Model processing method, device and equipment for reservoir prediction based on closed-loop network

Through a closed-loop network-based method, using two-dimensional label images to train the reservoir prediction model, the problem of low reservoir recognition accuracy in the prior art is solved, and higher reservoir recognition accuracy and spatial continuity capture are achieved.

CN114782744BActive Publication Date: 2024-11-22TSINGHUA UNIVERSITY
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202210406342.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-18
Publication Date
2024-11-22
Estimated Expiration
2042-04-18

AI Technical Summary

Technical Problem

In the prior art, reservoir prediction models have low accuracy in reservoir identification, mainly because convolutional neural networks trained using one-dimensional labels cannot effectively capture the spatial continuity and mutual influence of underground reservoirs.

Method used

Using a closed-loop network-based method, the model is trained and updated by obtaining the two-dimensional label image in the sample data, including the reservoir probability label image of the first and second regions, and the storage is generated using the encoder, converter and decoder. The layer probability predicts the image and updates the model parameters based on the predicted image and the label image.

Benefits of technology

It improves the accuracy of reservoir identification, can more accurately reflect the spatial continuity and probability relationship between reservoirs, and enhances the model's ability to identify reservoirs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114782744B_ABST
    Figure CN114782744B_ABST
Patent Text Reader

Abstract

The embodiment of the present application provides a model processing method, device and equipment for reservoir prediction based on a closed-loop network, the method comprising: obtaining sample data, the sample data comprising a sample image of a first region, a first label image and a second label image corresponding to the sample image, the first label image being used to indicate the probability of a reservoir existing in a first partial region of the first region, and the second label image being used to indicate the probability of a reservoir existing in a second partial region of the first region; processing the sample image by a first model to obtain a reservoir probability prediction image of the first region; updating the model parameters of the first model according to the first label image, the second label image and the reservoir probability prediction image. The accuracy of the model in identifying reservoirs is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of geophysical technology, and in particular to a model processing method, device and equipment for reservoir prediction based on a closed-loop network. Background Art

[0002] Reservoirs are rock formations that can store and permeate fluids. Reservoir prediction can help identify the distribution of underground oil reservoirs, thereby reducing the difficulty of oil and gas exploration and development.

[0003] At present, the convolutional neural network can be trained through the one-dimensional labels obtained from seismic data, and then the reservoir can be identified through the trained convolutional neural network. For example, multiple attributes of seismic data (such as ant body, texture attributes, etc.) are spliced ​​in the horizontal direction to obtain a one-dimensional label, and then the convolutional neural network is trained through the one-dimensional label. However, the underground reservoir has spatial continuity, and the probability of the existence of reservoirs in space will affect each other. The convolutional neural network is trained through one-dimensional labels, and only the longitudinal one-dimensional depth features are used, which leads to the low accuracy of the model in reservoir identification. Summary of the invention

[0004] The embodiments of the present application provide a closed-loop network-based reservoir prediction model processing method, device and equipment, which are used to solve the technical problem of low accuracy of reservoir identification by the model in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a model processing method for reservoir prediction based on a closed-loop network, the method comprising:

[0006] Acquire sample data, the sample data comprising a sample image of a first region, a first label image corresponding to the sample image, and a second label image, the first label image being used to indicate a probability that a reservoir exists in a first partial region of the first region, and the second label image being used to indicate a probability that a reservoir exists in a second partial region of the first region;

[0007] Processing the sample image by using a first model to obtain a reservoir probability prediction image of the first area;

[0008] The model parameters of the first model are updated according to the first label image, the second label image, and the reservoir probability prediction image.

[0009] In a possible implementation, obtaining a first label image corresponding to the sample image includes:

[0010] Processing the sample image by a second model to obtain a first label sub-image, wherein the second model is obtained by learning multiple groups of first samples, each group of first samples including a sample image of a first region and a sample label sub-image;

[0011] A first mask is added to the first label sub-image to obtain the first label image, wherein the first mask is used to cover the second partial area corresponding to the sample image.

[0012] In a possible implementation, obtaining a second label image corresponding to the sample image includes:

[0013] Performing polynomial linear regression processing on the sample image to obtain a second label sub-image;

[0014] A second mask is added to the second label sub-image to obtain the second label image, where the second mask is used to cover the first partial area corresponding to the sample image.

[0015] In a possible implementation manner, the first model includes an encoder, a converter, and a decoder; processing the sample image by the first model to obtain a reservoir probability prediction image of the first area includes:

[0016] Acquire the features of the sample image through the encoder, and send the features of the sample image to the converter;

[0017] Converting the features of the sample image into a feature vector by the converter, and sending the feature vector to the decoder;

[0018] The decoder converts the feature vector into partial features corresponding to the sample image, and generates the reservoir probability prediction image according to the partial features.

[0019] In a possible implementation, updating the model parameters of the first model according to the first label image, the label annotated image, and the reservoir probability prediction image includes:

[0020] Obtaining a first weight of a first loss function between the first label image and the reservoir probability prediction image;

[0021] Obtaining a second weight of a second loss function between the second label image and the reservoir probability prediction image;

[0022] Update the model parameters of the first model according to the first loss function, the second loss function, the first weight, and the second weight.

[0023] In a possible implementation, after updating the model parameters of the first model according to the first label image, the label annotated image, and the reservoir probability prediction image, the method further includes:

[0024] Processing the reservoir probability prediction image by a third model to obtain a target sample image corresponding to the reservoir probability prediction image, wherein the third model is obtained by learning a plurality of groups of second samples, each group of second samples including a sample reservoir probability prediction image and a sample image of the first area;

[0025] The model parameters of the first model are updated according to the sample image and the target sample image.

[0026] In a second aspect, the present application provides a model processing method for reservoir prediction based on a closed-loop network, the method comprising:

[0027] acquiring at least one reservoir image set of a first region;

[0028] The at least one reservoir image set is processed according to a first model to obtain a reservoir probability prediction image of the first area, and the first model is the first model described in any one of the first aspects.

[0029] In a third aspect, the present application provides a model processing device for reservoir prediction based on a closed-loop network, comprising an acquisition module, a processing module and an update module, wherein:

[0030] The acquisition module is used to acquire sample data, the sample data including a sample image of a first area, a first label image and a second label image corresponding to the sample image, the first label image is used to indicate the probability that a reservoir exists in a first partial area of ​​the first area, and the second label image is used to indicate the probability that a reservoir exists in a second partial area of ​​the first area;

[0031] The processing module is used to process the sample image through a first model to obtain a reservoir probability prediction image of the first area;

[0032] The updating module is used to update the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image.

