A method, apparatus, equipment, storage medium, and product for predicting gas saturation.

Abstract

This specification relates to the field of seismic data processing and interpretation technology, and provides a method, apparatus, device, storage medium, and product for predicting gas saturation. The method includes: acquiring pre-stack angle gather data of a target reservoir; performing pre-stack inversion on the pre-stack angle gather data using the Zoeppritz equation to obtain the elastic parameters of the target reservoir; and inputting the elastic parameters of the target reservoir into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on a KAN model, which uses a learnable activation function to capture the nonlinear relationship between the elastic parameters and the gas saturation. This specification's embodiments can improve the accuracy of gas saturation prediction.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of seismic data processing and interpretation technology, and in particular to a method, apparatus, equipment, storage medium, and product for predicting gas saturation. Background Technology

[0002] Gas saturation is a physical quantity describing the degree to which the pore space of a reservoir rock is filled with natural gas. In oil and gas exploration and development, accurate gas saturation data is fundamental to calculating natural gas reserves. By measuring gas saturation, the natural gas reserves of a reservoir can be estimated, providing a scientific basis for oil and gas field development. Furthermore, based on the distribution of gas saturation, reasonable oil and gas field development strategies can be formulated, thereby improving the economic benefits of oil and gas fields. Therefore, accurately predicting gas saturation is of great significance to oil and gas exploration and development.

[0003] Currently, numerous well-logging interpretation methods for gas saturation have been proposed in the field of geophysical exploration, but seismic interpretation methods for gas saturation are relatively few. Predicting gas saturation in natural gas reservoirs using seismic data is a complex, multifaceted, and highly nonlinear seismic inversion problem. Existing methods, based on seismic inversion, typically use rock physics models to establish a relationship between reservoir elastic parameters and reservoir parameters, and select appropriate inversion algorithms to obtain reservoir parameters. However, for natural gas exploration areas with increasingly complex geological conditions, the rock physics relationship between elastic parameters and physical property parameters often exhibits extremely strong nonlinearity and is easily affected by various factors such as lithology, pore structure, pressure, and temperature in the actual study area, making it difficult to establish accurate rock physics models in most cases. With the development of artificial intelligence, the emergence of machine learning methods provides a new approach to this type of nonlinear modeling problem. Machine learning methods can obtain a nonlinear mapping relationship through machine autonomous learning, achieving highly complex nonlinear function approximation, and possessing a powerful ability to learn the essence of datasets and highly abstract features. However, traditional deep learning typically relies on fixed activation functions. In complex nonlinear problems, fixed activation functions cannot optimally adapt to the inherent characteristics of the data, limiting the model's expressive and generalization abilities. This results in low accuracy when using traditional deep learning models for gas saturation prediction. Therefore, there is an urgent need for a gas saturation prediction method that can improve the accuracy of gas saturation prediction and provide strong technical support for oil and gas exploration and development under complex geological conditions. Summary of the Invention

[0004] In view of the above problems and the prior art, the purpose of the embodiments of this specification is to provide a method, apparatus, device, storage medium and product for predicting gas saturation, so as to solve the problem of low accuracy in predicting gas saturation in the prior art.

[0005] To solve the above-mentioned technical problems, the specific technical solutions of the embodiments in this specification are as follows:

[0006] On the one hand, embodiments of this specification provide a method for predicting gas saturation, the method comprising:

[0007] Acquire pre-stack corner gather data of the target reservoir;

[0008] The elastic parameters of the target reservoir are obtained by pre-stack inversion of the pre-stack angle gather data using the zoeppritz equation.

[0009] The elastic parameters of the target reservoir are input into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on the KAN model, which uses a learnable activation function to capture the nonlinear relationship between the elastic parameters and the gas saturation.

[0010] Further, acquiring the pre-stack angle gather data of the target reservoir includes:

[0011] Acquire pre-stack CMP gather data and forward modeling data of geological bodies for the target reservoir;

[0012] The navigation pyramid method was used to perform multi-scale decomposition and denoising on the pre-stack CMP gather data to obtain denoised multi-scale seismic data.

[0013] Based on the forward modeling characteristics of the geological body, angle gathers are extracted from the denoised multi-scale seismic data at scales that conform to the forward modeling characteristics of the geological body to obtain pre-stack angle gather data.

[0014] Furthermore, the step of using the Zoeppritz equation to perform pre-stack inversion on the pre-stack angle gather data to obtain the elastic parameters of the target reservoir includes:

[0015] A priori distribution is established based on well logging data and forward modeling characteristics of the target reservoir;

[0016] A likelihood function is constructed based on the zoeppritz equation and the pre-stack angle gather data.

