Optimal reservoir classification number permeability prediction method, system, medium, and apparatus

Abstract

The present application relates to a permeability prediction method, system, medium and equipment of optimal reservoir classification number, comprising: collecting data and preprocessing to obtain an initial unlabeled data set; clustering analysis is carried out on the initial unlabeled data set to determine the initial classification number; taking the initial classification number as the midpoint, expanding to both sides according to the expansion quantity n, forming a candidate classification number set; constructing a labeled data set based on the candidate classification number set; constructing a multi-classification number permeability calculation model according to the labeled data set; establishing a machine learning reservoir type prediction model and predicting the reservoir type; after the reservoir type prediction is completed, the optimal classification number is determined; based on the optimal classification number, the final reservoir type prediction model is established, and the classification number permeability calculation model is embedded in the final reservoir type prediction model for prediction. The present application realizes the optimal balance of reservoir classification number and prediction accuracy through the establishment of a closed-loop feedback mechanism of "candidate classification number generation-classification modeling verification-prediction accuracy optimization".

Description

Technical Field

[0001] This invention relates to the field of oil and gas exploration technology, and more specifically, to a method, system, medium, and device for predicting permeability based on the optimal reservoir classification number. Background Technology

[0002] Reservoir permeability is a core parameter for oil and gas reservoir evaluation, directly determining production capacity assessment and development plan design. For heterogeneous reservoirs, due to complex sedimentary environments and significant differences in diagenesis, the reservoirs are highly heterogeneous, making it difficult to establish a unified global permeability calculation model using traditional methods. Machine learning has emerged as a new solution.

[0003] Currently, the industry mainstream adopts two technical approaches for permeability prediction using machine learning: one is to directly predict permeability using well logging data; the other is a two-stage strategy of classification followed by prediction. The former is more suitable for homogeneous formations, while the latter is more suitable for heterogeneous low-permeability sandstone reservoirs. For the latter, the first and most crucial step is reservoir classification. Common methods for reservoir classification are based on core data (porosity, permeability, saturation, etc.) or well logging curves (GR, DEN, CNL, etc.), using methods such as lithology, sedimentary facies, diagenetic facies, petrological facies, well logging facies, or flow units to classify reservoirs into several types. Simultaneously, a classification permeability model is established based on the classification results, and the determination of the number of classifications (K) mainly relies on statistical indicators. The second stage is reservoir type prediction. A reservoir type prediction model (such as XGBoost, random forest, neural network, etc.) is established to predict the reservoir type, and finally, the reservoir permeability is calculated using the corresponding permeability model based on different reservoir types. Existing technologies have the following problems. First, the reservoir classification stage only pursues geological differences, or mathematically speaking, intra-class compactness and inter-class separation, while the prediction stage pursues the accuracy of permeability calculation. These differing optimization objectives mean that the optimal K for classification may not be the optimal K for prediction. Second, there is a lack of feedback verification mechanisms. Existing methods determine the K value all at once, without considering the impact of different initial K values ​​on the final permeability calculation task, and cannot assess the sensitivity of the number of classifications. Furthermore, for small sample core data, an excessive number of classifications can lead to insufficient training samples for some categories, overfitting of the machine learning model, and a lack of constraints in determining the number of classifications.

[0004] The more types of reservoirs there are, the more refined the classification permeability model becomes, but the corresponding machine learning reservoir type prediction accuracy will decrease, leading to a decrease in permeability prediction accuracy. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, system, medium and device for predicting the permeability of the optimal reservoir classification number, in view of the problems existing in the prior art.

[0006] The technical solution adopted by this invention to solve its technical problem is: to construct a permeability prediction method for optimal reservoir classification number, including the following steps: Step S10: Collect all core conventional physical property analysis data and corresponding depth logging curve data of the target area, and perform data preprocessing to obtain the initial unlabeled dataset; Step S20: Perform cluster analysis on the initial unlabeled dataset using a clustering algorithm to determine the initial number of categories; Step S30: Using the initial number of categories as the midpoint, expand outwards to both sides according to the expansion amount n to form a set of candidate categories; Step S40: Construct a labeled dataset based on the set of candidate classification numbers; Step S50: Construct a multi-class penetration rate calculation model based on the labeled dataset; Step S60: Establish a corresponding machine learning reservoir type prediction model for the multi-class permeability calculation model, and perform reservoir type prediction; Step S70: After reservoir type prediction is completed, determine the optimal number of classifications; Step S80: Establish the final reservoir type prediction model based on the optimal number of classifications, and embed the number of classifications permeability calculation model into the final reservoir type prediction model for prediction.