[0033] In a possible implementation manner, the acquisition module is specifically used to:

[0034] Processing the sample image by a second model to obtain a first label sub-image, wherein the second model is obtained by learning multiple groups of first samples, each group of first samples including a sample image of a first region and a sample label sub-image;

[0035] A first mask is added to the first label sub-image to obtain the first label image, wherein the first mask is used to cover the second partial area corresponding to the sample image.

[0036] In a possible implementation manner, the acquisition module is specifically used to:

[0037] Performing polynomial linear regression processing on the sample image to obtain a second label sub-image;

[0038] A second mask is added to the second label sub-image to obtain the second label image, where the second mask is used to cover the first partial area corresponding to the sample image.

[0039] In a possible implementation manner, the processing module is specifically used to:

[0040] Acquire the features of the sample image through the encoder, and send the features of the sample image to the converter;

[0041] Converting the features of the sample image into a feature vector by the converter, and sending the feature vector to the decoder;

[0042] The decoder converts the feature vector into partial features corresponding to the sample image, and generates the reservoir probability prediction image according to the partial features.

[0043] In a possible implementation manner, the update module is specifically used to:

[0044] Obtaining a first weight of a first loss function between the first label image and the reservoir probability prediction image;

[0045] Obtaining a second weight of a second loss function between the second label image and the reservoir probability prediction image;

[0046] Update the model parameters of the first model according to the first loss function, the second loss function, the first weight, and the second weight.

[0047] In a possible implementation manner, the processing module is further configured to:

[0048] Processing the reservoir probability prediction image by a third model to obtain a target sample image corresponding to the reservoir probability prediction image, wherein the third model is obtained by learning a plurality of groups of second samples, each group of second samples including a sample reservoir probability prediction image and a sample image of the first area;

[0049] The model parameters of the first model are updated according to the sample image and the target sample image.

[0050] In a fourth aspect, the present application provides a model processing device for reservoir prediction based on a closed-loop network, comprising an acquisition module and a processing module, wherein:

[0051] The acquisition module is used to acquire at least one reservoir image set of the first area;

[0052] The processing module is used to process the at least one reservoir image set according to a first model to obtain a reservoir probability prediction image of the first area, and the first model is the first model described in any one of claims 1-6.

[0053] In a fifth aspect, an embodiment of the present application provides a model processing device for reservoir prediction based on a closed-loop network, comprising: a memory, a processor and a communication interface, the memory being used to store program instructions, the processor being used to call the program instructions in the memory to execute the model processing method for reservoir prediction based on a closed-loop network as described in any one of the first aspect, or the model processing method for reservoir prediction based on a closed-loop network as described in any one of the second aspect.

[0054] In a sixth aspect, an embodiment of the present application provides a readable storage medium having a computer program stored thereon; the computer program is used to implement a model processing method for reservoir prediction based on a closed-loop network as described in any one of the first aspects.

[0055] The embodiment of the present application provides a model processing method, device and equipment for reservoir prediction based on a closed-loop network, obtaining sample data, wherein the sample data includes a sample image of a first region, a first label image and a second label image corresponding to the sample image, the first label image is used to indicate the probability of a reservoir existing in a first partial region of the first region, and the second label image is used to indicate the probability of a reservoir existing in a second partial region of the first region, the sample image is processed by the first model to obtain a reservoir probability prediction image of the first region, and the model parameters of the first model are updated according to the first label image, the second label image and the reservoir probability prediction image. In the above method, since the sample data of the first model includes a two-dimensional label image, and the first label image and the second label image can accurately reflect the probability of reservoirs existing in different partial regions under the first region, the label accuracy of the first model is high, and the label is a two-dimensional image label, which can accurately reflect the spatial continuity between reservoirs, thereby improving the accuracy of the first model in identifying reservoirs. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0057] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application;

[0058] Figure 2 A schematic flow chart of a model processing method for reservoir prediction based on a closed-loop network provided in an embodiment of the present application;

[0059] Figure 3 A schematic diagram of a sample image provided in an embodiment of the present application;

[0060] Figure 4 A schematic diagram of the structure of a first model provided in an embodiment of the present application;

[0061] Figure 5 A schematic diagram of updating model parameters of a first model provided in an embodiment of the present application;

[0062] Figure 6 A schematic flow chart of another closed-loop network-based reservoir prediction model processing method provided in an embodiment of the present application;

[0063] Figure 7 A schematic diagram of a process for obtaining a reservoir probability identification model provided in an embodiment of the present application;

[0064] Figure 8 A schematic diagram of a process for obtaining a first label sub-image of a sample provided in an embodiment of the present application;

[0065] Fig. 9 A schematic diagram of another process of updating model parameters of a first model provided in an embodiment of the present application;

[0066] Fig.10 A schematic flow chart of a model processing method for reservoir prediction based on a closed-loop network provided in an embodiment of the present application;

[0067] Fig.11 A schematic diagram of a reservoir probability prediction image provided in an embodiment of the present application;

[0068] Fig.12 A schematic diagram of the structure of a model processing device for reservoir prediction based on a closed-loop network provided in an embodiment of the present application;

[0069] Fig.13A schematic diagram of the structure of the terminal device provided for this application. DETAILED DESCRIPTION

[0070] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0071] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0072] In related technologies, reservoir prediction can help identify the distribution of underground oil reservoirs, thereby assisting oil and gas exploration and development and reducing the cost and difficulty of oil and gas exploration and development. At present, reservoirs can be identified through trained convolutional neural networks, where the training samples of the convolutional neural network can include one-dimensional labels obtained through seismic data. For example, multiple attributes of seismic data are spliced ​​in the horizontal direction to obtain a one-dimensional label, and then the one-dimensional label is used as a label in the training sample of the convolutional neural network to train the convolutional neural network. However, the convolutional neural network obtained by one-dimensional label training can only recognize the longitudinal one-dimensional depth features in the seismic data, while the underground reservoir has spatial continuity, and the probability of the existence of reservoirs between spaces will affect each other, which leads to the low accuracy of the model in reservoir identification.

