Oil and gas reservoir well location deployment method and device based on machine learning, equipment and medium

Abstract

The invention discloses an oil and gas reservoir well location deployment method, device and equipment based on machine learning and a storage medium. The method comprises the following steps: obtaining a yield control factor matched with a to-be-predicted well location; determining a well site yield prediction result of the to-be-predicted well site according to the yield control factor matched with the to-be-predicted well site and a pre-established well site yield prediction model, and performing well site deployment according to the well site yield prediction result; wherein the well location yield prediction model is a decision tree model obtained by regression training based on yield control factors of deployed well locations of oil and gas reservoir wells and well location yields matched with the yield control factors. According to the technical scheme, the problems of low oil and gas reservoir well location deployment accuracy, high dominant well location positioning difficulty and the like are solved, and the oil and gas exploitation efficiency can be improved while the oil and gas reservoir well location deployment precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of geophysical technology, and in particular to a method, device, electronic device and storage medium for oil and gas reservoir well location deployment based on machine learning. Background Art

[0002] At present, deepwater oilfields are mainly developed by sparse irregular well patterns. The investment per well in deepwater oilfields is high, and the conventional method of using regular well patterns to effectively deal with the heterogeneity of reservoirs in onshore oilfields is difficult to apply. The high-efficiency development model with few wells places higher demands on the well pattern design and well location deployment of deepwater oilfields. Therefore, a method for well location deployment of oil and gas reservoirs is urgently needed to realize the well location deployment of deepwater oilfields and support the efficient well layout and profitable development of deepwater oilfields. Summary of the invention

[0003] The present invention provides a method, device, equipment and storage medium for oil and gas reservoir well site deployment based on machine learning to solve the problems of low accuracy of oil and gas reservoir well site deployment and great difficulty in locating advantageous well sites. It can improve the accuracy of oil and gas reservoir well site deployment while improving the efficiency of oil and gas extraction.

[0004] According to one aspect of the present invention, a method for oil and gas reservoir well location deployment based on machine learning is provided, the method comprising:

[0005] Obtain production control factors matching the well positions to be predicted;

[0006] Determine the well production prediction result of the well to be predicted according to the production control factors matched with the well to be predicted and the well production prediction model established in advance, so as to deploy the well according to the well production prediction result;

[0007] The well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matching the production control factors.

[0008] According to another aspect of the present invention, there is provided an oil and gas reservoir well location deployment device based on machine learning, the device comprising:

[0009] A control factor acquisition module is used to obtain production control factors matching the well position to be predicted;

[0010] A prediction result determination module is used to determine the well production prediction result of the well to be predicted according to the production control factors matched with the well to be predicted and the well production prediction model established in advance, so as to deploy the well according to the well production prediction result;

[0011] The well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matching the production control factors.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the oil and gas reservoir well location deployment method based on machine learning as described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the oil and gas reservoir well location deployment method based on machine learning as described in any embodiment of the present invention when executed.

[0017] The technical solution of the embodiment of the present invention obtains the production control factors matched with the well location to be predicted, determines the well location production prediction results of the well location to be predicted according to the production control factors matched with the well location to be predicted and the well location production prediction model established in advance, and performs well location deployment according to the well location production prediction results. This technical solution solves the problems of low accuracy of well location deployment in oil and gas reservoirs and great difficulty in locating advantageous well locations, and can improve the efficiency of oil and gas production while improving the accuracy of well location deployment in oil and gas reservoirs.

[0018] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 is a flow chart of a method for deploying well locations in oil and gas reservoirs based on machine learning according to Embodiment 1 of the present invention;

[0021] Figure 2Ais a flow chart of a method for deploying well locations in oil and gas reservoirs based on machine learning according to Embodiment 2 of the present invention;

[0022] Figure 2B is a schematic diagram of characteristic correlation between yield control factors provided according to the second embodiment of the present invention;

[0023] Figure 2C is a schematic diagram of the relationship between the number of production control factors and the mean square error provided in the second embodiment of the present invention;

[0024] Figure 2D is a schematic diagram of the importance of production control factors to well production according to the second embodiment of the present invention;

[0025] Figure 3 It is a structural schematic diagram of an oil and gas reservoir well location deployment device based on machine learning provided according to Embodiment 3 of the present invention;

[0026] Figure 4 It is a structural schematic diagram of an electronic device for implementing the oil and gas reservoir well location deployment method based on machine learning according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices. The acquisition, storage, use, processing, etc. of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0029] Embodiment 1

[0030] Figure 1A flowchart of a method for well location deployment in oil and gas reservoirs based on machine learning is provided for the first embodiment of the present invention. This embodiment is applicable to well location deployment scenarios in oil and gas reservoirs. The method can be executed by a well location deployment device based on machine learning. The device can be implemented in the form of hardware and / or software, and the device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0031] S110, obtaining production control factors matching the well position to be predicted.

