Methods and related equipment for inverting the pore structure of carbonate rocks

Abstract

This invention provides a method and related equipment for inverting the pore structure of carbonate rocks, relating to the field of geological exploration technology. The method includes acquiring test data, and using a trained deep learning prediction model with a Bayesian estimation algorithm to invert the pore structure based on the test data, thereby obtaining pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer. According to the above technical solution, test data is input into the trained deep learning prediction model, and the Bayesian estimation algorithm is used to invert the pore structure, obtaining pore structure prediction results, including confidence intervals for pore structure parameters. This not only quantifies the uncertainty of the prediction results, thereby expressing the uncertainty and obtaining confidence intervals, but also effectively improves prediction accuracy, i.e., improves the accuracy of the pore structure prediction results. It has great application potential in the field of data-driven pore structure prediction, facilitating better decision support.

Description

Technical Field

[0001] This invention relates to the field of geological exploration technology, and more specifically, to a method for inverting the pore structure of carbonate rocks, a device for inverting the pore structure of carbonate rocks, an electronic device, and a storage medium. Background Technology

[0002] Seismic inversion is a powerful tool for studying the Earth's internal structure and describing reservoirs. Pore structure is a crucial parameter for evaluating carbonate reservoirs and a significant aspect of carbonate reservoir prediction. Seismic inversion methods are generally classified into two main categories: deterministic methods, including generalized linear inversion and sparse pulse inversion, and statistical inversion methods, which statistically sample the posterior distribution based on a Bayesian theoretical framework. Due to the band-limited nature of seismic data, seismic inversion suffers from non-uniqueness and instability. Both deterministic and statistical methods require prior information to regularize the inversion process; therefore, prior information plays a vital role in obtaining the optimal inversion solution. For mathematical convenience, simple prior assumptions are often made about the properties of the inversion solution, such as sparsity or minimum length assumptions. However, these prior assumptions can lead to overly smooth inversion results, limiting their effectiveness in resolving thin reservoirs.

[0003] Deep learning has been a hot research topic in recent years. In the fields of computer vision and pattern recognition, it is a technology that imitates the human brain's perceptual process. It adopts a multi-layered structure and imitates the hierarchical transmission characteristics of the human brain's nervous system. It has received widespread attention and application in the fields of computer vision and pattern recognition and has achieved good results.

[0004] However, traditional deep neural networks can only provide point estimates of pore structure predictions, which may lead to prediction bias and losses in practical applications.

[0005] Therefore, a new technical solution is urgently needed to solve the above-mentioned technical problems. Summary of the Invention

[0006] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. The summary section of this invention is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0007] In a first aspect, the present invention proposes a method for inverting the pore structure of carbonate rocks, comprising:

[0008] Obtain test data;

[0009] For the test data, the pore structure is inverted using a Bayesian estimation algorithm with a trained deep learning prediction model to obtain the pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer.

[0010] Optionally, the method further includes:

[0011] A labeling technique was constructed using rock physics analysis and forward modeling. A label training dataset was generated based on the original data. The label training dataset included seismic data and pore structure parameters.

[0012] Deep learning prediction models are trained based on labeled training datasets.

[0013] Optionally, a labeled training dataset is generated based on the original data, including:

[0014] Simulate the sedimentary environment of carbonate rocks;

[0015] Obtain the lithofacies and rock physical property parameters of the target reservoir;

[0016] Seismic records are generated based on lithofacies and rock physical property parameters, where there is a one-to-one correspondence between seismic data, pore structure parameters, and seismic records in the labeled training dataset.

[0017] Optionally, the method further includes:

[0018] Calculate the posterior distribution of the weights of a Bayesian convolutional neural network layer.

[0019] Optionally, the posterior distribution of the weights of the Bayesian convolutional neural network layer is calculated, including:

[0020] Variational inference is used to approximate the posterior distribution with the first distribution controlled by the first parameter in order to obtain the objective function;

[0021] Based on the objective function and the labeled training dataset, the loss function of the Bayesian convolutional neural network layer is obtained.

[0022] Obtain weighted sample values ​​to optimize the first parameter using the loss function and the weighted sample values.

[0023] Optionally, the weight sample values ​​are obtained, including:

[0024] Sample from each Gaussian distribution of the first parameter to obtain weighted sample values;

[0025] The method also includes:

[0026] After multiple samplings, multiple pore structure prediction results were obtained;

[0027] The final prediction result is obtained by averaging the prediction results of multiple pore structures.