[0017] Calculate the posterior distribution value based on the prior distribution and the likelihood function;

[0018] The maximum a posteriori estimate is obtained by sampling the posterior distribution value, and the maximum a posteriori estimate is used as the elasticity parameter estimate.

[0019] Based on the estimated elastic parameters, determine whether the inversion process meets the preset convergence condition;

[0020] If so, output the estimated value of the elasticity parameter;

[0021] If not, calculate the output value of the Zoeppritz equation and the preset inversion objective function value based on the estimated value of the elastic parameters, and iteratively update the estimated value of the elastic parameters based on the output value of the Zoeppritz equation and the preset inversion objective function value until the preset convergence condition is met.

[0022] Furthermore, the training process of the gas saturation prediction model includes:

[0023] Acquire pre-stack CMP gather data, geological body forward modeling data, and well logging data for several reservoirs;

[0024] The navigation pyramid method was used to perform multi-scale decomposition and denoising on the pre-stack CMP gather data to obtain denoised multi-scale seismic data.

[0025] Based on the forward modeling feature data of the geological body, select scale seismic data that conforms to the forward modeling feature of the geological body from the denoised multi-scale seismic data and perform corner gather extraction to obtain pre-stack corner gather data;

[0026] The elastic parameters are obtained by pre-stack inversion of the pre-stack angle gather data using the zoeppritz equation.

[0027] Several elastic moduli are calculated based on the elastic parameters;

[0028] Estimate gas saturation based on the well logging data;

[0029] The correlation between each elastic modulus and the gas saturation is calculated. Elastic moduli with a correlation higher than a preset threshold are selected as training data. The corresponding gas saturation is used as the training label to train the KAN model and obtain a gas saturation prediction model. The KAN model captures the nonlinear relationship between the elastic parameters and the gas saturation through a learnable activation function.

[0030] Further, estimating gas saturation based on the well logging data includes:

[0031] The gas saturation can be estimated using the following formula:

[0032]

[0033] Where a is the lithology coefficient, m is the cementation index, n is the saturation index, R0 is the formation resistivity, and R w φ represents the resistivity of formation water, and φ represents the porosity.

[0034] Furthermore, the training process of the KAN model includes:

[0035] The activation function of each edge in the KAN model is initialized;

[0036] The training data is input into the KAN model, the output value is calculated through forward propagation, and the loss value is calculated based on the difference between the output value and the training label.

[0037] The model parameters are updated using the backpropagation algorithm based on the loss value;

[0038] Obtain a visualization of the output value of each activation function. If there is a non-smooth curve, adjust the structure of the KAN model until a smooth curve is obtained.

[0039] Assign a corresponding activation function to each smooth curve based on its shape, and replace the original activation function with the newly assigned activation function.

[0040] Retrain the model using the updated parameters and activation function until the iteration terminates.

[0041] On the other hand, embodiments of this specification provide a gas saturation prediction device, the device comprising:

[0042] The acquisition module is used to acquire pre-stack corner gather data of the target reservoir;

[0043] The inversion module is used to perform pre-stack inversion on the pre-stack angle gather data using the zoeppritz equation to obtain the elastic parameters of the target reservoir.

[0044] The prediction module is used to input the elastic parameters of the target reservoir into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on the KAN model, which uses a learnable activation function to capture the nonlinear relationship between the elastic parameters and the gas saturation.

[0045] In another aspect, embodiments of this specification also provide a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the computer program, when executed by the processor, performs instructions of any of the methods described above.

[0046] In another aspect, embodiments of this specification also provide a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor of a computer device to perform instructions for any of the methods described above.

[0047] In another aspect, embodiments of this specification also provide a computer program product, which, when run by the processor of a computer device, executes instructions for any of the methods described above.

[0048] By employing the above technical solution, the gas saturation prediction method provided in this specification utilizes the Zoeppritz equation to perform pre-stack inversion on pre-stack angle gather data, avoiding errors that may arise from approximate equations. This results in more accurate elastic parameters extracted from seismic data, providing a reliable foundation for subsequent analysis and prediction. The elastic parameters are input into a gas saturation prediction model trained based on the KAN model to obtain the gas saturation. Compared to traditional deep learning networks, the KAN model, through a learnable activation function, can capture the complex nonlinear relationship between elastic parameters and gas saturation, thereby improving the accuracy of gas saturation prediction.

[0049] The above description is merely an overview of some embodiments of the technical solutions in this specification. In order to better understand the technical means of some embodiments of this specification and to implement them in accordance with the content of the specification, and to make the above and other objects, features and advantages of the embodiments of this specification more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This specification illustrates the steps of a gas saturation prediction method in some embodiments.