[0007] In the permeability prediction method for the optimal reservoir classification number described in this invention, step S30 includes: Determine the expansion quantity n; Using the initial number of categories as the midpoint, expand n units to both sides to form the candidate number of categories set.

[0008] In the permeability prediction method for the optimal reservoir classification number described in this invention, the expansion amount n is determined through the following steps: Determine the sample size and sample size weighting coefficients; Calculate the heterogeneity weighting coefficient; Calculate the cluster stability weight coefficients; Calculate the reservoir heterogeneity strength coefficient; Determine the cluster stability compensation term; The expansion quantity n is obtained by combining the sample size and sample size weight coefficient, the heterogeneity weight coefficient, the cluster stability weight coefficient, the reservoir heterogeneity intensity coefficient, and the cluster stability compensation term.

[0009] In the permeability prediction method for the optimal reservoir classification number described in this invention, step S40 includes: Based on the Gaussian mixture model, the core samples were classified into multiple reservoir types; A labeled dataset is constructed based on the division of multiple reservoir types; in the labeled dataset, well logging data is the feature and reservoir type is the label.

[0010] In the permeability prediction method for the optimal reservoir classification number described in this invention, step S60 includes: For each of the multi-class permeability calculation models, a corresponding machine learning reservoir type prediction model is established. For different types of reservoirs, the corresponding machine learning reservoir type prediction model is used to predict the reservoir type.

[0011] In the permeability prediction method for the optimal reservoir classification number described in this invention, step S70 includes: After the reservoir type is predicted, the permeability is calculated using the corresponding permeability calculation model to obtain the corresponding predicted permeability. The predicted permeability is compared with the actual permeability of the core sample to calculate the corresponding mean absolute logarithmic error of the permeability. The optimal number of classifications is determined based on the mean absolute logarithmic error of the penetration rate.

[0012] In the permeability prediction method for the optimal reservoir classification number described in this invention, step S80 includes: The final dataset is established based on the reservoir type corresponding to the optimal number of classifications. Based on the final dataset, a machine learning algorithm is used to train and construct the final reservoir type prediction model; The classification number permeability calculation model is embedded into the final reservoir type prediction model for prediction.

[0013] This invention also provides a permeability prediction system for the optimal reservoir classification number, comprising: The data collection and preprocessing unit is used to collect all conventional physical property analysis data of core samples and corresponding depth logging curve data of the target area, and to perform data preprocessing to obtain an initial unlabeled dataset; The clustering analysis calculation unit is used to perform clustering analysis on the initial unlabeled dataset using a clustering algorithm to determine the initial number of categories; The classification number set construction unit is used to expand n units to both sides of the initial classification number as the midpoint to form a candidate classification number set; A labeled dataset construction unit is used to construct a labeled dataset based on the set of candidate categories. A penetration rate model building unit is used to build a multi-class penetration rate calculation model based on the labeled dataset. The reservoir type prediction model construction and prediction unit is used to build a corresponding machine learning reservoir type prediction model based on the multi-classification penetration calculation model and to predict the reservoir type. The optimal number of classifications determination unit is used to determine the optimal number of classifications after reservoir type prediction is completed; The final model building and prediction unit is used to build the final reservoir type prediction model based on the optimal number of classifications, and to embed the classification number permeability calculation model into the final reservoir type prediction model for prediction.

[0014] The present invention also provides a storage medium storing a computer program adapted for loading by a processor to perform the steps of the permeability prediction method for the optimal reservoir classification number as described above.

[0015] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the permeability prediction method for the optimal reservoir classification number as described above by calling the computer program stored in the memory.