[0073] In order to solve the technical problem of low accuracy of reservoir identification by models in related technologies, an embodiment of the present application provides a model processing method for reservoir prediction based on a closed-loop network, wherein sample data is acquired, the sample data including a sample image of a first area, a first label image and a second label image corresponding to the sample image, the first label image being used to indicate the probability of a reservoir existing in a first partial area of ​​the first area, and the second label image being used to indicate the probability of a reservoir existing in a second partial area of ​​the first area, features of the sample image being acquired through an encoder of the first model, and the features of the sample image being sent to a converter of the first model, the features of the sample image being received through the converter, and the features of the sample image being converted into a feature vector, the feature vector being sent to a decoder of the first model, the feature vector being received through the decoder, the feature vector being converted into partial features corresponding to the sample image, and a reservoir probability prediction image being generated based on the partial features, and model parameters of the first model being updated based on the first label image, the second label image, and the reservoir probability prediction image. In this way, since the training sample data of the first model includes a two-dimensional label image, and the first label image and the second label image can accurately reflect the probability of the existence of reservoirs in different partial areas under the first area, the label accuracy of the first model is relatively high, and the label is a two-dimensional image label, which can accurately reflect the spatial continuity between reservoirs, thereby improving the accuracy of the first model in reservoir identification.

[0074] Next, combine Figure 1 , the application scenarios of this application are described.

[0075] Figure 1 This is a schematic diagram of an application scenario provided by the embodiment of this application. Figure 1 , including: a first model. Wherein, a sample image, a first label image and a second label image are input into the first model, and the first model can process the sample image to obtain a reservoir probability prediction image, and then update the model parameters of the first model through the reservoir probability prediction image, the first label image and the second label image. Since the training sample data of the first model includes a two-dimensional label image, and the first label image and the second label image can accurately reflect the probability of the existence of reservoirs in different partial areas under the first area, the label accuracy of the first model is relatively high, and the label is a two-dimensional image label, which can accurately reflect the spatial continuity between reservoirs, thereby improving the accuracy of the first model in reservoir identification.

[0076] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0077] Figure 2 A schematic diagram of a flow chart of a model processing method for reservoir prediction based on a closed-loop network provided in an embodiment of the present application. Figure 2 , the method may include:

[0078] S201. Obtain sample data.

[0079] The execution subject of the embodiment of the present application may be a server, or a closed-loop network-based reservoir prediction model processing device disposed in the server. The closed-loop network-based reservoir prediction model processing device may be implemented by software, or the closed-loop network-based reservoir prediction model processing device may be implemented by a combination of software and hardware.

[0080] Optionally, the sample data includes a sample image of the first area, a first label image corresponding to the sample image, and a second label image. The first area may be an area where reservoir prediction is to be performed. For example, if reservoir prediction is required for area A, the first area may be area A, and if reservoir prediction is required for area B, the first area may be area B. Optionally, the first area may be an area where the well logging is located. For example, the well logging is a wellway for extracting natural gas, and the location of the well logging can be determined in advance, and then the geological depth area where the well logging is located can be determined as the first area. For example, in actual application, a three-dimensional geological model including multiple well logging can be constructed, and the cross-section of the well logging can be determined as the first area.

[0081] Optionally, the sample image may be seismic data of the first region. The seismic data may include at least one of the following: seismic data, texture attributes, dip derivative attributes, attenuation attributes, absorption attributes, and ant bodies. Optionally, the seismic data may be information such as the intensity of the earthquake in the first region. For example, the seismic data may be a two-dimensional image, and the seismic data of the first region may be displayed by a seismic intensity image. Optionally, the texture attributes, dip derivative attributes, attenuation attributes, absorption attributes, and ant bodies are all geological attributes with a high degree of correlation with the reservoir. For example, the texture attributes, dip derivative attributes, attenuation attributes, absorption attributes, and ant bodies may all be represented by a two-dimensional image.

[0082] Optionally, seismic data of the first area may be obtained in a database. For example, when the server obtains a seismic intensity image of the first area, the image may be stored in the database, and when the reservoir probability of the first area needs to be predicted, the server may obtain the seismic intensity image in the database.

[0083] Optionally, the texture attribute, dip derivative attribute, attenuation attribute, absorption attribute and ant body of the first area can be obtained according to the seismic data. For example, the seismic data (two-dimensional image) is processed by a preset algorithm (such as a geological analysis algorithm) to obtain a texture attribute image, a dip derivative attribute image, an attenuation attribute image, an absorption attribute image and an ant body image of the first area.

[0084] Next, combine Figure 3 , a sample image is described.

[0085] Figure 3 A schematic diagram of a sample image provided in an embodiment of the present application. Figure 3 , including seismic data images. The seismic data image is an image of a cross-sectional geological surface of a well logging, and the seismic data image includes the well logging. The seismic data image is processed by a preset algorithm to obtain a texture attribute image, a dip derivative attribute image, an attenuation attribute image, an absorption attribute image and an ant body image of the cross-sectional geological surface.

[0086] Optionally, the first label image is used to indicate the probability of the presence of a reservoir in the first partial area of ​​the first area. For example, the first label image may be a distribution image of the reservoir probability of the first partial area, and the probability of the presence of a reservoir at each position in the first partial area may be obtained through the first label image. Optionally, the first partial area may be an area within a preset range of the well logging. For example, the first area may be a cross-sectional geological surface including the well logging, and the first partial area in the first area may be an area within a range of 5 meters around the well logging.

[0087] Optionally, the first label image corresponding to the sample image can be obtained by the following feasible implementation method: the sample image is processed by the second model to obtain the first label sub-image. The second model is obtained by learning multiple groups of first samples, and each group of first samples includes a sample image of the first area and a sample first label sub-image. Multiple groups of first samples can be pre-labeled samples. For example, for sample image 1, the sample first label sub-image 1 corresponding to the sample image 1 is obtained to obtain a group of samples, which includes the sample image 1 and the sample first label sub-image 1. Multiple groups of first samples can be obtained in this way. For example, multiple groups of first samples can be shown in Table 1:

[0088] Table 1

[0089] Multiple groups of first samples Sample Image Sample first label sub-image The first group of samples Sample Image 1 Sample first label sub-image 1 The second group of first sample Sample Image 2 Sample first label sub-image 2 The first sample of the third group Sample Image 3 Sample first label sub-image 3 …… …… ……

[0090] It should be noted that Table 1 is only an example of multiple groups of first samples, and is not a limitation to the multiple groups of first samples.

[0091] For example, if the seismic data input into the second model is sample image 1, the second model outputs the first label sub-image corresponding to the seismic data as sample first label sub-image 1; if the seismic data input into the second model is sample image 2, the second model outputs the first label sub-image corresponding to the seismic data as sample first label sub-image 2; if the seismic data input into the second model is sample image 3, the second model outputs the first label sub-image corresponding to the seismic data as sample first label sub-image 3.

[0092] Optionally, a first mask is added to the first label sub-image to obtain a first label image. The first mask is used to cover the second partial area corresponding to the sample image. For example, the first label sub-image can accurately reflect the probability that a reservoir exists in the first partial area of ​​the first area. Therefore, when training the first model, the second partial area can be covered by the first mask to improve the accuracy of the first model in identifying the reservoir.