[0028] Optionally, the deep learning prediction model may also include:

[0029] The system consists of fully connected layers, pooling layers, long short-term memory networks, and convolutional neural networks. The Bayesian convolutional neural network layer is connected to the pooling layer and then connected to the long short-term memory network and the convolutional neural network via the fully connected layer.

[0030] The method also includes:

[0031] Choose the appropriate sliding window size based on the size of the test data.

[0032] Secondly, a device for inverting the pore structure of carbonate rocks is also proposed, including:

[0033] The acquisition module is used to acquire test data;

[0034] The inversion module is used to perform pore structure inversion on test data using a trained deep learning prediction model and a Bayesian estimation algorithm to obtain pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer.

[0035] Thirdly, an electronic device is also proposed, including a processor and a memory, wherein the memory stores computer program instructions, which are executed by the processor to perform the carbonate rock pore structure inversion method as described above.

[0036] Fourthly, a storage medium is also proposed, on which program instructions are stored. When the program instructions are run, they are used to execute the carbonate rock pore structure inversion method as described above.

[0037] According to the above technical solution, test data is input into a trained deep learning prediction model, and a Bayesian estimation algorithm is used to invert the pore structure to obtain pore structure prediction results, including confidence intervals for pore structure parameters. This not only quantifies the uncertainty of the prediction results, thereby expressing the uncertainty and obtaining confidence intervals, but also effectively improves prediction accuracy, i.e., enhances the precision of pore structure prediction results. It has great application potential in the field of data-driven pore structure prediction, facilitating better decision support.

[0038] The method for inverting the pore structure of carbonate rocks of the present invention, and other advantages, objectives and features of the present invention will be partly apparent from the following description, and partly understood by those skilled in the art through study and practice of the present invention. Attached Figure Description

[0039] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0040] Figure 1 A schematic flowchart of a method for inverting the pore structure of carbonate rocks according to an embodiment of the present invention is shown;

[0041] Figure 2 A schematic diagram illustrating the probability distribution of predicted pore structures according to an embodiment of the present invention is shown.

[0042] Figure 3 A schematic diagram of pore structure prediction results according to an embodiment of the present invention is shown;

[0043] Figure 4 A schematic diagram of the structure of a deep learning prediction model according to an embodiment of the present invention is shown;

[0044] Figure 5 A schematic diagram is shown of the pore structure prediction results obtained by a deep learning prediction model including a Bayesian convolutional neural network layer or a deep learning prediction model without a Bayesian convolutional neural network layer according to an embodiment of the present invention.

[0045] Figure 6 A schematic diagram illustrating the generation of a labeled training dataset using a labeling technique based on rock physical analysis and forward modeling, according to an embodiment of the present invention, is shown.

[0046] Figure 7 A schematic block diagram of a carbonate rock pore structure inversion apparatus according to an embodiment of the present invention is shown; and

[0047] Figure 8 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0048] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application 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 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 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. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0049] According to a first aspect of the present invention, the present invention proposes a method for inverting the pore structure of carbonate rocks. Figure 1 A schematic flowchart of a method 100 for inverting the pore structure of carbonate rocks according to an embodiment of the present invention is shown. Method 100 may include the following steps.

[0050] Step S110: Obtain test data.

[0051] For example, seismic acquisition and development in carbonate rock areas can yield a large amount of test data, specifically including seismic data and drilling and logging information. Pore structure is a crucial parameter for reservoir evaluation in carbonate rock formations. Generally, pore structure parameters can be predicted on a single well.

[0052] Step S130: For the test data, the trained deep learning prediction model is used to perform pore structure inversion using a Bayesian estimation algorithm to obtain pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer.