[0016] The method, system, medium, and device for predicting permeability based on the optimal number of reservoir classifications according to this invention have the following beneficial effects: They include: collecting and preprocessing data to obtain an initial unlabeled dataset; performing cluster analysis on the initial unlabeled dataset to determine the initial number of classifications; expanding outwards from the initial number of classifications as the midpoint by an expansion amount n to form a candidate number of classifications set; constructing a labeled dataset based on the candidate number of classifications set; constructing a multi-class permeability calculation model based on the labeled dataset; establishing a machine learning reservoir type prediction model and predicting reservoir types; determining the optimal number of classifications after reservoir type prediction is completed; establishing a final reservoir type prediction model based on the optimal number of classifications, and embedding the permeability calculation model based on the number of classifications into the final reservoir type prediction model for prediction. This invention achieves an optimal balance between the number of reservoir classifications and prediction accuracy by establishing a closed-loop feedback mechanism of "candidate number of classifications generation—classification modeling verification—prediction accuracy optimization." Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart illustrating the permeability prediction method for the optimal reservoir classification number provided by the present invention. Figure 2 This is a schematic diagram of the elbow rule for fuzzy C-means clustering provided by the present invention; Figure 3 This is a schematic diagram of the fuzzy C-mean profile coefficient provided by the present invention; Figure 4 This invention provides a diagram showing the relationship between porosity and permeability in core analysis. =4); Figure 5 This invention provides a diagram showing the relationship between porosity and permeability in core analysis. =5); Figure 6 This invention provides a diagram showing the relationship between porosity and permeability in core analysis. =6); Figure 7 This invention provides a diagram showing the relationship between porosity and permeability in core analysis. =7); Figure 8 This invention provides a diagram showing the relationship between porosity and permeability in core analysis. =8); Figure 9 The permeability prediction accuracy provided by this invention ( A schematic diagram showing (=4, MALE = 0.72); Figure 10 The permeability prediction accuracy provided by this invention ( A schematic diagram showing (=5, MALE = 0.57); Figure 11 The permeability prediction accuracy provided by this invention ( A schematic diagram showing (=6, MALE = 0.63); Figure 12 The permeability prediction accuracy provided by this invention ( A schematic diagram showing (=7, MALE = 0.52); Figure 13 The permeability prediction accuracy provided by this invention ( A schematic diagram showing (=8, MALE = 0.65); Figure 14 This is a graph showing the permeability calculation results of well A provided by the present invention; Figure 15 This is a logic block diagram of the permeability prediction system for the optimal reservoir classification number provided by the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To address the challenge of balancing the number of reservoir classifications with prediction accuracy, this invention provides a permeability prediction method based on the optimal number of reservoir classifications. By establishing a closed-loop feedback mechanism of "candidate classification generation—classification modeling verification—prediction accuracy optimization," the optimal balance between the number of classifications and prediction accuracy is achieved.

[0020] Specifically, such as Figure 1 As shown, in a preferred embodiment, the permeability prediction method for the optimal reservoir classification number includes the following steps: Step S10: Collect all core conventional physical property analysis data and corresponding depth logging curve data of the target area, and perform data preprocessing to obtain an initial unlabeled dataset.

[0021] Specifically, all core conventional physical property analysis data and corresponding depth logging curve data from the study area (i.e., the target area) were collected. The data underwent preprocessing such as cleaning and overburden correction to obtain an initial unlabeled dataset. The core conventional physical property analysis data included, but was not limited to, porosity. φ Penetration rate k Features included in the initial unlabeled dataset are: GR (natural gamma curve), CNL (compensated neutron curve), DEN (density curve), RT (undisturbed formation resistivity curve), and RXO (flushed zone resistivity curve). φ and k .

[0022] Step S20: Use a clustering algorithm to perform cluster analysis on the initial unlabeled dataset to determine the initial number of categories.

[0023] In this embodiment of the invention, an unsupervised fuzzy C-means clustering algorithm (FCM) is used for cluster analysis, and the initial number of clusters (K0) is determined by the elbow rule.

[0024] Step S30: Using the initial number of categories as the midpoint, expand to both sides according to the expansion amount n to form a set of candidate categories.

[0025] Specifically, in step S30, the expansion amount n is first determined, and then, taking the initial number of categories K0 as the midpoint, n units are expanded to both sides to form a candidate category set. The candidate category set can be represented as: K set = {K 0-n , ..., K0, ..., K 0+n}

[0026] Where the expansion quantity n is a positive integer dynamically determined based on the sample size and heterogeneity.