[0093] Optionally, the second label image is used to indicate the probability of the existence of a reservoir in the second partial area of ​​the first area. For example, the second label image may be a distribution image of the reservoir probability of the second partial area, and the probability of the existence of a reservoir at each position in the second partial area can be obtained through the second label image. Optionally, the second partial area may be an area outside the preset range of the well logging. For example, the first area may be a cross-sectional geological surface including the well logging, and the second partial area in the first area may be an area outside the 5-meter range around the well logging. Optionally, the first area includes a first partial area and a second partial area. For example, the first partial area may be an area within a preset range around the well logging in the first area, and the second partial area may be an area outside the preset range around the well logging in the first area.

[0094] Optionally, the second label image corresponding to the sample image can be obtained according to the following feasible implementation method: polynomial linear regression processing is performed on the sample image to obtain the second label sub-image. For example, when the seismic data of the first area is obtained, the relationship between the seismic data and the reservoir probability distribution can be fitted by a regression analysis method (such as the least squares method, etc.), thereby obtaining a reservoir probability distribution map of the first area, that is, the second label sub-image.

[0095] A second mask is added to the second label sub-image to obtain a second label image. The second mask is used to mask the first partial area corresponding to the sample image. For example, the accuracy of the second label sub-image obtained by the polynomial linear regression method is unknown, that is, the second label sub-image is a pseudo-label. However, since the first label image can accurately reflect the probability of the existence of a reservoir in the first partial area, it is necessary to add a mask to the first partial area of ​​the second label sub-image to obtain the second label image, so that the first model can learn the reservoir probability of the second partial area in the second label image.

[0096] S202: Process the sample image using the first model to obtain a reservoir probability prediction image of the first area.

[0097] Optionally, the reservoir probability prediction image is used to indicate the probability of the presence of a reservoir in each portion of the first region. For example, the reservoir probability prediction image can be used to accurately determine the portion of the first region where the reservoir exists, thereby facilitating oil and gas exploration.

[0098] Optionally, the first model includes an encoder, a converter, and a decoder, and can obtain a reservoir probability prediction image of the first area through the following feasible implementation method: obtain the features of the sample image through the encoder, and send the features of the sample image to the converter. For example, the encoder can include 3 convolutional layers, and the input sample image can be downsampled through the 3 convolutional layers to obtain the features of the sample image.

[0099] The features of the sample image are converted into feature vectors through the converter, and the feature vectors are sent to the decoder. For example, the converter part may include 9 residual neural network ResNets structures, each ResNets structure includes 2 convolutional layers, and the features of the sample image can be converted into feature vectors through the ResNets structure, while retaining the original features of the sample image.

[0100] The feature vector is converted into partial features corresponding to the sample image by the decoder, and a reservoir probability prediction image is generated based on the partial features. The partial features may be low-level features in the sample image. For example, the decoder includes three deconvolution layers, through which the feature vector can be upsampled, and the low-level features of the sample image can be restored in the feature vector, and the reservoir probability prediction image can be obtained based on the low-level features.

[0101] Next, combine Figure 4 , the structure of the first model is explained.

[0102] Figure 4 This is a schematic diagram of the structure of a first model provided in an embodiment of the present application. Figure 4 , including: a first model. The first model includes an encoder, a converter and a decoder. The encoder includes convolutional layer A, convolutional layer B and convolutional layer C, the converter includes 9 residual neural networks, each residual neural network includes 2 convolutional layers, and the converter includes 18 convolutional layers. The decoder includes deconvolutional layer A, deconvolutional layer B and deconvolutional layer C.

[0103] See also Figure 4When the first model receives a sample image, the encoder in the first model can downsample the sample image to extract features in the sample image, the converter can convert the features in the sample image into feature vectors, and the decoder can upsample the feature vectors to restore low-level features in the sample image, and obtain a reservoir probability prediction image based on the low-level features.

[0104] S203: Update the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image.

[0105] Optionally, the model parameters of the first model may be updated according to the following feasible implementation method: obtaining a first weight of a first loss function between the first label image and the reservoir probability prediction image. The first loss function is used to adjust the model parameters of the first model. Optionally, the first weight is the proportion of the first loss function in adjusting the model parameters. Optionally, the first loss function may be obtained by the following formula:

[0106]

[0107] in, is the first loss function, Y aug is the first label sub-image (two-dimensional augmented logging label), mask aug is the first mask. In practical applications, the first mask is used to determine whether a point belongs to the first area (well logging and well bypass area). Wp (X) is the first model, and Wp is the model parameter of the first model.

[0108] Optional, mask aug The formula is as follows:

[0109]

[0110] Among them, j is the logging depth, the logging depth range is 0 to α, k is the column number corresponding to the logging position, and its range is r-β to r+β. The total number of well bypass channels is 2β. Optionally, β can be 20.

[0111] Obtain a second weight of a second loss function between the second label image and the reservoir probability prediction image. The second loss function is used to adjust the model parameters of the first model. Optionally, the second weight is the proportion of the second loss function in adjusting the model parameters. Optionally, the second loss function can be obtained by the following formula:

[0112]

[0113] in, is the second loss function, Y pvugis the second label sub-image, mask pvug is the second mask. In practical applications, the second mask is used to determine whether a point belongs to the second area (the area outside the well logging and well bypass). Wp (X) is the first model, and Wp is the model parameter of the first model.

[0114] Optional, mask pvug The formula is as follows:

[0115]

[0116] Among them, j is the logging depth, the logging depth range is 0 to α, k is the column number corresponding to the logging position, and its range is r-β to r+β. The total number of well bypass channels is 2β. Optionally, β can be 20.

[0117] The model parameters of the first model are updated according to the first loss function, the second loss function, the first weight, and the second weight. For example, the loss function of the first model as a whole is obtained through the first weight, the second weight, the first loss function, and the second loss function, and then the model parameters of the first model are updated through the overall loss function. For example, a first product of the first weight and the first loss function is obtained, a second product of the second weight and the second loss function is obtained, and the sum of the first product and the second product is determined as the loss function of the first model as a whole.

[0118] Next, combine Figure 5 , the process of updating the model parameters of the first model is explained.

[0119] Figure 5 A schematic diagram of updating the model parameters of the first model provided in an embodiment of the present application. Figure 5 , including a first model. A sample image is input into the first model, and the first model obtains a reservoir probability prediction image corresponding to the sample image. The first model obtains a first loss function through the first label image and the reservoir probability prediction image, and the first model obtains a second loss function through the second label image and the reservoir probability prediction image. The first model updates the model parameters of the first model through the first loss function and the second loss function. When updating the model parameters, the complete loss function of the first model can be obtained through the first weight corresponding to the first loss function and the second weight corresponding to the second loss function, thereby updating the model parameters.

