Reservoir lithology prediction method based on gaussian mixture model and electronic device

Abstract

The application discloses a reservoir lithology prediction method based on a Gaussian mixture model and electronic equipment, and the method comprises the following steps: obtaining multiple elastic parameters calculated by well logging according to well logging data; performing reservoir lithology category prediction training according to the multiple elastic parameters calculated by well logging and well logging lithology interpretation results, and obtaining a trained Gaussian mixture model; obtaining multiple elastic parameters calculated by prestack seismic inversion under well logging constraints; and obtaining a reservoir lithology category based on the trained Gaussian mixture model and the multiple elastic parameters calculated by prestack seismic inversion. The reservoir lithology prediction method based on the Gaussian mixture model uses elastic parameters calculated by well logging and interpretation results as training data, performs reservoir lithology category prediction training, obtains a trained Gaussian mixture model, and predicts a lithology category according to the trained Gaussian mixture model, so that the predicted lithology category is accurate, and the method is simple and easy to implement.

Description

Technical Field

[0001] This invention belongs to the field of seismic exploration and development of oil and gas and coalbed methane, and specifically relates to a reservoir lithology prediction method and electronic equipment based on a Gaussian mixture model. Background Technology

[0002] In the exploration and development of underground sedimentary minerals such as oil and coal, lithological studies are of paramount importance. However, because the target layer is buried deep underground, the research methods and approaches used differ significantly from those used in lithological studies of outcrop areas.

[0003] In subsurface lithological analysis, the target lithological markers can only be observed through rock data. However, well coring is generally not continuous, and the total coring rate of an exploration well is often only a few percent to a dozen percent, which poses a significant challenge to the study of lithological spatial distribution. While lithological analysis using well logging data can provide a continuous lithological interpretation of the entire well, it is highly susceptible to multiple interpretations. Therefore, in addition to the two types of data mentioned above, it is urgent to obtain more information from other data to improve the accuracy of lithological interpretation.

[0004] More importantly, even with sufficient data from single-well lithological analysis, traditional research methods only yield partial information, failing to utilize crucial details such as stratigraphic stacking patterns and sedimentary body shapes. Furthermore, even if the interpretation is entirely correct, it remains a "one-sided view." To further understand the planar distribution characteristics of lithology, a large number of sufficiently dense boreholes are required, which is difficult to achieve during the exploration phase. Therefore, there is an urgent need for new methods and techniques that can effectively grasp the planar variation characteristics of lithology using only a small number of boreholes.

[0005] Seismic reservoir lithology prediction arose precisely to meet these urgent needs. In oil exploration and the exploration of certain coalfields and salt mines, seismic exploration data is an indispensable and crucial foundational resource. This data is generally available in the early stages of exploration and typically covers the entire basin, containing extremely rich stratigraphic, structural, and sedimentary facies information; therefore, it is invaluable basic data for subsurface geological analysis.

[0006] Traditional seismic reservoir lithology prediction is based on seismic data, guided by geological principles, and constrained by drilling and logging data to study the spatial variation characteristics of lithology and reservoir properties in oil and gas-bearing reservoirs. There are two main approaches: One is based on seismic inversion, using cross-analysis of logging data to select sensitive elastic properties that are sensitive to the lithology of the target reservoir section and constructing classification criteria or standards based on lithology. Then, the sensitive elastic parameters are calculated using the seismic inversion results, and the spatial distribution characteristics of lithology are obtained using the classification criteria or standards. This method requires determining the classification criteria or standards, such as thresholds, during the cross-analysis stage. The other approach is based on impedance spectroscopy results, combined with lithology probability for geostatistical analysis to obtain a variogram, which is then used as an indicator to simulate the spatial distribution of lithology. This method essentially uses statistical graphs to obtain classification criteria or standards.

[0007] However, the two methods mentioned above are not accurate enough in obtaining classifications.

[0008] Therefore, an accurate lithology classification and prediction method is particularly needed. Summary of the Invention

[0009] The purpose of this invention is to propose an accurate method for predicting lithology classification.