[0053] For example, test data can be used as input to a pre-trained deep learning prediction model to predict pore structures using a Bayesian estimation algorithm. This deep learning prediction model includes a Bayesian convolutional neural network layer. For instance, the deep learning prediction model can be a traditional deep learning neural network with an added Bayesian convolutional neural network layer. It is understood that after introducing the Bayesian convolutional neural network layer, the model parameters can be replaced by probability distributions instead of specific numerical values. Therefore, by training the deep learning prediction model, the probability distribution of pore structure predictions can be output. Optionally, the probability distribution obtained during training can be used as a prior distribution. After inputting test data into the trained deep learning prediction model, a Bayesian estimation algorithm can be used, combined with the aforementioned prior distribution, to calculate the confidence interval of the pore structure parameters, thereby obtaining the posterior distribution. Figure 2 A schematic diagram illustrating the probability distribution of predicted pore structure according to an embodiment of the present invention is shown. Figure 2 As shown, the horizontal axis represents the confidence interval, and the vertical axis represents the frequency of occurrence within the confidence interval. "True" indicates the actual pore structure distribution, and "BCNN" indicates the probability distribution of the predicted pore structure. The probability distribution obtained using the trained deep learning prediction model generally follows a normal distribution. Compared to traditional neural network models, its mean is close to the point estimation results predicted by traditional models, indicating that the prediction accuracy is not affected. Furthermore, it adds the function of quantifying model uncertainty through probability distribution, i.e., obtaining the confidence interval. Figure 3 A schematic diagram of pore structure prediction results according to an embodiment of the present invention is shown. The horizontal axis represents time, and the vertical axis represents pore structure parameters. The solid line corresponding to "True" represents the actual pore structure value, and the dashed line corresponding to "BCNN" represents the predicted pore structure value. It is easy to see that the matching degree between the two is relatively high. Figure 3 The gray shaded area represents the 95% confidence interval. In conjunction with the preceding text, compared to traditional prediction models, using a deep learning prediction model with an added Bayesian convolutional neural network layer for pore structure prediction can additionally obtain a 95% confidence interval for the prediction results.

[0054] According to the above technical solution, test data is input into a trained deep learning prediction model, and a Bayesian estimation algorithm is used to invert the pore structure to obtain pore structure prediction results, including confidence intervals for pore structure parameters. This not only quantifies the uncertainty of the prediction results, thereby expressing the uncertainty and obtaining confidence intervals, but also effectively improves prediction accuracy, i.e., enhances the precision of pore structure prediction results. It has great application potential in the field of data-driven pore structure prediction, facilitating better decision support.

[0055] Optionally, the deep learning prediction model may also include: a fully connected layer (Dense Layer), a pooling layer, a Long Short-Term Memory (LSTM) network, and a Convolutional Neural Network (CNN), wherein the Bayesian convolutional neural network layer is connected to the pooling layer, and is connected to the LSTM network and the convolutional neural network via the fully connected layer. Figure 4 A schematic diagram of a deep learning prediction model according to an embodiment of the present invention is shown. The deep learning prediction model mainly includes a composite CNN-LSTM consisting of a Bayesian convolutional neural network layer, a pooling layer, and an LSTM layer. Specifically, combined with... Figure 4 Here, "Feature Dimension" represents the feature dimension, and "TimeSequence" represents the time sequence. First, a three-layer 1D Bayesian convolutional neural network extracts the deep spatial dimension of the input samples along the time axis. Optionally, the input samples can be test data or training data. Each Bayesian convolutional neural network layer is connected to a pooling layer, and then to a CNN and an LSTM through a fully connected layer. The LSTM is configured as a two-layer network, and finally, the model outputs through a fully connected layer.

[0056] As mentioned above, the application of Bayesian convolutional neural networks requires significant improvements to traditional deep neural networks. Since the model parameters are replaced by probability distributions instead of specific numerical values, many additional hyperparameters are introduced into the neural network. Furthermore, the inference process and the structure of the deep neural network itself are quite complex, which greatly increases the computational load of modeling and causes significant time consumption during model training and prediction. However, with the advancement of computer technology and the improvement of computing power, deep learning prediction models with the above-mentioned structure can effectively overcome the problem of computational load and achieve better accuracy in predicting pore structures while taking into account the measurement of prediction uncertainty.

[0057] Preferably, the method may further include: selecting a corresponding sliding window size based on the data size of the test data. It is understood that the data size may vary for different test data; therefore, selecting a suitable sliding window size for different data sizes enables dynamic control of the sliding window, ensuring data validity during data transmission and avoiding blocking. Optionally, during training or testing, the number of channels can be set to 24, the dropout rate to 0.2, and the activation function of all network layers can be set to ReLU.