[0027] In this embodiment of the invention, the expansion quantity n is determined through the following steps: determining the sample size and sample size weight coefficients; calculating the heterogeneity weight coefficients; calculating the clustering stability weight coefficients; calculating the reservoir heterogeneity intensity coefficients; determining the clustering stability compensation term; and comprehensively calculating the expansion quantity n by combining the sample size and sample size weight coefficients, heterogeneity weight coefficients, clustering stability weight coefficients, reservoir heterogeneity intensity coefficients, and clustering stability compensation term. The mathematical expression for the expansion quantity n is: In the formula: The sample size weighting coefficient is used. ; N is the sample size; These are weighting coefficients for heterogeneity. ; This represents the cluster stability weight coefficient. ; This represents the reservoir heterogeneity strength coefficient. ; This is a clustering stability compensation term. .

[0028] Among them, the reservoir heterogeneity strength coefficient The calculation is as follows: In the formula: , As weight, ; The permeability heterogeneity coefficient; The standard deviation of the penetration rate; The porosity heterogeneity coefficient; Porosity confidence level; This represents the average porosity.

[0029] Cluster stability compensation term It can be determined in the following ways: In the formula, This refers to the profile coefficient.

[0030] The constraints are: .

[0031] The boundary constraints are: ,in, It is a set of positive integers.

[0032] It should be noted that, in this embodiment of the invention, after generating the candidate classification set, it is also necessary to determine whether it meets the constraints. If it does not meet the constraints, return to step S20 to perform cluster analysis again; when the constraints are met, execute step S40.

[0033] Step S40: Construct a labeled dataset based on the set of candidate class numbers.

[0034] In this embodiment of the invention, step S40 includes: dividing the core sample into multiple reservoir types based on a Gaussian mixture model; constructing a labeled dataset based on the divided reservoir types; in the labeled dataset, well logging data are features and reservoir types are labels.

[0035] Specifically, for the candidate classification set K set In each K i (i.e., K) 0-n , ..., K0, ..., K 0+n First, based on the Gaussian mixture model, the core samples were divided into K... i Based on the reservoir types, a 2n+1 labeled dataset is then created. In this labeled dataset, well logging data serves as the feature, and reservoir type is the label.

[0036] Step S50: Construct a multi-class penetration calculation model based on the label dataset.

[0037] Specifically, for the labeled dataset in step S40, a penetration rate calculation model is established for each category based on the data labels, and the specific expression is as follows: In the formula: a and b These are the fitting coefficients. i = 1, 2, ..., m, m = , where represents the number of reservoir types; Let be the permeability of the i-th type of reservoir; Effective porosity.

[0038] Step S60: Establish a corresponding machine learning reservoir type prediction model for the multi-class permeability calculation model, and perform reservoir type prediction.

[0039] In this embodiment of the invention, step S60 includes: establishing corresponding machine learning reservoir type prediction models for the multi-class permeability calculation model; and predicting reservoir type using the corresponding machine learning reservoir type prediction models for different types of reservoirs.

[0040] Specifically, a machine learning reservoir type prediction model is established by classifying the data. The dataset is divided into a training set and a test set in a 7:3 ratio to predict the reservoir type. The prediction accuracy should be greater than 80%.

[0041] Step S70: After the reservoir type prediction is completed, determine the optimal number of classifications.

[0042] In this embodiment of the invention, step S70 includes: after the reservoir type prediction is completed, permeability is calculated using the corresponding permeability calculation model to obtain the corresponding predicted permeability; the predicted permeability is compared with the actual permeability of the core to calculate the corresponding mean absolute logarithmic error of permeability; and the optimal number of classifications is determined based on the mean absolute logarithmic error of permeability.

[0043] Specifically, after the reservoir type prediction is completed, the permeability is calculated according to the corresponding model classification, and compared with the actual permeability of the core sample to calculate the mean absolute logarithmic error of the permeability: In the formula: C represents the number of core points; This represents the true core permeability, in mD. To predict permeability, mD; MALE represents the mean absolute logarithmic error of permeability.

[0044] Compare each of MALE Choose the one with the lowest error The optimal number of classifications is determined by the number of classifications. If the errors are the same, the number of classifications is selected. The smaller number is used as the final classification number.

[0045] Step S80: Establish the final reservoir type prediction model based on the optimal number of classifications, and embed the number of classifications permeability calculation model into the final reservoir type prediction model for prediction.