[0010] In a first aspect, the present invention provides a reservoir lithology prediction method based on a Gaussian mixture model, comprising: acquiring well logging data and well logging lithology interpretation results; obtaining elastic parameters calculated from the well logging data, wherein the well logging lithology interpretation results include the lithology category of the reservoir; training the Gaussian mixture model for reservoir lithology category prediction based on the elastic parameters calculated from the well logging and the interpretation results, thereby obtaining a trained Gaussian mixture model; calculating the elastic parameters of the target reservoir based on pre-stack seismic inversion under well logging constraints; and obtaining the lithology category of the target reservoir based on the trained Gaussian mixture model and the elastic parameters of the target reservoir.

[0011] Optionally, the various elastic parameters include: sound wave velocity, transverse wave velocity, density, longitudinal wave impedance, transverse wave impedance, longitudinal-to-transverse wave velocity ratio, λρ, μρ, Poisson's ratio, and elastic wave impedance, where ρ is density, and λ and μ are Lamé constants, respectively.

[0012] Optionally, the trained Gaussian mixture model is obtained using the following method: Based on well logging data, multiple data points are obtained; the various elastic parameters calculated from the well logging corresponding to each data point are represented in vector form and used as model parameters; based on the well logging lithology interpretation results, the number of lithology categories is obtained; among the multiple data points, K data points are randomly selected, where K is the number of lithology categories, and each selected data point is used as an initial cluster center for one category; for each category, based on the model parameters of each data point, the initial covariance matrix between each data point and the cluster center of the category is calculated; based on the model parameters of each data point, the initial cluster centers and the initial covariance matrix are optimized and adjusted to obtain the final cluster centers and covariance matrix; based on the final cluster centers and covariance matrix, the trained Gaussian mixture model is obtained.

[0013] Optionally, the final cluster centers and covariance matrix can be obtained using the following steps: Step 1: For each category, based on the initial covariance matrix, calculate the membership degree of each data point to the category; based on the membership degree of each data point to the category, calculate the total membership degree of the data points belonging to the category; based on the total membership degree and the membership degree of each data point, calculate the new cluster centers; based on the new cluster centers, the total membership degree, and the membership degree of each data point, calculate the new covariance matrix between each data point and the cluster centers of the category; Step 2: Use the new cluster centers as the initial cluster centers and the new covariance matrix as the initial covariance matrix; Step 3: Repeat steps 1-2 until the cluster centers no longer change or the objective function value is less than a threshold; use the cluster centers and covariance matrix at this point as the final cluster centers and covariance matrix, and calculate the probability of each category appearing at this point.

[0014] Optionally, the trained Gaussian mixture model is:

[0015]

[0016] Among them, X i Let μ be the model parameter for the i-th data point. k ' is the final cluster center for the k-th lithological category, Σ k ' is the final covariance matrix of the k-th lithology category, where k is the k-th lithology category.

[0017] Optionally, the elastic parameters of the target reservoir are calculated according to the following steps: obtaining pre-stack inversion attribute parameters based on pre-stack seismic inversion under well logging constraints; and obtaining various elastic parameters calculated by pre-stack inversion based on the pre-stack inversion attribute parameters.

[0018] Optionally, based on the pre-stack seismic inversion under well logging constraints, multiple data points are determined, and the target reservoir elastic parameters of each data point are used as new model parameters. Based on the trained Gaussian mixture model and the new model parameters, the degree of membership of each data point to each category is calculated. From multiple degrees of membership, the maximum degree of membership is obtained, and the category corresponding to the maximum degree of membership is used as the reservoir lithology category corresponding to the data point.

[0019] Optionally, the degree of membership of a data point to a category can be calculated using the following formula:

[0020]

[0021] Where, r i,k p represents the degree of membership of the i-th data point to the k-th category. k μ is the probability of the k-th category occurring. k ' is the final cluster center for the k-th lithological category, Σ k ' is the final covariance moment for the k-th lithology category, p j Let N(X) be the probability of the j-th category appearing. i '|μ k ',Σ k ')for X i ' represents the model parameters for the i-th data point, N(X) i '|μ j ',Σ j ')for

[0022] Optionally, the objective function is:

[0023]

[0024] Among them, X j Let μ be the model parameter for the j-th data point. k S is the initial cluster center for the k-th lithological category. k Let K be the dataset for each category after classification, where each category represents a cluster, each cluster represents a type of lithology, k is the kth category, and K is the total number of categories.