[0120] The embodiment of the present application provides a model processing method for reservoir prediction based on a closed-loop network, wherein sample data is obtained, the sample data includes a sample image of a first region, a first label image and a second label image corresponding to the sample image, the first label image is used to indicate the probability that a reservoir exists in a first partial region of the first region, and the second label image is used to indicate the probability that a reservoir exists in a second partial region of the first region, the features of the sample image are obtained through the encoder of the first model, and the features of the sample image are sent to the converter of the first model, the features of the sample image are received through the converter, and the features of the sample image are converted into feature vectors, the feature vectors are sent to the decoder of the first model, the feature vectors are received through the decoder, the feature vectors are converted into partial features corresponding to the sample image, and a reservoir probability prediction image is generated according to the partial features, and the model parameters of the first model are updated according to the first label image, the second label image and the reservoir probability prediction image. In this way, since the training sample data of the first model includes a two-dimensional label image, and the first label image and the second label image can accurately reflect the probability that a reservoir exists in different partial regions under the first region, the label accuracy of the first model is high, and the label is a two-dimensional image label, which can accurately reflect the spatial continuity between reservoirs, thereby improving the accuracy of the first model in identifying reservoirs.

[0121] exist Figure 2 Based on the embodiment shown below, combined with Figure 6 , the above-mentioned closed-loop network-based reservoir prediction model processing method is further explained.

[0122] Figure 6 A flow chart of another closed-loop network-based reservoir prediction model processing method provided in an embodiment of the present application. Figure 6 The method flow includes:

[0123] S601: Obtain sample data.

[0124] Optionally, the sample data includes a sample image of the first area, a first label image corresponding to the sample image, and a second label image, wherein the first label image is used to indicate the probability of the existence of a reservoir in a first partial area in the first area, and the second label image is used to indicate the probability of the existence of a reservoir in a second partial area in the first area.

[0125] Optionally, the second model can be trained by the sample image and the sample first label sub-image to obtain the second model, and the second model can also be obtained by iterative training of the model. Optionally, the second model can be obtained by the following feasible implementation method: further training the reservoir probability recognition model to obtain the second model. Among them, the reservoir probability recognition model is a model obtained by training the sample image and the sample reservoir probability prediction image. For example, the training samples of the reservoir probability recognition model are the sample image and the manually annotated reservoir probability prediction image. Since the accuracy of the manually annotated reservoir probability prediction image is low, the reservoir probability recognition model can be further trained to improve the accuracy of reservoir recognition. Optionally, when further training the reservoir probability identification model to obtain the second model, a one-dimensional well logging label that crosses the geological surface can be obtained, wherein the one-dimensional well logging label can be determined by existing seismic data. Therefore, the one-dimensional well logging label can accurately reflect the probability of the existence of a reservoir in a preset area around the well logging. Since the reservoir probability identification model is a two-dimensional training model, a mask can be added to the surface of the one-dimensional well logging label to obtain a two-dimensional well logging label, wherein only the probability of the reservoir in the preset area around the well logging is accurate (the one-dimensional label is accurate) in the two-dimensional well logging label. Then, the reservoir probability identification model is further trained through sample images and two-dimensional well logging labels (one-dimensional well logging labels with masks added). After the training is completed, the second model is obtained. Next, combined with Figure 7 , the process of obtaining the reservoir probability identification model is explained.

[0126] Figure 7 A schematic diagram of a process for obtaining a reservoir probability identification model provided in an embodiment of the present application. Figure 7 , including a reservoir probability recognition model. A sample image is input into the reservoir probability recognition model, and the reservoir probability recognition model can output a reservoir probability prediction image, but the accuracy of the reservoir probability prediction image is low. The loss function of the reservoir probability recognition model is determined by the pseudo-label and the reservoir probability prediction image, and then the model parameters of the reservoir probability recognition model are updated by the loss function. Although the accuracy of the pseudo-label is also low, the reservoir probability recognition model can process the two-dimensional image through the above training method. Optionally, the loss function of the reservoir probability recognition model can be determined by the following formula:

[0127]

[0128] Among them, L open is the loss function of the reservoir probability identification model; Y pvug is a pseudo label; f Wp (X) is the reservoir probability identification model, and Wp is the model parameter of the reservoir probability identification model.

[0129] Next, combine Figure 8 , the process of training the second model is explained.

[0130] Figure 8 A schematic diagram of the second model training process provided in the embodiment of the present application. Figure 8 , including: a second model. The second model is a trained reservoir probability recognition model. A sample image is input into the second model, and the second model outputs a first label sub-image (a reservoir probability prediction image output by the reservoir recognition model). A mask is added to the surface of the one-dimensional logging label to obtain a two-dimensional logging label, and a loss function is determined by the first label sub-image and the two-dimensional logging label. The model parameters in the second model are updated by the loss function so that the second model can output an accurate first label sub-image.

[0131] Optionally, the loss function of the second model can be determined by the following formula:

[0132]

[0133] Among them, L open is the loss function of the second model; Y well is the one-dimensional logging label; f Wp (X) is the second model, Wp is the model parameter of the second model; ⊙ is the dot product; mask well is the mask corresponding to the one-dimensional well logging label.

[0134] The mask formula can be:

[0135]

[0136] Among them, mask well is the mask corresponding to the one-dimensional well logging label, j is the depth of the well logging, the depth range of the well logging is 0 to α, k is the column number corresponding to the well logging position, and r is the column number corresponding to the well logging.

[0137] Optionally, when obtaining the sample first label sub-image, each set of seismic data can train a reservoir probability recognition model, and then multiple sample first label sub-images corresponding to multiple seismic data can be obtained by multi-model processing. For example, if 79 one-dimensional well logging labels are determined by seismic data, 79 second models are trained respectively.

[0138] S602: Process the sample image using the first model to obtain a reservoir probability prediction image of the first area.

[0139] It should be noted that the execution process of step S602 can refer to step S202, and this embodiment of the present application will not be described in detail.

[0140] S603: Update the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image.

[0141] It should be noted that the execution process of step S603 can refer to step S203, and this embodiment of the present application will not be further described.

[0142] S604: Process the reservoir probability prediction image using the third model to obtain a target sample image corresponding to the reservoir probability prediction image.