[0052] Figure 2 This specification illustrates the steps for obtaining pre-stack corner gather data of a target reservoir in some embodiments.

[0053] Figure 3 This specification shows schematic diagrams illustrating the steps of pre-stack inversion in some embodiments;

[0054] Figure 4 This specification shows schematic diagrams illustrating the training process of the gas saturation prediction model in some embodiments;

[0055] Figure 5 This specification shows schematic diagrams of the KAN model training process in some embodiments;

[0056] Figure 6 This specification shows a schematic diagram of the structure of a gas saturation prediction device in some embodiments;

[0057] Figure 7A schematic diagram of the structure of a computer device is shown in this specification.

[0058] Explanation of symbols in the attached drawings:

[0059] 601. Acquisition Module;

[0060] 602. Inversion Module;

[0061] 603. Prediction Module;

[0062] 702. Computer equipment;

[0063] 704, Processor;

[0064] 706. Memory;

[0065] 708. Drive mechanism;

[0066] 710. Input / Output Module;

[0067] 712. Input devices;

[0068] 714. Output devices;

[0069] 716. Presentation equipment;

[0070] 718. Graphical User Interface;

[0071] 720. Network interface;

[0072] 722. Communication link;

[0073] 724. Communication bus. Detailed Implementation

[0074] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0075] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the acquisition, storage, use, and processing of data in the technical solutions described in the embodiments of this application all comply with relevant regulations.

[0076] It should be noted that the terms "first," "second," etc., used in this specification, claims, and the foregoing drawings 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 a non-exclusive inclusion; for example, a process, method, apparatus, product, or device 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 devices.

[0077] To address the aforementioned problems, this specification provides a method for predicting gas saturation. Figure 1 This diagram illustrates the steps of a gas saturation prediction method provided in the embodiments of this specification. While this specification provides the operational steps of the method as described in the embodiments or flowcharts, more or fewer operational steps may be included based on conventional or non-inventive methods. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only possible execution order. In actual system or device products, the methods shown in the embodiments or accompanying drawings can be executed sequentially or in parallel. Specifically, as shown in the diagram... Figure 1 As shown, the method may include:

[0078] S101: Obtain pre-stack angle gather data of the target reservoir.

[0079] In this embodiment, the pre-stack angle gather data of the target reservoir contains information about incident waves reflected back from the subsurface medium at different angles. Compared with traditional offset gathers, angle gather data has a higher signal-to-noise ratio, better continuity, and a more uniform amplitude distribution. By analyzing the amplitude and phase information at different angles, the physical characteristics of the reservoir can be identified, thereby providing more accurate and reliable data support for subsequent gas saturation prediction. Pre-stack angle gather data can be obtained by denoising, filtering, and migration imaging of the original seismic data.

[0080] S102: Use the zoeppritz equation to perform pre-stack inversion on the pre-stack angle gather data to obtain the elastic parameters of the target reservoir.

[0081] In this embodiment, the elastic parameters include P-wave velocity, S-wave velocity, and density. The Zoeppritz equation is a mathematical model describing the reflection and transmission of seismic waves at different media interfaces. By utilizing the Zoeppritz equation, elastic parameters of the subsurface reservoir can be extracted from seismic data. Compared with approximate formulas of the Zoeppritz equation (such as the Aki-Richards equation and the Shuey equation), pre-stack inversion based on the exact solution of the Zoeppritz equation and large-angle information in pre-stack angle gather data can avoid errors caused by approximate formulas at specific angles or conditions, thereby improving the reliability and accuracy of the inversion results.

[0082] S103: Input the elastic parameters of the target reservoir into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir, wherein the gas saturation prediction model is trained based on the KAN model, and the KAN model captures the nonlinear relationship between the elastic parameters and the gas saturation through a learnable activation function.

[0083] Since conventional neural network models typically rely on fixed activation functions, these functions cannot optimally adapt to the inherent characteristics of the data in complex nonlinear problems, limiting the model's expressive and generalization capabilities. This results in low accuracy when using traditional deep learning models for gas saturation prediction. Therefore, this specification's embodiments are based on the KAN (Kolmogorov-Arnold Network) model for gas saturation prediction. Unlike traditional MLP (Multilayer Perceptron) models, the KAN model improves performance by shifting activation functions from nodes to edges and uses spline functions to parameterize univariate activation functions on the edges. This design allows the KAN model to automatically learn customized activation functions for different features, greatly enhancing its flexibility. Furthermore, the KAN model completely abandons linear weights, replacing all linear weights with learnable univariate functions, thus enabling the modeling of arbitrarily complex nonlinear relationships.