[0025] Secondly, the present invention also provides an electronic device, the electronic device comprising: a memory storing executable instructions; and a processor that executes the executable instructions in the memory to implement the above-described reservoir lithology prediction method based on the Gaussian mixture model.

[0026] The beneficial effects of this invention are as follows: The reservoir lithology prediction method based on the Gaussian mixture model of this invention uses the elastic parameters and interpretation results calculated by well logging as training data to perform reservoir lithology category prediction training, obtain the trained Gaussian mixture model, and predict the lithology category based on the trained Gaussian mixture model. The predicted lithology category is accurate, and the method is simple and easy to implement.

[0027] The present invention has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0028] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the invention.

[0029] Figure 1 A flowchart of a reservoir lithology prediction method based on a Gaussian mixture model according to an embodiment of the present invention is shown. Detailed Implementation

[0030] Preferred embodiments of the invention will now be described in more detail. While preferred embodiments of the invention are described below, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0031] This invention provides a reservoir lithology prediction method based on a Gaussian mixture model, comprising: acquiring well logging data and well logging lithology interpretation results; obtaining elastic parameters calculated from the well logging data, wherein the well logging lithology interpretation results include the lithology category of the reservoir; training the Gaussian mixture model for reservoir lithology category prediction based on the elastic parameters calculated from the well logging and the interpretation results, thereby obtaining a trained Gaussian mixture model; calculating the elastic parameters of the target reservoir based on pre-stack seismic inversion under well logging constraints; and obtaining the lithology category of the target reservoir based on the trained Gaussian mixture model and the elastic parameters of the target reservoir.

[0032] Specifically, various elastic parameters calculated from well logging data are obtained, and the elastic parameters and interpretation results are used as training data to train reservoir lithology category prediction, thereby obtaining a trained Gaussian mixture model. The elastic parameters of the target reservoir calculated by pre-stack inversion are obtained based on well logging constraints. Finally, the lithology category is predicted using the elastic parameters of the target reservoir calculated by pre-stack inversion and the trained Gaussian mixture model.

[0033] According to an exemplary implementation, the reservoir lithology prediction method based on the Gaussian mixture model uses the elastic parameters and interpretation results calculated from well logging as training data to train the reservoir lithology category prediction, thereby obtaining the trained Gaussian mixture model. The lithology category is then predicted based on the trained Gaussian mixture model, and the predicted lithology category is accurate. Moreover, this method is simple and easy to implement.

[0034] As an optional option, various elastic parameters include: sound wave velocity, transverse wave velocity, density, longitudinal wave impedance, transverse wave impedance, longitudinal-to-transverse wave velocity ratio, λρ, μρ, Poisson's ratio, and elastic wave impedance, where ρ is density, and λ and μ are Lamé constants, respectively.

[0035] Specifically, the elastic parameters include 10 parameters such as sound wave velocity, transverse wave velocity, density, longitudinal wave impedance, transverse wave impedance, longitudinal wave velocity ratio, λρ, μρ, Poisson's ratio, and elastic wave impedance. These elastic parameters at each point form P-dimensional data at the data point, where P = 10.

[0036] As an optional approach, the trained Gaussian mixture model is obtained using the following method: Based on well logging data, multiple data points are obtained. The various elastic parameters calculated from the well logging for each data point are represented in vector form and used as model parameters. The number of lithology categories is obtained based on the well logging lithology interpretation results. K data points are randomly selected from the multiple data points, where K is the number of lithology categories. Each selected data point is used as an initial cluster center for one category. For each category, based on the model parameters for each data point, the initial covariance matrix between each data point and the cluster center of the category is calculated. Based on the model parameters for each data point, the initial cluster centers and the initial covariance matrix are optimized and adjusted to obtain the final cluster centers and covariance matrix. Based on the final cluster centers and covariance matrix, the trained Gaussian mixture model is obtained.