[0143] Optionally, the third model is obtained by learning multiple groups of second samples. Each group of second samples includes a sample reservoir probability prediction image and a sample image of the first area. Multiple groups of second samples can be pre-labeled samples. For example, for sample reservoir probability prediction image 1, sample image 1 corresponding to sample reservoir probability prediction image 1 is obtained to obtain a group of samples, which includes sample reservoir probability prediction image 1 and sample image. Multiple groups of second samples can be obtained in this way. For example, multiple groups of second samples can be shown in Table 2:

[0144] Table 2

[0145] Multiple second samples Sample reservoir probability prediction image Sample Image The first group, the second sample Sample reservoir probability prediction image 1 Sample Image 1 The second group of samples Sample reservoir probability prediction image 2 Sample Image 2 The second sample of the third group Sample reservoir probability prediction image 3 Sample Image 3 …… …… ……

[0146] It should be noted that Table 2 is only an example of multiple groups of second samples, and is not a limitation on multiple groups of first samples.

[0147] For example, if the image input into the third model is sample reservoir probability prediction image 1, the third model outputs sample image 1 corresponding to the sample reservoir probability prediction image 1; if the image input into the third model is sample reservoir probability prediction image 2, the third model outputs sample image 2 corresponding to the sample reservoir probability prediction image 1; if the image input into the third model is sample reservoir probability prediction image 3, the third model outputs sample image 3 corresponding to the sample reservoir probability prediction image 1.

[0148] S605: Update the model parameters of the first model according to the sample image and the target sample image.

[0149] Optionally, the reservoir probability image is input into the third model, and the third model can generate a target sample image corresponding to the reservoir probability image. Optionally, the target sample image can be seismic data corresponding to the reservoir probability image. For example, the target sample image can include a seismic data image, a texture attribute image, a dip derivative attribute image, an attenuation attribute image, an absorption attribute image, and an ant body image corresponding to the reservoir probability image.

[0150] Through the sample image and the target sample image, the third loss function can be determined, and then the model parameters of the first model can be updated through the third loss function. For example, through the sample image and the target sample image, the third loss function corresponding to the first model can be obtained, and then according to the third weight of the third loss function in adjusting the model parameters, the complete loss function of the first model is updated, and then the model parameters of the first model are updated through the updated complete loss function. In this way, the better the cycle consistency, the higher the accuracy of the reservoir probability prediction image output by the first model.

[0151] Optionally, when updating the model parameters of the first model, the model parameters of the first model can also be updated through the first label image and the second label image. For example, the first label image and the second label image are input into the third model, and the third model outputs the corresponding first sample image, and then the fifth loss function and the sixth loss function, as well as the fifth weight of the fifth loss function and the sixth weight of the sixth loss function are determined according to the sample image and the first sample image, and then the complete loss function of the first model is updated according to the fifth weight and the fifth loss function, and the model parameters of the first model are updated.

[0152] Optionally, after obtaining the first sample image through the third model, the first sample image can also be processed through the first model to obtain a target reservoir probability prediction image, and then the first model parameters are updated according to the target reservoir probability image, the first label image and the second label image. For example, after processing the first sample image through the first model, the target reservoir probability prediction image can be obtained, and the seventh loss function and the seventh weight corresponding to the seventh loss function can be obtained through the target reservoir probability prediction image and the first label image, and the eighth loss function and the eighth weight can be obtained through the target reservoir probability prediction image and the second label image, and then the complete loss function of the first model is updated through the seventh loss function, the seventh weight, the eighth loss function and the eighth weight, and then the model parameters of the first model are updated.

[0153] Next, combine Fig. 9 , the process of updating the model parameters of the first model is explained.

[0154] Fig. 9 This is another schematic diagram of a process for updating the model parameters of the first model provided in an embodiment of the present application. Fig. 9 , including the first model and the third model. Where X is the sample image, X cycle is the target sample image, is the reservoir probability prediction image, Y aug is the first label sub-image, mask aug is the first mask, Y pvugis the second label sub-image, mask pvug The sample image is input to the first model, and the first model outputs a reservoir probability prediction image.

[0155] The first model obtains a first label image by synthesizing the first label sub-image and the first mask, and obtains a second label image by synthesizing the second label sub-image and the second mask. The first model determines a first loss function through the first label image and the reservoir probability prediction image. The first model determines a second loss function through the second label image and the reservoir probability prediction image.

[0156] See also Fig. 9 , the reservoir probability prediction image, the first label image, and the composite image of the second label image are input to the third model, and the third model can output a target sample image, and the composite image of the reservoir probability prediction image and the second label image is input to the third model, and the third model can output another target sample image. The third loss function and the fourth loss function of the first model can be determined by the two target sample images and the sample image.

[0157] See also Fig. 9 , the first label image and the second label image with mask are simultaneously input into the third model to obtain the corresponding two target sample images The fifth loss function and the sixth loss function can be determined by the sample image, the two target sample images, the first mask and the second mask. The two target sample images are input into the first model again to obtain two reservoir probability prediction images Y after one cycle. cycle , and then through one of the Y cycle , Y aug And the first mask, get the seventh loss function, through another Y cycle , Y pvug and the second mask to obtain the eighth loss function. Then, the overall loss function of the first model is determined through the first loss function, the second loss function, the third loss function, the fourth loss function, the fifth loss function, the sixth loss function, the seventh loss function, the eighth loss function, and the weights corresponding to each loss function, and the model parameters of the first model are updated through the overall loss function.

[0158] Optionally, the formula for the third loss function is as follows:

[0159]

[0160] in, is the third loss function, Y aug is the first label sub-image, mask aug is the first mask, Y pvug is the second label sub-image, maskpvuh is the second mask, f Wp (X) is the first model, Wp is the model parameter of the first model, f WG is the third model, and WG is the model parameter of the third model.

[0161] The formula of the fourth loss function is as follows:

[0162]

[0163] in, is the fourth loss function, Y pvug is the second label sub-image, mask pvug is the second mask, f Wp (X * ) is the first model, X * is the target sample image, Wp is the model parameter of the first model, f WG is the third model, and WG is the model parameter of the third model.

[0164] The formula of the fifth loss function is as follows:

[0165]

[0166] in, is the fifth loss function, Y aug is the first label sub-image, X is the sample image, mask aug is the first mask, f WG For the third model.

[0167] The formula of the sixth loss function is as follows:

[0168]

[0169] in, is the sixth loss function, Y pvug is the second label sub-image, X is the sample image, mask pvug is the second mask, f WG For the third model.

[0170] The formula of the seventh loss function is as follows:

[0171]

[0172] in, is the seventh loss function, Y aug is the first label sub-image, mask aug is the first mask, f WF is the first model of the loop input, f WG For the third model.

[0173] The formula of the eighth loss function is as follows:

[0174]

[0175] in, is the eighth loss function, Y pvug is the second label sub-image, mask pvug is the second mask, f WF is the first model of the loop input, f WG For the third model.

