of sedimentary microfacies.
[0070] See Figure 4In another feasible embodiment of the present invention, the following modifications are made as needed. It includes a seismic attribute screening module, a sand-to-soil ratio map acquisition module, and a model building module connected in sequence. Specifically: the seismic attribute screening module uses the sand-to-soil ratio parameters and seismic attributes at well points to screen seismic attributes with strong correlation to the sand-to-soil ratio parameters using a grey relational algorithm; the sand-to-soil ratio map acquisition module, based on the strongly correlated seismic attributes, extracts data from multiple profiles of the main survey line and connecting survey lines in the target layer to form a 6×6 seismic attribute grid, visualizes it to obtain a preferred seismic attribute image, predicts the sub-layer sand-to-soil ratio value based on the preferred seismic attribute two-dimensional image and the sand-to-soil ratio prediction model, and visualizes the sub-layer sand-to-soil ratio value to obtain a sub-layer sand-to-soil ratio map; the model building module uses the well facies map obtained based on the single-well sedimentary microfacies interpretation results and the sub-layer sand-to-soil ratio map as conditional information, and uses existing sedimentary microfacies images as learning data to train a conditional generative adversarial network model. These modules work together to achieve accurate modeling of sedimentary microfacies, generating more precise and controllable sedimentary microfacies maps, providing better support for the exploration and development of oil and gas fields.
[0071] An electronic device includes: a processor; a memory for storing computer program instructions; and steps for implementing a depositional microphase modeling method based on adversarial networks when executing the computer program.
[0072] A storage medium storing computer program instructions, which are loaded and executed by a processor, wherein the processor performs a sedimentary microfacies modeling method based on an adversarial network.
[0073] 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, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0074] 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.
[0075] 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.
[0076] 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 function specified in one or more boxes.
[0077] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.
Claims
1. A sedimentary microfacies modeling method based on adversarial networks, characterized in that, Includes the following steps: Using the sand-to-soil ratio parameter and seismic attributes at the well point, the seismic attributes with strong correlation to the sand-to-soil ratio parameter were selected based on the grey relational algorithm. Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey line are extracted from the target layer to form a 6×6 seismic attribute grid. The grid is then visualized to obtain a two-dimensional seismic attribute image. The sub-layer sand-to-land ratio is predicted based on the optimized two-dimensional seismic attribute image and the sand-to-land ratio prediction model. The sub-layer sand-to-land ratio is then visualized to obtain a sub-layer sand-to-land ratio map. Using well facies maps and small-layer sand-soil ratio maps obtained from single-well sedimentary microfacies interpretation as conditional information, and existing sedimentary microfacies images as learning data, a conditional generative adversarial network model is trained to achieve accurate modeling of sedimentary microfacies.
2. The sedimentary microfacies modeling method based on adversarial networks according to claim 1, characterized in that, The process involves using the sand-soil ratio parameter and seismic attributes at the well point, and then using a grey relational algorithm to filter out seismic attributes that are strongly correlated with the sand-soil ratio parameter, as detailed below: The sand-to-soil ratio parameter at the well point is used as the reference sequence, and the seismic attributes at the well point are used as the comparison sequence. The grey relational degree between the reference sequence and the comparison sequence is calculated using the grey relational analysis algorithm. Based on the grey relational degree, the seismic attributes with strong correlation to the sand-to-soil ratio parameter are selected.
3. The sedimentary microfacies modeling method based on adversarial networks according to claim 2, characterized in that, The process of obtaining the seismic attributes at the well point is as follows: Based on the coordinate information of the seismic attribute data, the n nearest locations to the well point are calculated and selected. The seismic attributes of the n nearest neighbor locations are then weighted and summed to obtain the seismic attributes at the well point.
4. The sedimentary microfacies modeling method based on adversarial networks according to claim 3, characterized in that, The formula for calculating the seismic attributes at the well point is as follows: Where Z0 represents the seismic attribute at the well point, Z i Let ω be the seismic attribute from the well point to the i-th nearest neighbor location, i = 1, 2, ..., n. i d represents the weight of the i-th nearest neighbor position. i It represents the distance from the well point to the i-th nearest neighbor.