[0037] Assuming that well logging lithology interpretation results indicate that there are K lithology categories in the study area, meaning that the P-dimensional dataset X across all N well logging data points corresponds to K lithologies; then the well logging data can be represented as X ij Where i = 1, 2, ..., N represents the data point index, and j = 1, 2, ..., P represents the variable index (data dimension) of the data at each data point.

[0038] Suppose that the probability of a certain category k appearing in the entire category is: p k Then 0 < p k <1, and

[0039] Further assuming that the data in each category follows a Gaussian distribution, the probability that each data point belongs to each category, and the probability that the i-th data point belongs to the k-th category, is: Among them, N(X)i |μ k ,Σ k ), N(X i |μ j ,Σ j ) is a vector Gaussian function with the following form

[0040]

[0041] Where, μ k Let represent the cluster center of the k-th class, i.e., the mean, and Σ k Let μ represent the covariance matrix between the data points of the k-th class and the cluster centers. Then, all classes constitute a Gaussian mixture model, where the data of each class follows a Gaussian distribution, but its μ... k and Σ k different.

[0042] Specifically, from multiple data points, K data points are randomly selected, and each data point is used as the initial cluster center of a class, μ. k Let r be the initial cluster center of the k-th category; for each data point, let r be the degree of membership to each category. i,k Initialize to 0; based on the model parameters of each data point, calculate the initial covariance matrix of the cluster center of each data point and the category respectively. According to the model parameters of each data point, optimize and adjust the initial cluster center and the initial covariance matrix to obtain the final cluster center and covariance matrix, and then obtain the trained Gaussian mixture model.

[0043] As an optional approach, the following steps are used to obtain the final cluster centers and covariance matrix: Step 1: For each category, based on the initial covariance matrix, calculate the membership degree of each data point to the category. Based on the membership degree of each data point to the category, calculate the total membership degree of the data points belonging to the category. Based on the total membership degree and the membership degree of each data point, calculate the new cluster centers. Based on the new cluster centers, the total membership degree, and the membership degree of each data point, calculate the new covariance matrix between each data point and the cluster centers of the category. Step 2: Use the new cluster centers as the initial cluster centers and the new covariance matrix as the initial covariance matrix. Step 3: Repeat steps 1-2 until the cluster centers no longer change or the objective function value is less than the threshold. Use the cluster centers and covariance matrix at this point as the final cluster centers and covariance matrix, and calculate the probability of each category appearing at this point.

[0044] Specifically, (1) First, perform initialization settings:

[0045] ① From multiple data points, randomly select K data points, and use each data point as an initial cluster center for a category, μ k The cluster center of the k-th category;

[0046] ② For each data point, the degree of membership r of its belonging to each category i,k Initialize to 0;

[0047] ③ The probability of each category occurring is initialized to a uniform value: j = 1, 2, ..., K;

[0048] (2) Based on the cluster centers of each category, calculate the covariance matrix Σ between each data point and each cluster center. k Then, based on the covariance matrix, the membership degree of each data point to its category is calculated, and the probability that the i-th data point belongs to the k-th category is:

[0049]

[0050] (3) ① Based on the r calculated in step 2 i,k Calculate for each category k = 1, 2, ..., K;

[0051] ② For each category, calculate its cluster center: k = 1, 2, ..., K

[0052] ③ For each category, calculate the covariance matrix between each data point and the cluster center:

[0053] ④ Calculate the probability of occurrence for each category: k = 1, 2, ..., K

[0054] (4) Substitute the new cluster centers and covariance matrix calculated in step 3 into step 2 to calculate the new membership degree. Then substitute the new membership degree and cluster centers into step 3 again to calculate the new cluster centers and covariance matrix. Repeat steps 2 and 3 until the cluster centers no longer change or the objective function is no longer valid. It is less than a certain threshold. This determines the final cluster centers and the new covariance matrix.

[0055] The probability of each category is calculated based on the final cluster center and the degree of membership of each data point to the category corresponding to the new covariance matrix.

[0056] As an alternative, the trained Gaussian mixture model is as follows:

[0057]

[0058] Among them, X i Let μ be the model parameter for the i-th data point. k ' is the final cluster center for the k-th lithological category, Σ k' is the final covariance matrix of the k-th lithology category, where k is the k-th lithology category.