[0032] This solution can be executed by electronic devices such as computers and servers, and the electronic devices can obtain the production control factors of the wells to be predicted based on the logging data of the oil reservoir, production dynamic data and other information. The production control factors can be the well attributes that affect the well production, for example, factors such as porosity, permeability, net-to-gross ratio, oil layer thickness, formation inclination, reserve abundance, and well-controlled reserves.

[0033] S120, determining the well production prediction result of the well location to be predicted according to the production control factors matched with the well location to be predicted and the well production prediction model established in advance, so as to perform well location deployment according to the well production prediction result.

[0034] The electronic device can use the production control factors matched with the well location to be predicted as the input of the well location production prediction model to obtain the well location production prediction result of the well location to be predicted. Among them, the well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matched with the production control factors. For example, the decision tree model can be a decision tree structure constructed based on XGBoost (Extreme Gradient Boosting). XGBoost is an integrated learning method based on CART (Classification And Regression Tree), which improves the accuracy of the prediction model by constructing multiple CART decision trees, and sums the prediction results of the decision trees obtained from each round of training to obtain the final prediction value. Compared with GBDT, XGBoost adds a regular term in the objective function, which can effectively prevent overfitting. Among them, the objective function of the CART decision tree is defined as follows:

[0035]

[0036] Among them, t represents the decision tree index, i represents the sample index, and n represents the number of samples. represents the predicted value of sample i by the first t-1 decision trees, y i represents the actual value of sample i, f t (x i) represents the predicted value of the t-th decision tree for sample i, Ω(f t ) represents the model complexity of the t-th decision tree, w represents the vector composed of the predicted values ​​of each leaf node, T represents the number of leaf nodes, γ represents the first coefficient, and λ represents the second coefficient.

[0037] When performing feature selection, traverse the values ​​of all feature variables in the training set, subtract the sum of the objective function values ​​of the two leaf nodes after the split from the objective function value before the node split, calculate the gain value, and obtain the optimal split point of the tree model. The gain value calculation formula is:

[0038]

[0039] Among them, I L Represents the index set of the left child node after the split, I R represents the right child node index set after the split, I represents the child node index set, I=I L UI R , i represents the child node index, g i represents the first-order partial derivative of the sample corresponding to child node i, h i Represents the second-order partial derivative of the sample corresponding to child node i.

[0040] In this solution, the well location production prediction model can include multiple weak prediction structures, and the generation of each weak prediction structure can be based on the gradient direction of the preset loss function. Based on the multiple weak prediction structures, the prediction results output by each weak prediction structure are subjected to partial forward addition operations to obtain the well location production prediction results of the well location to be predicted.

[0041] The technical solution of the embodiment of the present invention obtains the production control factors matched with the well location to be predicted, determines the well location production prediction results of the well location to be predicted according to the production control factors matched with the well location to be predicted and the well location production prediction model established in advance, and performs well location deployment according to the well location production prediction results. This technical solution solves the problems of low accuracy of well location deployment in oil and gas reservoirs and great difficulty in locating advantageous well locations, and can improve the efficiency of oil and gas production while improving the accuracy of well location deployment in oil and gas reservoirs.

[0042] Embodiment 2

[0043] Figure 2A This is a flow chart of a method for deploying oil and gas reservoir wells based on machine learning provided in Embodiment 2 of the present invention. This embodiment is based on the above embodiment and is refined. Figure 2A As shown, the method includes:

[0044] S210, obtaining production control factors of deployed well locations of oil and gas reservoir wells, and generating a target feature set based on the production control factors.

[0045] The production control factors include at least one of porosity, permeability, net-to-gross ratio, oil layer thickness, formation inclination, reserve abundance and well-controlled reserves.

[0046] Optionally, after obtaining at least one production control factor of a deployed well position of an oil and gas reservoir well, and before generating a target feature set according to the production control factor, the method further includes:

[0047] Data cleaning is performed on each yield control factor; wherein the data cleaning includes at least one of duplicate value processing, missing value processing, outlier processing, standardization processing and normalization processing.