[0058] Figure 5This diagram illustrates pore structure prediction results obtained using a deep learning prediction model including or excluding Bayesian convolutional neural network layers, according to an embodiment of the present invention. Figure 3 Similarly, the horizontal axis represents time, and the vertical axis represents pore structure parameters. For example... Figure 5 As shown, the dashed line represents the pore structure value predicted by CNN-LSTM, while the solid line represents the pore structure value predicted by CNN-LSTM with a Bayesian convolutional neural network layer. From the prediction results, the overall trend of reservoir pore structure is consistent, with some differences in details. This is permissible in geophysical reservoir prediction, as the CNN-LSTM prediction structure with a Bayesian convolutional neural network layer is closer to the true value.

[0059] Optionally, the method may also include:

[0060] Step S121: Using rock physics analysis and forward modeling to construct label technology, a label training dataset is generated based on the original data. The label training dataset includes seismic data and pore structure parameters.

[0061] Figure 6 A schematic diagram illustrating the generation of a labeled training dataset using a labeling technique based on rock physics analysis and forward modeling, according to an embodiment of the present invention, is shown.

[0062] exist Figure 6 In the illustrated embodiment, step S121, which generates a labeled training dataset based on the original data, may include the following steps: step S121a, simulating the depositional environment of carbonate rocks; step S121b, obtaining the lithofacies and rock physical property parameters of the target reservoir; and step S121c, generating seismic records based on the lithofacies and rock physical property parameters, wherein the seismic data, pore structure parameters, and seismic records in the labeled training dataset have a one-to-one correspondence.

[0063] Specifically, the pore structure obtained from well logging and the seismic data of that well point can constitute a sample label. However, the number of wells is usually limited, greatly reducing the amount of available label training data. Therefore, rock physics and geological methods can be used to analyze the observed well data to artificially generate a label training dataset for training the model. In fact, from a geological perspective, the formation of subsurface strata is a physical, non-random process. Therefore, artificially adding well data and generating synthetic reservoir models must at least consider the spatial correlation caused by locational changes. Therefore, in the label construction process, a carbonate sedimentary environment is first assumed, with two main lithofacies in the reservoir profile: limestone (referred to as facies 1). Figure 6 The middle part is shown as Facise1), and the mud (called phase 2) is shown as phase 2. Figure 6(Shown as Facie2). In this case, lithofacies and rock physical property parameters can be obtained from the logging curves: porosity (using...) The model uses the finite element (represented by the clay volume) and the volume of clay (Vclay). Then, geostatistical techniques are used to simulate 2000 one-dimensional geological models, each 200 meters thick. Optionally, sequential indicator simulation can generate several lithofacies with predefined lithofacies proportions and variograms. Similarly, sequential Gaussian simulation is used to populate the physical properties of each lithofacies using predefined variograms; see [link to Gaussian simulation]. Figure 6 In the 'b', "Dolomite" represents marble and "Limestone" represents limestone. Both simulations assume the variogram type is spherical. The generated physical properties are then correlated with the product of P-wave velocity and density using a multifacies petrophysical model to obtain the porosity structure; see [link to relevant documentation]. Figure 6 c. Based on this, using the reflectivity method, given a seismic wavelet, the corresponding seismogram can be synthesized. See [link to seismogram]. Figure 6 In the equation 'd', 'Amplitude' represents the amplitude and 'Time' represents the time. These synthesized seismic records have a one-to-one correspondence with pore structure parameters and seismic data.

[0064] Step S122: Train the deep learning prediction model based on the labeled training dataset.

[0065] After artificially synthesizing seismic records, the corresponding pore structure parameters and seismic data are combined and used as a labeled training dataset, which is then used as input training data for a deep learning prediction model, thereby training the deep learning prediction model.

[0066] This effectively increases the amount of data available for training, thereby improving training accuracy and thus enhancing the model's prediction results. It also strengthens the model's robustness.

[0067] Optionally, the method further includes: step S140, calculating the posterior distribution of the weights of the Bayesian convolutional neural network layer.

[0068] For example, for a training set D, assuming the input variable is X, the output is the distribution of the predicted value Y, and W is the weight in the Bayesian convolutional neural network, the Bayesian convolutional neural network can be regarded as a model satisfying the conditional distribution P(Y|X, W). As mentioned earlier, the Bayesian estimation algorithm can introduce a prior distribution, and then the posterior distribution P(W|D) of W can be obtained, introducing uncertainty into the prediction of the deep learning prediction model and optimizing the pore structure prediction results.

[0069] Optionally, step S140, which calculates the posterior distribution of the weights of the Bayesian convolutional neural network layer, may include the following steps.