[0176] Optionally, the overall loss function of the first model can be expressed as follows:

[0177]

[0178] Among them, L close is the overall loss function of the first model, and λ1, λ2, and λ3 are weights used to adjust the eight loss functions.

[0179] The embodiment of the present application provides a model processing method for reservoir prediction based on a closed-loop network, which obtains sample data, processes the sample image through a first model to obtain a reservoir probability prediction image of a first region, updates the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image, processes the reservoir probability prediction image through a third model to obtain a target sample image corresponding to the reservoir probability prediction image, and updates the model parameters of the first model according to the sample image and the target sample image. In this way, since the training sample data of the first model includes a two-dimensional label image, and the first label image and the second label image can accurately reflect the probability that a reservoir exists in different partial regions under the first region, the label accuracy of the first model is relatively high, and the label is a two-dimensional image label, which can accurately reflect the spatial continuity between reservoirs, and the model parameters of the first model can be updated through a closed-loop network, thereby improving the accuracy of the first model in identifying reservoirs.

[0180] Based on any of the above embodiments, Fig.10 , the process of performing model processing on the first model is explained.

[0181] Fig.10 A schematic diagram of a flow chart of a model processing method for reservoir prediction based on a closed-loop network provided in an embodiment of the present application. Fig.10 The method flow includes:

[0182] S1001. Acquire at least one reservoir image set in a first area.

[0183] Optionally, the reservoir image set is Figure 2 Sample images in the illustrated embodiment. For example, the reservoir image set may include seismic data images, texture attribute images, dip derivative attribute images, attenuation attribute images, absorption attribute images, and ant body images of the first region.

[0184] S1002. Process at least one reservoir image set according to a first model to obtain a reservoir probability prediction image of a first area.

[0185] Optionally, the first model is the first model in any one of the above embodiments. A reservoir image set of a first region is input into the first model, and the first model can output a reservoir probability prediction image corresponding to the first region.

[0186] Next, combine Fig.11 , the reservoir probability prediction image is explained.

[0187] Fig.11 This is a schematic diagram of a reservoir probability prediction image provided in the embodiment of this application. Fig.11 , including a first image and a second image. Among them, the first image is a reservoir probability prediction image obtained according to the prior art, and the second image is a reservoir probability prediction image obtained according to the first model in the present application. The position indicated by the arrow is the location of the well logging. In the first image, due to the use of one-dimensional labels as training samples of the model, the spatial continuity between the obtained reservoirs is poor, which leads to a low accuracy of reservoir prediction. In the second image obtained by the first model of the present application, the spatial continuity between the reservoirs is good, and the accuracy of reservoir prediction is high.

[0188] The embodiment of the present application provides a model processing method for reservoir prediction based on a closed-loop network, obtaining at least one reservoir image set of a first area, processing at least one reservoir image set according to a first model, and obtaining a reservoir probability prediction image of the first area. The first model is the first model in any of the above embodiments. Since the first model can accurately determine the probability of the existence of a reservoir in the first area, the accuracy of reservoir prediction can be improved by the first model.

[0189] Fig.12 A schematic diagram of the structure of a closed-loop network-based reservoir prediction model processing device provided in an embodiment of the present application. Fig.12 The model processing device 10 for reservoir prediction based on a closed-loop network includes an acquisition module 11, a processing module 12 and an update module 13, wherein:

[0190] The acquisition module 11 is used to acquire sample data, the sample data including a sample image of a first area, a first label image and a second label image corresponding to the sample image, the first label image is used to indicate the probability that a reservoir exists in a first partial area of ​​the first area, and the second label image is used to indicate the probability that a reservoir exists in a second partial area of ​​the first area;

[0191] The processing module 12 is used to process the sample image through the first model to obtain a reservoir probability prediction image of the first area;

[0192] The updating module 13 is used to update the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image.

[0193] In a possible implementation manner, the acquisition module 11 is specifically used for:

[0194] Processing the sample image by a second model to obtain a first label sub-image, wherein the second model is obtained by learning multiple groups of first samples, each group of first samples including a sample image of a first region and a sample label sub-image;

[0195] A first mask is added to the first label sub-image to obtain the first label image, wherein the first mask is used to cover the second partial area corresponding to the sample image.

[0196] In a possible implementation manner, the acquisition module 11 is specifically used for:

[0197] Performing polynomial linear regression processing on the sample image to obtain a second label sub-image;

[0198] A second mask is added to the second label sub-image to obtain the second label image, where the second mask is used to cover the first partial area corresponding to the sample image.

[0199] In a possible implementation manner, the processing module 12 is specifically configured to:

[0200] Acquire the features of the sample image through the encoder, and send the features of the sample image to the converter;

[0201] Converting the features of the sample image into a feature vector by the converter, and sending the feature vector to the decoder;

[0202] The decoder converts the feature vector into partial features corresponding to the sample image, and generates the reservoir probability prediction image according to the partial features.

[0203] In a possible implementation manner, the updating module 13 is specifically configured to:

[0204] Obtaining a first weight of a first loss function between the first label image and the reservoir probability prediction image;

[0205] Obtaining a second weight of a second loss function between the second label image and the reservoir probability prediction image;

[0206] Update the model parameters of the first model according to the first loss function, the second loss function, the first weight, and the second weight.

[0207] In a possible implementation manner, the processing module 12 is further configured to:

[0208] Processing the reservoir probability prediction image by a third model to obtain a target sample image corresponding to the reservoir probability prediction image, wherein the third model is obtained by learning a plurality of groups of second samples, each group of second samples including a sample reservoir probability prediction image and a sample image of the first area;

[0209] The model parameters of the first model are updated according to the sample image and the target sample image.

[0210] The closed-loop network-based reservoir prediction model processing device provided in the embodiment of the present application can execute the technical solution shown in the above method embodiment. Its implementation principle and beneficial effects are similar and will not be repeated here.

[0211] The present application also includes another structure of a model processing device for reservoir prediction based on a closed-loop network, the model processing device for reservoir prediction based on a closed-loop network includes an acquisition module and a processing module, wherein:

[0212] The acquisition module is used to acquire at least one reservoir image set of the first area;

[0213] The processing module is used to process the at least one reservoir image set according to a first model to obtain a reservoir probability prediction image of the first area, and the first model is the first model described in any one of claims 1-6.

[0214] Fig.13 The schematic diagram of the terminal device provided for this application. Fig.13The terminal device 20 may include: a processor 21 and a memory 22, wherein the processor 21 and the memory 22 can communicate; illustratively, the processor 21 and the memory 22 communicate via a communication bus 23, the memory 22 is used to store program instructions, and the processor 21 is used to call the program instructions in the memory to execute the model processing method for reservoir prediction based on a closed-loop network as shown in any of the above method embodiments.