[0084] In summary, by adopting the above technical solution, the gas saturation prediction method provided in this specification utilizes the Zoeppritz equation to perform pre-stack inversion on pre-stack angle gather data, avoiding errors that may be caused by approximate equations. This makes the elastic parameters extracted from seismic data more accurate, providing a reliable foundation for subsequent analysis and prediction. The elastic parameters are input into the gas saturation prediction model trained based on the KAN model to obtain the gas saturation. Compared with traditional deep learning networks, the KAN model, through a learnable activation function, can capture the complex nonlinear relationship between elastic parameters and gas saturation, thereby improving the accuracy of gas saturation prediction.

[0085] In the embodiments of this specification, refer to Figure 2 In step S101, obtaining the pre-stack angle gather data of the target reservoir includes:

[0086] S201: Obtain pre-stack CMP gather data and forward modeling characteristic data of geological bodies for the target reservoir.

[0087] Pre-stack CMP gather data refers to the extraction of gathers from different shot gathers that share a common midpoint, forming a new set. In seismic data acquisition, if the subsurface interface is horizontal, the projection of the common reflection point onto the ground must be the center point of the receiver distance of the common reflection points in the shot gather; hence, it is called the common midpoint. These gathers record the reflection information of seismic waves under different combinations of shot and receiver points, containing rich information about the subsurface medium, such as lithology and structure.

[0088] Forward modeling data of geological bodies is typically obtained through geological modeling and forward simulation, and includes information such as the morphology, structure, and physical properties of the geological body. Combining pre-stack CMP gather data and forward modeling data of geological bodies for inversion can comprehensively consider the propagation characteristics of seismic waves and the actual characteristics of the geological body, thereby obtaining more accurate reservoir prediction results.

[0089] S202: The pre-stack CMP gather data is decomposed and denoised using the navigation pyramid method to obtain denoised multi-scale seismic data.

[0090] The navigation pyramid can decompose pre-stack CMP gather data into sub-components of different scales. Then, noise is identified and suppressed at each scale to remove noise and interference from the sub-components, thereby improving the signal-to-noise ratio and resolution of the seismic data and providing a better data foundation for subsequent seismic interpretation and inversion. Multi-scale decomposition and denoising of pre-stack seismic CMP gathers first utilizes a polar coordinate filter. The polar coordinate filter formula is as follows:

[0091]

[0092] in, K a and K b These are the start and cutoff wavenumbers of the filter's tapered region, respectively.

[0093] Secondly, a directionally controllable filter is used for multi-directional decomposition. This filter has arbitrary rotation capabilities, generating multiple directionally controllable sub-band sets at each layer. Any direction can be formed by a linear combination of several basis filters, whose corresponding filter bases overlap to some extent in the frequency domain. The formula is as follows:

[0094] G θ =G 0° cosθ+G90° sinθ

[0095] Where θ is the directional input of the directionally controllable filter, and G 0° For the base filter with a direction of 0°, G 90° The base filters are oriented at 90°. After multi-scale decomposition using polar coordinate filters and multi-directional decomposition using directionally controllable filters, the decomposition coefficients of the pre-stack seismic CMP gather at different scales and directions are obtained. These coefficients contain characteristic information of the seismic signal at different frequencies and directions, as well as noise characteristics. Finally, it is necessary to find the optimal direction for interpolation reconstruction to select the part that best represents the seismic signal characteristics from the decomposed coefficients while suppressing noise. Optionally, the optimal direction can be determined by methods such as coherence analysis or energy ratio analysis. After determining the optimal direction, an appropriate interpolation algorithm needs to be selected for reconstruction. To obtain an accurate interpolation function, the number of base filters required needs to be accurately known during interpolation reconstruction to obtain the weight of each base filter. After interpolation reconstruction is completed, the denoised pre-stack seismic CMP gather is obtained.

[0096] S203: Based on the forward modeling feature data of the geological body, select scale seismic data that conforms to the forward modeling feature of the geological body from the denoised multi-scale seismic data and extract corner gathers to obtain pre-stack corner gather data.

[0097] In the embodiments of this specification, selecting scale-based seismic data that conforms to the forward modeling characteristics of the geological body from the denoised multi-scale seismic data based on the forward modeling characteristic data of the geological body includes:

[0098] Seismic features are extracted from the denoised multi-scale seismic data to obtain seismic features at each scale.