[0070] Step S141: Using variational inference, the first distribution controlled by the first parameter is made to approximate the posterior distribution to obtain the objective function.

[0071] It is understandable that, since the prediction result is a probability distribution, the probability model for predicting Y from the input variable X based on the weight W becomes formula (1): P(Y|X)=E P(w|D) [P(Y|X,W)]. Using variational inference, the true posterior P(W|D) can be approximated by a distribution q(W|θ) controlled by a set of parameters θ. Thus, the problem of finding the posterior distribution can be transformed into an optimization problem of θ. For example, this can be achieved by minimizing the KL divergence (relative entropy) of the two distributions, specifically through formula (2):

[0072]

[0073]

[0074] This leads to formula (3), which is the objective function:

[0075] F(D, θ) = D KL [q(W|θ)||P(W)]-E q(W|θ) [log P(D|W)]. Here, the first term on the right is the complexity cost, which is the KL divergence between the variational inference posterior and prior, describing the degree of fit between the weights and the prior. The second term is the likelihood cost, whose value depends on the training data and describes the degree of fit to the samples.

[0076] Step S142: Based on the objective function and the labeled training dataset, obtain the loss function of the Bayesian convolutional neural network layer.

[0077] Generally, the training set D can include the input variable X and the output predicted value Y. Therefore, we can obtain formula (4): E q(W|θ) [log P(D|W)]=E q(W|θ) [log q(Y|W,X)]+P(X). It can be understood that for a random variable ε and a probability density q(ε), as long as q(ε)dε=q(W|θ)dW is satisfied, an unbiased estimate of the expected derivative can be obtained. Since P(X) is a known distribution and can be ignored, combining formula (4), the Monte Carlo approximation of formula (3) can be obtained as formula (5):

[0078] Among them, W i This represents the weight sampling when processing the i-th data point, where the Gaussian distribution of the i-th weight is θ. i={μ i , σ i When m=1, W i Independent of each other, W i With θ i They are independent of each other, θ i They are also independent of each other. Therefore, given the training data {x} j y j From this, we can obtain formula (6), which is the loss function:

[0079]

[0080] Step S143: Obtain weight sample values ​​to optimize the first parameter using the loss function and weight sample values.

[0081] For example, W can be obtained through sampling. i The specific value of θ can be substituted into formula (6) to calculate the loss function value, and then backpropagation can be used to calculate the first parameter θ. i Optimization is performed to achieve overall network parameter optimization during training.

[0082] Therefore, by modeling the posterior probability of the parameters, we can effectively reduce overlearning during training, reduce the impact of noise during training, and improve the generalization ability to unknown data, that is, improve the generalization ability of the model.

[0083] Optionally, step S143, obtaining the weight sample value, may include: sampling from each Gaussian distribution of the first parameter to obtain the weight sample value.

[0084] For example, during prediction, it can be derived from each Gaussian distribution θ i ={μ i ,σ i Sampling is performed in} to obtain the weighted sample value W. i .

[0085] In this embodiment, the method may further include:

[0086] Step S150: After multiple samplings, multiple pore structure prediction results are obtained.

[0087] As can be understood, Bayesian neural networks differ from general neural networks in that their weight parameters are random variables following a certain distribution, such as a Gaussian distribution, rather than deterministic values. Therefore, the weight parameters can be sampled multiple times, thus obtaining multiple prediction results as described above.

[0088] Step S160: Average the prediction results of multiple pore structures to obtain the final prediction result.

[0089] It is understandable that the prediction result is a probability distribution, and the final prediction result can be obtained by averaging multiple probability distribution results.

[0090] This avoids the occurrence of accidental events and ensures the accuracy of the prediction results.

[0091] According to a second aspect of the invention, a device for inverting the pore structure of carbonate rocks is also proposed. Figure 7 A schematic block diagram of a carbonate rock pore structure inversion apparatus 700 according to an embodiment of the present invention is shown. Figure 7 As shown, the device 700 may include:

[0092] Module 710 is used to acquire test data;

[0093] The inversion module 720 is used to perform pore structure inversion using a Bayesian estimation algorithm on test data with a trained deep learning prediction model to obtain pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer.