5. The sedimentary microfacies modeling method based on adversarial networks according to claim 2, characterized in that, The process of using the grey relational analysis algorithm to filter seismic attributes with strong correlation to the sand-land ratio parameter is as follows: Using the sand-to-land ratio parameter as the reference sequence and the seismic attributes as the comparison sequence, the grey relational degree of the reference sequence and the comparison sequence is calculated and sorted. Based on the sorted grey relational degree, m seismic attributes that are sensitive to the sand-to-land ratio parameter are selected. The selected m seismic attributes are the seismic attributes with strong correlation.
6. The sedimentary microfacies modeling method based on adversarial networks according to claim 1, characterized in that, The training process for the sand-to-land ratio prediction model is as follows: Based on the strongly correlated seismic attributes, data from multiple profiles of the main survey line and connecting survey lines are extracted from the target layer to form a 6×6 seismic attribute grid, which is then visualized to obtain the optimal seismic attribute image. Using the preferred seismic attribute images as training data and the sand-to-soil ratio values at well points as training labels, a sand-to-soil ratio prediction sample set is constructed. Each sand-to-soil ratio prediction sample set contains m preferred seismic attribute two-dimensional images and 1 sand-to-soil ratio value. A convolutional neural network was used to train a sand-land ratio prediction model based on a sand-land ratio prediction sample set.
7. The sedimentary microfacies modeling method based on adversarial networks according to claim 1, characterized in that, The well facies map and the sand-soil ratio map obtained based on the interpretation of single-well sedimentary microfacies are used as conditional information, and existing sedimentary microfacies images are used as learning data to train a conditional generative adversarial network model, as detailed below: The generator inputs are random noise added to the small-layer sand-land ratio map and the well facies map obtained based on the single-well sedimentary microfacies interpretation results. The sedimentary microfacies image generated by the generator, the real sedimentary microfacies image and the conditional data are used as the inputs of the discriminator. The generator attempts to generate more realistic sedimentary microfacies images that match the conditional data, while the discriminator attempts to more accurately distinguish between the real sedimentary microfacies image and the generated sedimentary microfacies image. The generator and the discriminator are trained against each other to finally obtain the desired generator. This generator learns the distribution patterns of sedimentary microfacies and the relationship between constraints and sedimentary microfacies models, enabling the generation of more accurate sedimentary microfacies models.
8. A sedimentary microfacies modeling system based on adversarial networks for implementing the method of any one of claims 1 to 7, characterized in that, It includes a seismic attribute filtering module, a sand-to-land ratio map acquisition module, and a model building module, which are connected in sequence. Seismic attribute filtering module: This module uses the sand-soil ratio parameter and seismic attributes at the well point to filter seismic attributes that are strongly correlated with the sand-soil ratio parameter based on the grey relational algorithm. Sand-to-land ratio acquisition module: Based on the seismic attributes with strong correlation, it extracts data from multiple profiles of the main survey line and connecting survey line in the target layer to form a 6×6 seismic attribute grid, visualizes it to obtain the preferred seismic attribute image, predicts the sub-layer sand-to-land ratio value based on the preferred seismic attribute 2D image and sand-to-land ratio prediction model, and visualizes the sub-layer sand-to-land ratio value to obtain the sub-layer sand-to-land ratio map. Model building module: It is used to use the well facies map and the sand-soil ratio map obtained from the interpretation results of single-well sedimentary microfacies as conditional information, and the existing sedimentary microfacies images as learning data to train the conditional generative adversarial network model, so as to achieve accurate modeling of sedimentary microfacies.
9. An electronic device, comprising: processor; A memory for storing computer program instructions; characterized in that, when executing the computer program, it implements the steps of the depositional microphase modeling method based on any one of claims 1-7.
10. A storage medium storing computer program instructions, characterized in that, When the computer program instructions are loaded and run by the processor, the processor executes the sedimentary microfacies modeling method based on adversarial networks as described in any one of claims 1-7.