[0059] As an alternative, the elastic parameters of the target reservoir are calculated according to the following steps: obtain the pre-stack inversion attribute parameters based on the pre-stack seismic inversion under well logging constraints; obtain various elastic parameters calculated by the pre-stack inversion based on the pre-stack inversion attribute parameters.

[0060] Specifically, pre-stack inversion under well logging constraints is performed, and the inversion results are used to calculate various elastic property data. This includes conventional pre-stack inversion procedures and steps, such as seismic data structural interpretation, well-seismic calibration, wavelet extraction, pre-stack AVO inversion, and pre-stack elastic wave impedance inversion. Using the same calculation formulas as those used for various elastic parameters calculated by well logging, various elastic parameters are calculated from the inversion results to obtain the elastic parameter data volume to be classified.

[0061] As an alternative, multiple data points are determined based on pre-stack seismic inversion under well logging constraints. The target reservoir elastic parameters of each data point are used as new model parameters. Based on the trained Gaussian mixture model and the new model parameters, the degree of membership of each data point to each category is calculated. From multiple degrees of membership, the maximum degree of membership is obtained, and the category corresponding to the maximum degree of membership is used as the reservoir lithology category corresponding to the data point.

[0062] Specifically, multiple elastic parameters calculated by pre-stack inversion are used as new model parameters, and the new model parameters and the trained Gaussian mixture model are substituted into the formula. Calculate the degree of membership of each data point to each category, and take the category corresponding to the maximum value among these K membership degrees as the lithology category corresponding to the data point, or directly output the maximum lithology probability corresponding to the data point.

[0063] As an optional approach, the degree of membership of data points to a category can be calculated using the following formula:

[0064]

[0065] Where, r i,k p represents the degree of membership of the i-th data point to the k-th category. k μ is the probability of the k-th category occurring. k ' is the final cluster center for the k-th lithological category, Σ k ' is the final covariance moment for the k-th lithology category, p j Let N(X) be the probability of the j-th category appearing. i '|μ k ',Σ k ')for X i 'Model parameters for the i-th data point, N(X)i '|μ j ',Σ j ')for

[0066] As an optional approach, the objective function is:

[0067]

[0068] Among them, X j Let μ be the model parameter for the j-th data point. k S is the initial cluster center for the k-th lithological category. k Let K be the dataset for each category after classification, where each category represents a cluster, each cluster represents a type of lithology, k is the kth category, and K is the total number of categories.

[0069] Example 1

[0070] Figure 1 A flowchart of a reservoir lithology prediction method based on a Gaussian mixture model according to an embodiment of the present invention is shown.

[0071] like Figure 1 This reservoir lithology prediction method based on the Gaussian mixture model includes:

[0072] Step 1: Obtain well logging data and well logging lithology interpretation results, and obtain the elastic parameters calculated from the well logging data. The well logging lithology interpretation results include the lithology category of the reservoir.

[0073] Step 2: Based on the elastic parameters calculated from the well logging and the interpretation results, train the Gaussian mixture model to predict reservoir lithology categories, and obtain the trained Gaussian mixture model;

[0074] Step 3: Calculate the elastic parameters of the target reservoir based on the pre-stack seismic inversion under logging constraints;

[0075] Step 4: Based on the trained Gaussian mixture model and the elastic parameters of the target reservoir, obtain the lithology category of the target reservoir.

[0076] Among them, various elastic parameters include: sound wave velocity, transverse wave velocity, density, longitudinal wave impedance, transverse wave impedance, longitudinal-to-transverse wave velocity ratio, λρ, μρ, Poisson's ratio, and elastic wave impedance, where ρ is density, and λ and μ are Lamé constants.

[0077] The trained Gaussian mixture model was obtained using the following method: Multiple data points were obtained from well logging data. The various elastic parameters calculated from the well logging for each data point were represented in vector form and used as model parameters. The number of lithology categories was obtained based on the well logging lithology interpretation results. K data points were randomly selected from the multiple data points, where K is the number of lithology categories. Each selected data point was used as an initial cluster center for one category. For each category, based on the model parameters for each data point, the initial covariance matrix between each data point and the cluster center of the category was calculated. Based on the model parameters for each data point, the initial cluster centers and the initial covariance matrix were optimized and adjusted to obtain the final cluster centers and covariance matrix. Based on the final cluster centers and covariance matrix, the trained Gaussian mixture model was obtained.