[0048] In this embodiment, after obtaining the production control factors of the deployed wells of the oil and gas reservoir, the electronic device can perform data cleaning on the production control factors to improve the reliability of the well production prediction model. Correspondingly, the electronic device can also perform data cleaning on the production control factors matched with the well to be predicted to ensure the accuracy of the input data of the well production prediction model, thereby ensuring the reliability of the well production prediction results.

[0049] It is understandable that data cleaning may include processing of values ​​such as duplicate values, missing values, and outliers. Specifically, for duplicate values, the electronic device may perform elimination processing. For missing values, the electronic device may perform supplementary processing according to the distribution law of non-missing values. The electronic device may also correct outliers. In order to avoid the dimensional influence of singular value data and different factors, the electronic device may standardize and normalize the production control factors to ensure the prediction stability of the well production prediction model.

[0050] In a feasible solution, the target feature set is generated according to the yield control factor data, including:

[0051] If the number of yield control factors is greater than 1, the feature correlation between every two yield control factors is determined, and based on the feature correlation between every two groups of yield control factors, a target feature set is determined; wherein the target feature set includes at least two target yield control factors determined in each yield control factor.

[0052] The more production control factors there are, the more input data the well production prediction model needs to have, and the more calculations the well production prediction model needs to complete, so the prediction time of the well production is longer, resulting in lower working efficiency of the well production prediction model. The electronic device can calculate the feature correlation of every two production control factors and screen the production control factors with similar features, thereby streamlining the number of production control factors and improving the prediction efficiency of the well production prediction model.

[0053] It is easy to understand that the electronic device can use the Pearson correlation coefficient calculation formula to calculate the feature correlation of every two yield control factors. If the absolute value of the feature correlation is greater than or equal to the preset correlation threshold, it means that the feature correlation of the two yield control factors is strong, and one of the two yield control factors can be deleted, and only one yield control factor can be retained. If the absolute value of the feature correlation is less than the preset correlation threshold, it means that the feature correlation of the two yield control factors is weak, and both yield control factors can be retained. After completing the feature correlation processing for every two yield control factors, the retained yield control factors can be used to form a target feature set.

[0054] S220: recursively eliminate the target yield control factors in the target feature set to determine a target feature subset.

[0055] The electronic device can perform an importance assessment on the target yield control factors in the target feature set, and screen key target yield control factors to form a target feature subset. Specifically, the electronic device can perform an importance assessment on the target yield control factors in the target feature set based on a random forest method to obtain a target feature subset. The target feature subset includes at least two target yield control factors.

[0056] In this solution, optionally, recursively eliminating the target yield control factors in the target feature set to determine the target feature subset includes:

[0057] Inputting each target yield control factor in the target feature set into the random forest regression simulator in turn to determine the first importance evaluation result of each target control factor;

[0058] According to the first importance evaluation result, at least one target control factor in the target feature set that does not meet the preset evaluation condition is removed to obtain the target feature set for this recursive matching, and the regression accuracy of the target feature set is determined;

[0059] If the target feature set of this recursive matching is not empty, then return to execute and input the yield control factors of each target group in the target feature set into the random forest regression simulator in turn;

[0060] If the target feature set of this recursive matching is empty, the target feature sets of all previous recursive matching are output, and the target feature subset is determined according to the regression accuracy of each target feature set.

[0061] It can be understood that the target yield control factors in the target feature set are input into the random forest regression simulator, the regression accuracy of each target yield factor is calculated, and the regression accuracy is used as the first importance evaluation result of the target yield factor. The electronic device can remove the target control factor with the lowest regression accuracy in the target feature set to obtain the target feature set for this recursive matching, and calculate the regression accuracy of the target feature set for this recursive matching.

[0062] If the target feature set of this recursive matching is not empty, then return to execute and input the yield control factors of each target group in the target feature set into the random forest regression simulator in turn, so as to continuously remove the target control factors with the lowest regression accuracy in the target feature set, and obtain the target feature set of all previous recursive matchings. If the target feature set of this recursive matching is empty, then output the target feature sets of all previous recursive matchings, and determine the target feature subset based on the regression accuracy of each target feature set. For example, the electronic device can use the target feature set with the highest regression accuracy as the target feature subset.

[0063] S230, performing at least one regression training on the decision tree model according to the target feature subset and the well production matched by each production control factor in the target feature subset to obtain a well production prediction model.

[0064] After obtaining the target feature subset, the electronic device can use the target production control factor in the target feature subset as the input of the decision tree model to obtain the well production prediction result output by the decision tree model. According to the prediction result of the well production and the well production matched by the target production control factor, the decision coefficient is calculated, and the decision tree model is regressed according to the decision coefficient. According to the evaluation results of the decision tree model obtained through previous training, the well production prediction model is determined.