[0215] Optionally, the terminal device 20 may further include a communication interface, which may include a transmitter and / or a receiver.

[0216] Optionally, the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the present application may be directly implemented as being executed by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor.

[0217] The present application provides a readable storage medium having a computer program stored thereon; the computer program is used to implement a model processing method for reservoir prediction based on a closed-loop network as described in any of the above embodiments.

[0218] An embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed, a computer executes the above-mentioned closed-loop network-based model processing method for reservoir prediction.

[0219] All or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a readable memory. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned memory (storage medium) includes: read-only memory (English: read-only memory, abbreviated: ROM), RAM, flash memory, hard disk, solid state drive, magnetic tape (English: magnetic tape), floppy disk (English: floppy disk), optical disc (English: optical disc) and any combination thereof.

[0220] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processing unit of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0221] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0222] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0223] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0224] In the present application, the term "include" and its variations may refer to non-restrictive inclusion; the term "or" and its variations may refer to "and / or". The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. In the present application, "plurality" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previously associated objects are in an "or" relationship.

Claims

1. A model processing method for reservoir prediction based on a closed-loop network, characterized in that: include: Acquire sample data, the sample data comprising a sample image of a first area, a first label image and a second label image corresponding to the sample image, the first label image being used to indicate a probability that a reservoir exists in a first partial area of ​​the first area, the second label image being used to indicate a probability that a reservoir exists in a second partial area of ​​the first area, the sample image comprising: seismic data, the seismic data being a two-dimensional image, and the sample data comprising a two-dimensional label image; Processing the sample image by using a first model to obtain a reservoir probability prediction image of the first area; updating the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image; The first model includes an encoder, a converter and a decoder; the sample image is processed by the first model to obtain a reservoir probability prediction image of the first area, including: Acquire the features of the sample image through the encoder, and send the features of the sample image to the converter; Converting the features of the sample image into a feature vector by the converter, and sending the feature vector to the decoder; The decoder converts the feature vector into partial features corresponding to the sample image, and generates the reservoir probability prediction image according to the partial features; The updating of the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image comprises: Obtaining a first weight of a first loss function between the first label image and the reservoir probability prediction image, wherein the first loss function is obtained by the first model through the first label image and the reservoir probability prediction image; Acquire a second weight of a second loss function between the second label image and the reservoir probability prediction image, where the second loss function is obtained by the first model through the second label image and the reservoir probability prediction image; Update the model parameters of the first model according to the first loss function, the second loss function, the first weight, and the second weight.

2. The method according to claim 1, characterized in that Acquiring a first label image corresponding to the sample image includes: Processing the sample image by a second model to obtain a first label sub-image, wherein the second model is obtained by learning multiple groups of first samples, each group of first samples including a sample image of a first region and a sample label sub-image; A first mask is added to the first label sub-image to obtain the first label image, wherein the first mask is used to cover the second partial area corresponding to the sample image.

3. The method according to claim 1 or 2, characterized in that: Acquiring a second label image corresponding to the sample image includes: Performing polynomial linear regression processing on the sample image to obtain a second label sub-image; A second mask is added to the second label sub-image to obtain the second label image, where the second mask is used to cover the first partial area corresponding to the sample image.

4. The method according to any one of claims 1 to 3, characterized in that: After updating the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image, the method further includes: Processing the reservoir probability prediction image by a third model to obtain a target sample image corresponding to the reservoir probability prediction image, wherein the third model is obtained by learning a plurality of groups of second samples, each group of second samples including a sample reservoir probability prediction image and a sample image of the first area; The model parameters of the first model are updated according to the sample image and the target sample image.

5. A model processing method for reservoir prediction based on a closed-loop network, characterized in that: include: acquiring at least one reservoir image set of a first region; The at least one reservoir image set is processed according to the first model to obtain a reservoir probability prediction image of the first area, and the first model is the first model in the model processing method for reservoir prediction based on a closed-loop network as described in any one of claims 1-4.

6. A model processing device for reservoir prediction based on a closed-loop network, characterized in that: It includes acquisition module, processing module and update module, among which: The acquisition module is used to acquire sample data, the sample data including a sample image of a first area, a first label image and a second label image corresponding to the sample image, the first label image is used to indicate the probability that a reservoir exists in a first partial area of ​​the first area, and the second label image is used to indicate the probability that a reservoir exists in a second partial area of ​​the first area, the sample image includes: seismic data, the seismic data is a two-dimensional image, and the sample data includes a two-dimensional label image; The processing module is used to process the sample image through a first model to obtain a reservoir probability prediction image of the first area; The updating module is used to update the model parameters of the first model according to the first label image, the second label image, and the reservoir probability prediction image; The first model includes an encoder, a converter and a decoder; the processing module is specifically used to obtain the features of the sample image through the encoder and send the features of the sample image to the converter; Converting the features of the sample image into a feature vector by the converter, and sending the feature vector to the decoder; The decoder converts the feature vector into partial features corresponding to the sample image, and generates the reservoir probability prediction image according to the partial features; The updating module is specifically used to obtain a first weight of a first loss function between the first label image and the reservoir probability prediction image, where the first loss function is obtained by the first model through the first label image and the reservoir probability prediction image; Acquire a second weight of a second loss function between the second label image and the reservoir probability prediction image, where the second loss function is obtained by the first model through the second label image and the reservoir probability prediction image; Update the model parameters of the first model according to the first loss function, the second loss function, the first weight, and the second weight.

7. A terminal device, characterized in that: include: a processor coupled to the memory; The memory is used to store computer programs; The processor is used to execute the computer program stored in the memory so that the terminal device executes the model processing method for reservoir prediction based on a closed-loop network as described in any one of claims 1 to 4 above, or the model processing method for reservoir prediction based on a closed-loop network as described in claim 5.

8. A readable storage medium, characterized in that: It comprises a program or an instruction, which, when executed on a computer, is a model processing method for reservoir prediction based on a closed-loop network as described in any one of claims 1 to 4 above, or a model processing method for reservoir prediction based on a closed-loop network as described in claim 5.

Citation Information

Patent Citations

  • A simulation-to-seismic workflow construed from core based rock typing and enhanced by rock replacement modeling

    CN105593703A

  • Wave impedance inversion method using neural network and neural network system

    CN113109874A

  • Method for Dynamic Calibration and Simultaneous Closed-Loop Inversion of Simulation Models of Fractured Reservoirs

    US20200292722A1