[0099] Calculate the correlation between the seismic features at each scale and the forward modeling features of the geological body, and select seismic data at scales that conform to the forward modeling features of the geological body based on the correlation and a preset threshold.

[0100] After determining the appropriate scale, pre-stack CMP data corresponding to that scale is extracted. To extract angle gathers, the data in the CMP gathers needs to be divided according to the incident angle. This typically involves converting the seismic data by offset and angle to obtain seismic gathers at different incident angles. Specifically, pre-stack angle gather extraction usually needs to be performed after migration imaging. Migration imaging can simulate the propagation path of seismic waves in the subsurface medium, thereby obtaining the imaging results of the subsurface medium. After migration imaging, offset-domain co-imaging gathers can be extracted according to imaging conditions (such as non-zero offset imaging conditions), and then converted into angle-domain co-imaging gathers, i.e., pre-stack angle gathers, using tilt stacking techniques.

[0101] In the embodiments of this specification, refer to Figure 3 In step S102, the step of using the Zoeppritz equation to perform pre-stack inversion on the pre-stack angle gather data to obtain the elastic parameters of the target reservoir includes:

[0102] S301: Establish a priori distribution based on well logging data and forward modeling characteristics of geological bodies of the target reservoir;

[0103] S302: Construct a likelihood function based on the zoeppritz equation and pre-stack angle gather data;

[0104] S303: Calculate the posterior distribution value based on the prior distribution and the likelihood function;

[0105] S304: Sample the posterior distribution value to obtain the maximum a posteriori estimate, and use the maximum a posteriori estimate as the elasticity parameter estimate;

[0106] S305: Determine whether the inversion process meets the preset convergence condition based on the estimated value of the elastic parameters;

[0107] S306: If so, output the estimated value of the elasticity parameter;

[0108] S307: If not, calculate the output value of the zoeppritz equation and the preset inversion objective equation function value based on the estimated value of the elastic parameters, and iteratively update the estimated value of the elastic parameters based on the output value of the zoeppritz equation and the preset inversion objective equation function value until the preset convergence condition is met.

[0109] Pre-stack inversion can be understood as using angle gather data, well logging data, and geological information to infer subsurface elastic parameters. A very common problem in geophysics is ill-posedness, and pre-stack inversion is one such ill-posed problem. Bayesian theory is a popular method for controlling inversion stability; it uses prior information to constrain pre-stack inversion, reducing ill-posedness and improving stability. Commonly used prior distributions include Gaussian, Cauchy, and Laplace distributions. This specification uses a Gaussian prior distribution as an example to obtain elastic parameters; therefore, the posterior distribution can be expressed as:

[0110]

[0111] Where m is the estimated value of the elasticity parameter, n is the number of pre-stack angle gather data d, B is the block diagonal matrix formed by the wavelet convolution matrix, μ is the coefficient of the regularization constraint term, ∑e is the error covariance matrix, and R(m) is the regularization constraint term obtained from the prior distribution function. Taking the derivative of the above equation with respect to m and setting it to zero, the final inversion objective equation is obtained as follows:

[0112]

[0113] Where B is represented as:

[0114]

[0115] Where W is the P-wave reflection coefficient, G is the partial derivative matrix, N is the number of angular discretizations, m is the estimated value of the elastic parameter, L is the half-wavelength, K is the number of inversion parameters, M is the number of stratigraphic layers, and r pp Let θ be the reflection coefficient. i (i = 1, 2, ... N) is the angle of incidence.

[0116] In step S304, by finding the maximum point of the posterior distribution, a most representative parameter estimate can be obtained, which reduces the uncertainty of the parameters to a certain extent. In seismic exploration or geophysical inversion, this means that the elastic parameters of the subsurface medium can be estimated more accurately. In step S305, the Zoeppritz equation output value and the objective equation function value are solved based on the current elastic parameter estimate. The convergence condition is judged based on the objective equation function value. If the convergence condition is met, the iterative process is considered to have ended, and the final elastic parameter estimate is obtained. If not, the parameter update direction needs to be calculated based on the objective equation function value. The update direction can be calculated using optimization algorithms such as the conjugate gradient method. Then, the elastic parameter estimate is updated according to the update direction. This update process can be regarded as moving one step in the parameter space along a specific direction to approximate the true parameter value.

[0117] Accurate pre-stack inversion using Zoeppritz can yield precise elastic parameters (P-wave and S-wave velocities and densities) of the target reservoir. To accurately fit the complex nonlinear relationship between elastic parameters and gas saturation, the embodiments in this specification select the KAN model as the fitting model.