[0078] The final cluster centers and covariance matrix are obtained using the following steps: Step 1: For each category, based on the initial covariance matrix, calculate the membership degree of each data point to the category. Based on the membership degree of each data point to the category, calculate the total membership degree of the data points belonging to the category. Based on the total membership degree and the membership degree of each data point, calculate the new cluster centers. Based on the new cluster centers, the total membership degree, and the membership degree of each data point, calculate the new covariance matrix between each data point and the cluster centers of the category. Step 2: Use the new cluster centers as the initial cluster centers and the new covariance matrix as the initial covariance matrix. Step 3: Repeat steps 1-2 until the cluster centers no longer change or the objective function value is less than the threshold. Use the cluster centers and covariance matrix at this point as the final cluster centers and covariance matrix, and calculate the probability of each category appearing at this point.

[0079] The trained Gaussian mixture model is as follows:

[0080]

[0081] Among them, X i Let μ be the model parameter for the i-th data point. k ' is the final cluster center for the k-th lithological category, Σ k ' is the final covariance matrix of the k-th lithology category, where k is the k-th lithology category.

[0082] The elastic parameters of the target reservoir are calculated according to the following steps: obtaining pre-stack inversion attribute parameters based on pre-stack seismic inversion under well logging constraints; and obtaining various elastic parameters calculated by pre-stack inversion based on the pre-stack inversion attribute parameters.

[0083] In this process, multiple data points are determined based on pre-stack seismic inversion under well logging constraints. The target reservoir elastic parameters of each data point are used as new model parameters. Based on the trained Gaussian mixture model and the new model parameters, the degree of membership of each data point to each category is calculated. From multiple degrees of membership, the maximum degree of membership is obtained, and the category corresponding to the maximum degree of membership is used as the reservoir lithology category corresponding to the data point.

[0084] The degree of membership of a data point to a category is calculated using the following formula:

[0085]

[0086] Where, r i,k p represents the degree of membership of the i-th data point to the k-th category. k μ is the probability of the k-th category occurring. k ' is the final cluster center for the k-th lithological category, Σ k ' is the final covariance moment for the k-th lithology category, p j Let N(X) be the probability of the j-th category appearing. i '|μ k ',Σ k ')for X i ' represents the model parameters for the i-th data point, N(X) i '|μ j ',Σ j ')for

[0087] The objective function is:

[0088]

[0089] Among them, X j Let μ be the model parameter for the j-th data point. k S is the initial cluster center for the k-th lithological category. k Let K be the dataset for each category after classification, where each category represents a cluster, each cluster represents a type of lithology, k is the kth category, and K is the total number of categories.

[0090] Example 2

[0091] This disclosure provides an electronic device comprising: a memory storing executable instructions; and a processor executing the executable instructions in the memory to implement the aforementioned reservoir lithology prediction method based on a Gaussian mixture model.

[0092] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.

[0093] This memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0094] The processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of this disclosure, the processor is used to execute computer-readable instructions stored in the memory.

[0095] Those skilled in the art should understand that, in order to solve the technical problem of how to obtain a good user experience, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included within the protection scope of this disclosure.