[0094] According to a third aspect of the present invention, an electronic device is also provided. Figure 8 A schematic block diagram of an electronic device 800 according to an embodiment of the present invention is shown. Figure 8 As shown, the electronic device 800 includes a processor 810 and a memory 820, wherein the memory 820 stores computer program instructions, which are executed by the processor 810 to perform the carbonate rock pore structure inversion method as described above.

[0095] According to a fourth aspect of the invention, a storage medium is also provided, on which program instructions are stored, which, when executed, are used to perform the carbonate rock pore structure inversion method as described above. The storage medium may, for example, include a storage component of a tablet computer, a hard disk of a computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0096] Those skilled in the art can understand the specific details and beneficial effects of the carbonate rock pore structure inversion device, electronic equipment, and storage medium by reading the above description of the relevant method for inverting the pore structure of carbonate rocks, and will not be elaborated here for the sake of brevity.

[0097] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and / or device can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0098] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0100] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0101] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for inverting the pore structure of carbonate rocks, characterized in that, include: Obtain test data; For the test data, a Bayesian estimation algorithm is used to invert the pore structure using a trained deep learning prediction model to obtain pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer. The posterior distribution of the weights of the Bayesian convolutional neural network layer is calculated. This calculation includes: using variational inference to approximate the posterior distribution with a first distribution controlled by a first parameter to obtain an objective function; obtaining a loss function for the Bayesian convolutional neural network layer based on the objective function and the labeled training dataset; and acquiring weight sample values ​​to apply the loss function and the weight sample values ​​to the first... The parameters are optimized; obtaining the weight sampling value includes: sampling from each Gaussian distribution of the first parameter to obtain the weight sampling value; the method further includes: obtaining multiple pore structure prediction results after multiple samplings; averaging the multiple pore structure prediction results to obtain the final prediction result; the deep learning prediction model further includes: a fully connected layer, a pooling layer, a long short-term memory network, and a convolutional neural network, wherein the Bayesian convolutional neural network layer is connected to the pooling layer, and is connected to the long short-term memory network and the convolutional neural network via the fully connected layer; the method further includes: selecting the corresponding sliding window size according to the data scale of the test data.

2. The method for inverting the pore structure of carbonate rocks as described in claim 1, characterized in that, The method further includes: A labeling technique is constructed using rock physics analysis and forward modeling. A label training dataset is generated based on the original data, wherein the label training dataset includes seismic data and pore structure parameters. The deep learning prediction model is trained based on the labeled training dataset.

3. The method for inverting the pore structure of carbonate rocks as described in claim 2, characterized in that, The labeled training dataset generated based on the original data includes: Simulate the depositional environment of the carbonate rocks; Obtain the lithofacies and rock physical property parameters of the target reservoir; Based on the lithofacies and rock physical property parameters, seismic records are generated, wherein the seismic data, the pore structure parameters, and the seismic records in the labeled training dataset have a one-to-one correspondence.

4. A device for inverting the pore structure of carbonate rocks, characterized in that, include: The acquisition module is used to acquire test data; The inversion module is used to perform pore structure inversion on the test data using a trained deep learning prediction model and a Bayesian estimation algorithm to obtain pore structure prediction results. The deep learning prediction model includes a Bayesian convolutional neural network layer. The module calculates the posterior distribution of the weights of the Bayesian convolutional neural network layer. This calculation includes: using variational inference to approximate the posterior distribution with a first distribution controlled by a first parameter to obtain an objective function; obtaining the loss function of the Bayesian convolutional neural network layer based on the objective function and the labeled training dataset; and acquiring weight sample values ​​to use the loss function and the weight sample values ​​to... The first parameter is optimized; obtaining the weight sampling value includes: sampling from each Gaussian distribution of the first parameter to obtain the weight sampling value; the method further includes: obtaining multiple pore structure prediction results after multiple samplings; averaging the multiple pore structure prediction results to obtain a final prediction result; the deep learning prediction model further includes: a fully connected layer, a pooling layer, a long short-term memory network, and a convolutional neural network, wherein the Bayesian convolutional neural network layer is connected to the pooling layer, and is connected to the long short-term memory network and the convolutional neural network via the fully connected layer; the method further includes: selecting the corresponding sliding window size according to the data scale of the test data.

5. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores computer program instructions, which, when executed by the processor, are used to perform the carbonate rock pore structure inversion method as described in any one of claims 1 to 3.

6. A storage medium storing program instructions that, when executed, perform the carbonate rock pore structure inversion method as described in any one of claims 1 to 3.

Images