[0118] In the embodiments of this specification, refer to Figure 4 The training process of the gas saturation prediction model includes:

[0119] S401: Acquire pre-stack CMP gather data, geological body forward modeling data, and well logging data for several reservoirs;

[0120] S402: The pre-stack CMP gather data is decomposed into multiple scales and denoised using the navigation pyramid method to obtain denoised multi-scale seismic data.

[0121] S403: Based on the forward modeling feature data of the geological body, select scale seismic data that conforms to the forward modeling feature of the geological body from the denoised multi-scale seismic data and perform corner gather extraction to obtain pre-stack corner gather data;

[0122] S404: Elastic parameters are obtained by pre-stack inversion of the pre-stack angle gather data using the zoeppritz equation;

[0123] S405: Several elastic moduli are calculated based on the elastic parameters;

[0124] S406: Estimate gas saturation based on the well logging data;

[0125] S407: Calculate the correlation between each elastic modulus and the gas saturation, select the elastic modulus with a correlation higher than a preset threshold as training data, use the corresponding gas saturation as training labels, train the KAN model to obtain a gas saturation prediction model, wherein the KAN model captures the nonlinear relationship between the elastic parameters and the gas saturation through a learnable activation function.

[0126] The specific implementation methods of steps S401-S404 are consistent with those described above, and therefore will not be repeated here. In step S405, Poisson's ratio, the ratio of P-wave to S-wave velocity, and the Lamé coefficient, among other elastic moduli, can be calculated based on the P-wave and S-wave velocities and density. Due to the complex and variable characteristics of underground reservoirs, different elastic moduli reflect different physical and chemical properties within the reservoir. Through correlation analysis, the variable most correlated with gas saturation can be selected from numerous elastic moduli, thereby avoiding the introduction of unnecessary noise and interference in subsequent analysis and prediction. In the embodiments of this specification, by calculating the correlation between each elastic modulus and the gas saturation, elastic moduli with a correlation higher than a preset threshold are selected as training parameter variables.

[0127] In step S406, to accurately estimate the gas saturation, this embodiment of the specification uses Archie's equation to calculate the gas saturation. Archie's equation is a commonly used method for calculating water saturation Sw. After obtaining the water saturation Sw, the gas saturation Sg can be calculated using the relationship between oil, gas, and water saturation (S + S + Sg = 1). Specifically, the gas saturation is estimated using the following formula:

[0128]

[0129] Where a is the lithology coefficient, m is the cementation index, n is the saturation index, R0 is the formation resistivity, and R w φ represents the resistivity of formation water, and φ represents the porosity.

[0130] The KAN model is trained using the gas saturation value calculated using Archie's formula as the training label and the selected elastic modulus parameters as the training data. Once the model is trained, the inverted P-wave and S-wave velocities and densities are used as inputs, and the output is the predicted gas saturation value. The KAN model applies a learnable activation function to the weights, rather than using a fixed activation function on the nodes (neurons), resulting in a more accurate network representation and higher gas saturation prediction precision.

[0131] In the embodiments of this specification, refer to Figure 5 The training process of the KAN model includes:

[0132] S501: Initialize the activation function of each edge in the KAN model;

[0133] S502: Input the training data into the KAN model, calculate the output value through forward propagation, and calculate the loss value based on the difference between the output value and the training label;

[0134] S503: Update the model parameters using the backpropagation algorithm based on the loss value;

[0135] S504: Obtain a visualization of the output value of each activation function. If there is a non-smooth curve, adjust the KAN model structure until a smooth curve is obtained.

[0136] S505: Assign a corresponding activation function to each smooth curve based on its shape, and replace the original activation function with the newly assigned activation function.

[0137] S506: Retrain based on the updated model parameters and activation function until the iteration termination condition is met.

[0138] This can be understood as follows: since the KAN model places learnable activation functions on the edges (weights), the activation function of each edge is initialized first. In this embodiment, B-splines are used for initialization. Splines are piecewise polynomial functions that maintain high smoothness at the intersections (nodes) of polynomial blocks, thus providing better data fitting capabilities. Next, the standard process of inputting data, calculating loss, backpropagating based on the loss, and updating network parameters (weights, biases) is performed. Regularization techniques can be used to prevent overfitting, and pruning operations can be used to zero out unnecessary activation functions to simplify the model and improve generalization. Since each activation function can be visualized, its shape can be observed intuitively. For some non-smooth activation functions, the model structure can be adjusted to smooth them out without retraining the entire network. After continuous adjustments, a smooth curve for each activation function can be obtained, and a matching activation function can be directly assigned based on the curve's shape. In some embodiments, the best-fit mathematical function can be automatically selected to fit the curve shape by calculating the goodness of fit (e.g., R^2), thus ensuring the accuracy of activation function selection. Therefore, by continuously updating the model parameters and activation functions, the KAN model can be trained to better learn the complex nonlinear relationship between input and output variables, thereby improving the accuracy of gas saturation prediction.