[0096] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0097] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method of reservoir lithology prediction based on Gaussian mixture model, characterized in that, The method comprises the following steps: obtaining well logging data and well logging lithology interpretation results, and obtaining elastic parameters calculated by well logging according to the well logging data, wherein the well logging lithology interpretation results comprise a lithology category of a reservoir; training a Gaussian mixture model for reservoir lithology category prediction according to the elastic parameters calculated by well logging and the interpretation results, and obtaining a trained Gaussian mixture model; calculating elastic parameters of a target reservoir according to pre-stack seismic inversion under well logging constraints; obtaining a lithology category of the target reservoir based on the trained Gaussian mixture model and the elastic parameters of the target reservoir; wherein the trained Gaussian mixture model is obtained by the following method: obtaining a plurality of data points according to the well logging data, and representing the elastic parameters calculated by well logging corresponding to each data point in a vector form as model parameters, obtaining the number of lithology categories according to the well logging lithology interpretation results, randomly selecting K data points from the plurality of data points, wherein K is the number of lithology categories, and taking each selected data point as an initial cluster center of a category, for each category, calculating an initial covariance matrix of each data point and the cluster center of the category based on the model parameters of each data point, optimizing and adjusting the initial cluster center and the initial covariance matrix based on the model parameters of each data point to obtain a final cluster center and a covariance matrix, obtaining the trained Gaussian mixture model based on the final cluster center and the covariance matrix, wherein the final cluster center and the covariance matrix are obtained by the following steps: Step 1: for each category, calculating the membership degree of each data point belonging to the category according to the initial covariance matrix, calculating the total membership degree of the data points belonging to the category according to the membership degree of each data point belonging to the category, and calculating a new cluster center according to the total membership degree and the membership degree of each data point; calculating a new covariance matrix of each data point and the cluster center of the category according to the new cluster center, the total membership degree and the membership degree of each data point; Step 2: taking the new cluster center as the initial cluster center and taking the new covariance matrix as the initial covariance matrix; Step 3: repeating steps 1-2 until the cluster center no longer changes or the target function value is less than a threshold, taking the cluster center and the covariance matrix at this time as the final cluster center and the covariance matrix, and calculating the probability of each category appearing at this time.

2. The method of claim 1, wherein the Gaussian mixture model-based reservoir lithology prediction method is characterized by, The elastic parameters comprise: acoustic velocity, shear wave velocity, density, P-wave impedance, S-wave impedance, P-S wave velocity ratio, λρ, μρ, Poisson's ratio and elastic wave impedance, wherein ρ is density, and λ and μ are Lame constants.

3. The method of claim 1, wherein the Gaussian mixture model-based reservoir lithology prediction method is characterized by, The trained Gaussian mixture model is: where X i is the model parameter of the i-th data point, μ k is the final cluster center of the k-th lithology class, ∑ k is the final covariance matrix of the k-th lithology class, and k is the k-th lithology class.

4. The method of claim 3, wherein the Gaussian mixture model-based reservoir lithology prediction method is characterized by, The elastic parameters of the target reservoir are calculated according to the following steps: obtaining pre-stack inversion attribute parameters according to pre-stack seismic inversion under well logging constraints; obtaining a plurality of elastic parameters calculated by pre-stack inversion according to the pre-stack inversion attribute parameters.

5. The method of claim 4, wherein the Gaussian mixture model-based reservoir lithology prediction method is characterized by, According to the pre-stack seismic inversion under the well logging constraint, a plurality of data points are determined, target reservoir elastic parameters of each data point are taken as new model parameters, membership degrees of each data point belonging to each category are calculated according to the trained Gaussian mixture model and the new model parameters, a maximum membership degree is obtained from the plurality of membership degrees, and a category corresponding to the maximum membership degree is taken as a reservoir lithology category corresponding to the data point.

6. The method of reservoir lithology prediction based on Gaussian mixture model according to claim 5, characterized in that, The membership degree of the data point belonging to the category is calculated by using the following formula: where r i,k is the membership of the ith data point to the kth class, p k is the probability of the kth class to occur, μ k is the final cluster center of the kth lithology class, ∑ k is the final covariance matrix of the kth lithology class, p j is the probability of the jth class to occur, N(X i | μ k ,∑ k ) is the probability density function of the jth class X i is the model parameter of the ith data point, N(X i | μ j ,∑ j ) is the probability density function of the jth class 7. The method of claim 1, wherein the Gaussian mixture model-based reservoir lithology prediction method is characterized by, The objective function is: where X j is the model parameter of the jth data point, μ k is the initial cluster center of the kth lithology class, S k is the data set of each class after classification, where each class represents a cluster, each cluster represents a lithology class, k is the kth class, and K is the total number of classes.

8. An electronic device, comprising: The electronic device comprises: a memory, which stores executable instructions; a processor, which runs the executable instructions in the memory to implement the reservoir lithology prediction method based on the Gaussian mixture model according to any one of claims 1-7.

Images