[0046] In this embodiment of the invention, step S80 includes: establishing a final dataset based on the reservoir type corresponding to the optimal number of classifications; using the XGBoost machine learning algorithm to train and construct a final reservoir type prediction model based on the final dataset; and embedding the classification number penetration calculation model into the final reservoir type prediction model for prediction.

[0047] Specifically, reservoir types are reclassified based on the optimal number of classifications, a final reservoir type prediction model is trained, and a classification permeability calculation model is embedded. This is applied to logging data across the entire well section to predict and output reservoir types and continuous permeability curves.

[0048] Compared with traditional solutions, the present invention has the following advantages: The optimal number of classifications can be adjusted based on reservoir type and homogeneity. By using permeability calculation accuracy feedback for screening, classification schemes that are statistically reasonable but have low permeability calculation accuracy are eliminated. Compared with the fixed number of classifications method, the midpoint expansion strategy can reduce the permeability prediction error. Compared with global classification search, the computational load is reduced by more than 60%, while covering the possible optimal range.

[0049] refer to Figure 15 The present invention also provides a permeability prediction system for the optimal number of reservoir classifications.

[0050] Specifically, such as Figure 15 As shown, the permeability prediction system for the optimal reservoir classification number includes: The data collection and preprocessing unit 10 is used to collect all the conventional physical property analysis data of the core samples and the corresponding depth logging curve data of the target area, and to perform data preprocessing to obtain the initial unlabeled dataset.

[0051] The clustering analysis calculation unit 20 is used to perform clustering analysis on the initial unlabeled dataset using a clustering algorithm to determine the initial number of categories.

[0052] The classification number set construction unit 30 is used to expand n units to both sides of the initial classification number as the midpoint to form a candidate classification number set.

[0053] Labeled dataset building unit 40 is used to build a labeled dataset based on the set of candidate class numbers.

[0054] The penetration rate model building unit 50 is used to build a multi-class penetration rate calculation model based on the label dataset.

[0055] The type prediction model construction and prediction unit 60 is used to build a corresponding machine learning reservoir type prediction model based on the multi-classification penetration calculation model and to predict the reservoir type.

[0056] The optimal number of classifications determination unit 70 is used to determine the optimal number of classifications after the reservoir type prediction is completed.

[0057] The final model building and prediction unit 80 is used to build the final reservoir type prediction model based on the optimal number of classifications, and to embed the classification number permeability calculation model into the final reservoir type prediction model for prediction.

[0058] Specifically, the specific coordination process between the units in the optimal reservoir classification number permeability prediction system can be referred to the above-mentioned optimal reservoir classification number permeability prediction method, and will not be repeated here.

[0059] The permeability prediction method for the optimal reservoir classification number of the present invention will be described below with specific embodiments.

[0060] Taking oil field A in a certain basin as an example, the invention is described in detail, and the specific implementation steps are as follows: Step 1: Data collection and preprocessing.

[0061] A total of 994 core samples from 36 wells in the study area were collected, and routine physical property analysis data, including porosity, were obtained from the core samples. φ Penetration rate k Anomalies (mudstone points, core fragmentation) and outliers in the cores were removed, resulting in a total of 757 core points. Preprocessing of the core porosity and permeability, including overburden correction and depth correction, was performed. Well logging data at corresponding depths were extracted to obtain an initial unlabeled dataset. The dataset features included GR, CNL, DEN, RT, RXO, etc. φ and k .

[0062] Step 2: Determine the initial number of categories.

[0063] Unsupervised fuzzy C-means clustering (FCM) algorithm is used for cluster analysis: ; In the formula: Let n be the objective function of FCM; n is the total number of samples. Let be the i-th sample; k is the total number of clusters. Let j be the j-th cluster center.

[0064] Calculate the sum of squared errors within the class: ; In the formula: SSE The sum of squared errors within each cluster; k is the total number of clusters; Let x be the i-th cluster; x represents the cluster to which x belongs. Sample data points; For the i-th cluster Cluster centers.

[0065] According to the elbow rule, the curve shows a clear inflection point when K=6, which is the point where the marginal benefit is maximized. Therefore, the initial number of categories... =6.

[0066] Figure 2 The elbow rule for fuzzy C-means clustering is shown. Figure 3The fuzzy C-mean profile coefficients are shown.

[0067] Step 3: Generate the candidate category set.

[0068] by Using the midpoint as the center, extend n units to both sides to form a set of candidate classification numbers.