[0139] Based on the gas saturation prediction method described above, this specification also provides a gas saturation prediction device. The device may include a system (including a distributed system), software (application), module, component, server, client, etc., using the method described in this specification, combined with necessary hardware implementation. Based on the same innovative concept, the devices in one or more embodiments provided in this specification are as described in the following embodiments. Since the implementation schemes and methods for solving the problem are similar, the implementation of specific devices in this specification can refer to the implementation of the aforementioned method, and repeated details will not be repeated. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0140] Specifically, Figure 6 This is a schematic diagram of the module structure of one embodiment of the gas saturation prediction device provided in this specification. (Refer to...) Figure 6 As shown in the embodiments of this specification, a gas saturation prediction device includes:

[0141] Acquisition module 601 is used to acquire pre-stack angle gather data of the target reservoir;

[0142] Inversion module 602 is used to perform pre-stack inversion on the pre-stack angle gather data using the zoeppritz equation to obtain the elastic parameters of the target reservoir;

[0143] The prediction module 603 is used to input the elastic parameters of the target reservoir into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on the KAN model, and the KAN model captures the nonlinear relationship between the elastic parameters and the gas saturation through a learnable activation function.

[0144] The beneficial effects obtained by the apparatus provided in the embodiments of this specification are consistent with the beneficial effects obtained by the methods described above, and will not be repeated here.

[0145] Reference Figure 7 As shown, based on the gas saturation prediction method described above, one embodiment of this specification also provides a computer device 702, wherein the above method operates on the computer device 702. The computer device 702 may include one or more processors 704, such as one or more central processing units (CPUs), each processing unit capable of implementing one or more hardware threads. The computer device 702 may also include any memory 706 for storing any kind of information such as code, settings, data, etc. Non-limitingly, for example, the memory 706 may include any type of RAM, any type of ROM, flash memory, hard disk, optical disk, etc. More generally, any memory can use any technology to store information. Further, any memory can provide volatile or non-volatile retention of information. Further, any memory can represent a fixed or removable component of the computer device 702. In one case, when the processor 704 executes associated instructions stored in any memory or combination of memories, the computer device 702 can perform any operation of the associated instructions. The computer device 702 also includes one or more drive mechanisms 708 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0146] Computer device 702 may also include an input / output module 710 (I / O) for receiving various inputs (via input device 712) and providing various outputs (via output device 714). A specific output mechanism may include a presentation device 716 and an associated graphical user interface (GUI) 718. In other embodiments, the input / output module 710 (I / O), input device 712, and output device 714 may be omitted, and the device may function solely as a computer device within a network. Computer device 702 may also include one or more network interfaces 720 for exchanging data with other devices via one or more communication links 722. One or more communication buses 724 couple the components described above together.

[0147] Communication link 722 can be implemented in any way, such as via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, or any combination thereof. Communication link 722 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0148] Corresponding to, for example Figures 1 to 5 In addition to the method shown, embodiments of this specification also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the above-described method.

[0149] This specification also provides computer-readable instructions, wherein when a processor executes the instructions, the program therein causes the processor to perform the following... Figures 1 to 5 The method shown.

[0150] This specification also provides a computer program product, including at least one instruction or at least one program segment, wherein the at least one instruction or the at least one program segment is loaded and executed by a processor to achieve the following: Figures 1 to 5 The method shown.

[0151] It should be understood that in the various embodiments of this specification, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this specification.

[0152] It should also be understood that, in the embodiments of this specification, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this specification generally indicates that the preceding and following related objects have an "or" relationship.

[0153] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this specification can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this specification.

[0154] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0155] In the several embodiments provided in this specification, it should be understood that the disclosed systems, apparatuses, and methods 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 couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, devices, or units, or they may be electrical, mechanical, or other forms of connection.

[0156] 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 the embodiments described in this specification, depending on actual needs.

[0157] Furthermore, the functional units in the various embodiments of this specification 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.

[0158] 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 specification, 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 described in the various embodiments of this specification. 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.

[0159] This specification uses specific embodiments to illustrate the principles and implementation methods of this specification. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this specification. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this specification. Therefore, the content of this specification should not be construed as a limitation of this specification.