[0069] First, determine the expansion amount n based on the data characteristics.

[0070] Determine the sample size and weighting coefficients : =0.18; Calculate the permeability heterogeneity coefficient : standardization: Calculate the porosity heterogeneity coefficient : standardization: Calculate the reservoir heterogeneity strength coefficient : Determine the cluster stability compensation term: With a profile coefficient of 0.22, we can obtain: ; Finally, after comprehensive calculation, we can obtain: Here, the value is rounded down.

[0071] Centered on 6, expand outwards by 2 values ​​on each side, resulting in the final set K of candidate categories. set = {K 0-n , ..., K0,..., K 0+n The condition} = {4, 5, 6, 7, 8} satisfies the constraints and boundary constraints.

[0072] Figure 4 The graph showing the relationship between porosity and permeability from core analysis (K) is shown. i = 4), Figure 5 The graph showing the relationship between porosity and permeability from core analysis (K) is shown. i =5); Figure 6 The graph showing the relationship between porosity and permeability from core analysis (K) is shown. i = 6); Figure 7The graph showing the relationship between porosity and permeability from core analysis (K) is shown. i = 7); Figure 8 The graph showing the relationship between porosity and permeability from core analysis (K) is shown. i =8).

[0073] Step 4: Construct a labeled dataset.

[0074] First, based on the Gaussian mixture model, the core samples were divided into reservoir types {4, 5, 6, 7, 8}. Based on this, five labeled datasets were established, in which the features are well logging data and the labels are reservoir types.

[0075] Step 5: Establish a multi-class penetration rate calculation model.

[0076] In the five labeled datasets, corresponding permeability calculation models were established based on the data labels, as shown in Table 1 below. Among them, Class I represents the worst reservoirs, characterized by poor porosity and poor connectivity; the higher the class, the higher the reservoir quality.

[0077] Table 1. Calculation model for reservoir permeability in oilfield A of a certain basin. Step 6: Predict reservoir types using XGBoost machine learning models. Train XGBoost reservoir type prediction models for each of the {4, 5, 6, 7, 8} reservoir types.

[0078] Figures 9-13 The permeability prediction accuracy for different reservoir types is shown.

[0079] Step 7: Determine the optimal number of categories.

[0080] After the reservoir type prediction is completed, the permeability is calculated using the corresponding permeability calculation model and compared with the actual permeability of the core. The average absolute logarithmic error of the calculated permeability is shown in Table 2.

[0081] Table 2. Calculation error of penetration rate for different categories Finally, the optimal number of classifications, 7, with the lowest error, was selected.

[0082] Step 8: Final model building and prediction.

[0083] A final dataset was established based on seven reservoir types, with the training and test sets divided in a 7:3 ratio. The XGBoost hyperparameters were optimized using a Bayesian optimization algorithm to determine the final reservoir type prediction model. Simultaneously, a classification permeability calculation model was embedded in the machine learning output, achieving integrated classification-prediction-computation. This model was applied to full-well logging data from new wells to predict reservoir types and output continuous permeability curves. Figure 14 The permeability calculation results for well A are shown.

[0084] Furthermore, an electronic device according to the present invention includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program to implement the permeability prediction method for the optimal reservoir classification number as described above. Specifically, according to embodiments of the present invention, the processes described above with reference to the flowchart can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, when the computer program is downloaded, installed, and executed by an electronic device, it performs the functions defined above in the methods of the embodiments of the present invention. The electronic device in the present invention can be a terminal such as a laptop, desktop computer, tablet computer, or smartphone, or it can be a server.

[0085] Furthermore, one type of storage medium of the present invention stores a computer program thereon, which, when executed by a processor, implements the permeability prediction method for the optimal reservoir classification number as described above. Specifically, it should be noted that the storage medium described above in the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein computer-readable program code is carried. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0086] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0087] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0088] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein 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 the various examples 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 implementations should not be considered beyond the scope of this invention.

[0089] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0090] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They do not limit the scope of protection of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for predicting permeability of optimal reservoir classification number, characterized in that, Includes the following steps: step S10: Collect all core conventional physical property analysis data and corresponding depth logging curve data of the target area, and perform data preprocessing to obtain the initial unlabeled dataset; Step S20: Perform cluster analysis on the initial unlabeled dataset using a clustering algorithm to determine the initial number of categories; Step S30: Using the initial number of categories as the midpoint, expand outwards to both sides according to the expansion amount n to form a set of candidate categories; Step S40: Construct a labeled dataset based on the set of candidate classification numbers; Step S50: Construct a multi-class penetration rate calculation model based on the labeled dataset; Step S60: Establish a corresponding machine learning reservoir type prediction model for the multi-class permeability calculation model, and perform reservoir type prediction; Step S70: After reservoir type prediction is completed, determine the optimal number of classifications; Step S80: Establish the final reservoir type prediction model based on the optimal number of classifications, and embed the number of classifications permeability calculation model into the final reservoir type prediction model for prediction.

2. The permeability prediction method for optimal reservoir classification number according to claim 1, characterized in that, Step S30 includes: Determine the expansion quantity n; Using the initial number of categories as the midpoint, expand n units to both sides to form the candidate number of categories set.

3. The permeability prediction method for the optimal reservoir classification number according to claim 2, characterized in that, The expansion amount n is determined through the following steps: Determine the sample size and sample size weighting coefficients; Calculate the heterogeneity weighting coefficient; Calculate the cluster stability weight coefficients; Calculate the reservoir heterogeneity strength coefficient; Determine the cluster stability compensation term; The expansion quantity n is obtained by combining the sample size and sample size weight coefficient, the heterogeneity weight coefficient, the cluster stability weight coefficient, the reservoir heterogeneity intensity coefficient, and the cluster stability compensation term.

4. The permeability prediction method for optimal reservoir classification number according to claim 1, characterized in that, Step S40 includes: Based on the Gaussian mixture model, the core samples were classified into multiple reservoir types; A labeled dataset is constructed based on the division of multiple reservoir types; in the labeled dataset, well logging data is the feature and reservoir type is the label.

5. The permeability prediction method for optimal reservoir classification number according to claim 1, characterized in that, Step S60 includes: For each of the multi-class permeability calculation models, a corresponding machine learning reservoir type prediction model is established. For different types of reservoirs, the corresponding machine learning reservoir type prediction model is used to predict the reservoir type.

6. The permeability prediction method for optimal reservoir classification number according to claim 1, characterized in that, Step S70 includes: After the reservoir type is predicted, the permeability is calculated using the corresponding permeability calculation model to obtain the corresponding predicted permeability. The predicted permeability is compared with the actual permeability of the core sample to calculate the corresponding mean absolute logarithmic error of the permeability. The optimal number of classifications is determined based on the mean absolute logarithmic error of the penetration rate.

7. The permeability prediction method for the optimal reservoir classification number according to any one of claims 1-6, characterized in that, Step S80 includes: The final dataset is established based on the reservoir type corresponding to the optimal number of classifications. Based on the final dataset, a machine learning algorithm is used to train and construct the final reservoir type prediction model; The classification number permeability calculation model is embedded into the final reservoir type prediction model for prediction.

8. A permeability prediction system for optimal reservoir classification number, characterized in that, include: The data collection and preprocessing unit is used to collect all conventional physical property analysis data of core samples and corresponding depth logging curve data of the target area, and to perform data preprocessing to obtain an initial unlabeled dataset; The clustering analysis calculation unit is used to perform clustering analysis on the initial unlabeled dataset using a clustering algorithm to determine the initial number of categories; The classification number set construction unit is used to expand n units to both sides of the initial classification number as the midpoint to form a candidate classification number set; A labeled dataset construction unit is used to construct a labeled dataset based on the set of candidate categories. A penetration rate model building unit is used to build a multi-class penetration rate calculation model based on the labeled dataset. The reservoir type prediction model construction and prediction unit is used to build a corresponding machine learning reservoir type prediction model based on the multi-classification penetration calculation model and to predict the reservoir type. The optimal number of classifications determination unit is used to determine the optimal number of classifications after reservoir type prediction is completed; The final model building and prediction unit is used to build the final reservoir type prediction model based on the optimal number of classifications, and to embed the classification number permeability calculation model into the final reservoir type prediction model for prediction.

9. A storage medium, characterized in that, The storage medium stores a computer program adapted for loading by a processor to perform the steps of the permeability prediction method for the optimal reservoir classification number as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the permeability prediction method for the optimal reservoir classification number as described in any one of claims 1 to 7 by calling the computer program stored in the memory.

Images