Claims

1. A method for predicting gas saturation, characterized in that, The method includes: Acquire pre-stack corner gather data of the target reservoir; The elastic parameters of the target reservoir are obtained by pre-stack inversion of the pre-stack angle gather data using the zoeppritz equation. The elastic parameters of the target reservoir are input into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on the KAN model, which uses a learnable activation function to capture the nonlinear relationship between the elastic parameters and the gas saturation.

2. The method according to claim 1, characterized in that, The acquisition of pre-stack angle gather data of the target reservoir includes: Acquire pre-stack CMP gather data and forward modeling data of geological bodies for the target reservoir; The navigation pyramid method was used to perform multi-scale decomposition and denoising on the pre-stack CMP gather data to obtain denoised multi-scale seismic data. Based on the forward modeling characteristics of the geological body, angle gathers are extracted from the denoised multi-scale seismic data at scales that conform to the forward modeling characteristics of the geological body, thus obtaining pre-stack angle gather data.

3. The method according to claim 1, characterized in that, The process of using the Zoeppritz equation to perform pre-stack inversion on the pre-stack angle gather data to obtain the elastic parameters of the target reservoir includes: A priori distribution is established based on well logging data and forward modeling characteristics of the target reservoir; A likelihood function is constructed based on the zoeppritz equation and the pre-stack angle gather data. Calculate the posterior distribution value based on the prior distribution and the likelihood function; The maximum a posteriori estimate is obtained by sampling the posterior distribution value, and the maximum a posteriori estimate is used as the elasticity parameter estimate. Based on the estimated elastic parameters, determine whether the inversion process meets the preset convergence condition; If so, output the estimated value of the elasticity parameter; If not, calculate the output value of the Zoeppritz equation and the preset inversion objective function value based on the estimated value of the elastic parameters, and iteratively update the estimated value of the elastic parameters based on the output value of the Zoeppritz equation and the preset inversion objective function value until the preset convergence condition is met.

4. The method according to claim 1, characterized in that, The training process of the gas saturation prediction model includes: Acquire pre-stack CMP gather data, geological body forward modeling data, and well logging data for several reservoirs; The navigation pyramid method was used to perform multi-scale decomposition and denoising on the pre-stack CMP gather data to obtain denoised multi-scale seismic data. Based on the forward modeling characteristics of the geological body, select scale seismic data that conform to the forward modeling characteristics of the geological body from the denoised multi-scale seismic data and perform corner gather extraction to obtain pre-stack corner gather data; The elastic parameters are obtained by pre-stack inversion of the pre-stack angle gather data using the zoeppritz equation. Several elastic moduli are calculated based on the elastic parameters; Estimate gas saturation based on the well logging data; The correlation between each elastic modulus and the gas saturation is calculated. Elastic moduli with a correlation higher than a preset threshold are selected as training data. The corresponding gas saturation is used as the training label to train the KAN model and obtain a gas saturation prediction model. The KAN model captures the nonlinear relationship between the elastic parameters and the gas saturation through a learnable activation function.

5. The method according to claim 4, characterized in that, The estimation of gas saturation based on the well logging data includes: The gas saturation can be estimated using the following formula: Where a is the lithology coefficient, m is the cementation index, n is the saturation index, R0 is the formation resistivity, and R w φ represents the resistivity of formation water, and φ represents the porosity.

6. The method according to claim 4, characterized in that, The training process of the KAN model includes: The activation function of each edge in the KAN model is initialized; The training data is input into the KAN model, the output value is calculated through forward propagation, and the loss value is calculated based on the difference between the output value and the training label. The model parameters are updated using the backpropagation algorithm based on the loss value; Obtain a visualization of the output value of each activation function. If there is a non-smooth curve, adjust the structure of the KAN model until a smooth curve is obtained. Assign a corresponding activation function to each smooth curve based on its shape, and replace the original activation function with the newly assigned activation function. Retrain the model using the updated parameters and activation function until the iteration terminates.

7. A gas saturation prediction device, characterized in that, The device includes: The acquisition module is used to acquire pre-stack corner gather data of the target reservoir; The inversion module is used to perform pre-stack inversion on the pre-stack angle gather data using the zoeppritz equation to obtain the elastic parameters of the target reservoir. The prediction module is used to input the elastic parameters of the target reservoir into a pre-trained gas saturation prediction model to obtain the gas saturation of the target reservoir. The gas saturation prediction model is trained based on the KAN model, which uses a learnable activation function to capture the nonlinear relationship between the elastic parameters and the gas saturation.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes at least one instruction or at least one program segment, said at least one instruction or said at least one program segment being loaded and executed by a processor to implement the method as claimed in any one of claims 1 to 6.

Images