[0065] In a feasible solution, the decision tree model is subjected to at least one regression training based on the target feature subset and the well production matched by each production control factor in the target feature subset to obtain a well production prediction model, including:

[0066] The target production control factors in the target feature subset are used as inputs of the decision tree model to obtain a prediction result of the well production, and a decision coefficient is calculated according to the prediction result and the well production matched with the pre-acquired target production control factors;

[0067] According to the decision coefficient, at least one regression training is performed to obtain a well production prediction model.

[0068] It should be noted that the electronic device can divide the data samples consisting of the target yield control factors in the target feature subset into a training set and a test set according to a preset ratio, and train the decision tree model using an N-fold cross-validation method based on the data samples in the training set. Training the decision tree model using the N-fold cross-validation method can ensure the model training effect while greatly reducing the calculation cost and time cost.

[0069] In a preferred solution, the decision tree model is subjected to at least one regression training based on the target feature subset and the well production matched by each production control factor in the target feature subset to obtain a well production prediction model, and the method further includes:

[0070] Generate a parameter population based on a genetic algorithm within a preset model parameter range; wherein the parameter population includes at least two model parameter combinations; each model parameter combination includes at least two model parameter items;

[0071] The model parameter combinations in the parameter population are sequentially used as the parameters of the decision tree model, and the decision tree model is subjected to at least one regression training according to the target feature subset and the well location production matched by each production control factor in the target feature subset, so as to determine the candidate well location production prediction model matched by each model parameter combination and the error of each candidate well location production prediction model;

[0072] According to the candidate well location production prediction models matched by each model parameter combination and the errors of each candidate well location production prediction model, the parameter population is genetically iterated at least once to determine the target model parameter combination, and the candidate well location production prediction model matched by the target model parameter combination is used as the well location production prediction model.

[0073] It is easy to understand that the electronic device can determine the optimal parameters of the well production prediction model based on the genetic algorithm to maximize the prediction accuracy of the well production prediction model. Specifically, the electronic device can generate a parameter population according to a preset model parameter range. Among them, the parameter population can include at least two model parameter combinations, each model parameter combination in the parameter population can be used as a parameter individual, and each model parameter combination can include at least two model parameter items, and the model parameter items can include model parameters such as the number of sub-models, learning rate, maximum depth, minimum sub-tree weight, column sampling rate, sub-sampling rate and hyperparameters in the decision tree model.

[0074] The electronic device can sequentially use the model parameter combinations in the parameter population as the parameters of the decision tree model, perform at least one regression training on the decision tree model according to the target feature subset and the well location production matched by each production control factor in the target feature subset, and determine the candidate well location production prediction model matched by each model parameter combination and the error of each candidate well location production prediction model. According to the candidate well location production prediction model matched by each model parameter combination and the error of each candidate well location production prediction model, the parameter population is genetically iterated at least once. According to each candidate well location production prediction model obtained from the last iteration, the electronic device can use the model parameter combination matched by the candidate well location production prediction model with the smallest error as the target model parameter combination, and use the candidate well location production prediction model matched by the target model parameter combination as the well location production prediction model.

[0075] Optionally, the error of the candidate well location production prediction model can be expressed by a root mean square error, and the calculation formula of the root mean square error is expressed as:

[0076] Among them, f(x i ) represents the predicted production value of the well location of sample i, Y i represents the actual value of the well production of sample i, and n represents the number of samples.

[0077] This scheme can determine the optimal model parameters for the well location production prediction model, which is beneficial to improving the prediction accuracy of the well location production prediction model.

[0078] S240, obtaining production control factors matching the well position to be predicted.

[0079] S250, determining the well production prediction result of the well location to be predicted according to the production control factors matched with the well location to be predicted and the well production prediction model established in advance, so as to perform well location deployment according to the well production prediction result.

[0080] The technical solution of the embodiment of the present invention obtains the production control factors matched with the well location to be predicted, determines the well location production prediction results of the well location to be predicted according to the production control factors matched with the well location to be predicted and the well location production prediction model established in advance, and performs well location deployment according to the well location production prediction results. This technical solution solves the problems of low accuracy of well location deployment in oil and gas reservoirs and great difficulty in locating advantageous well locations, and can improve the efficiency of oil and gas production while improving the accuracy of well location deployment in oil and gas reservoirs.

[0081] In a specific example, the oil and gas reservoir well location deployment process can be implemented in the following steps:

[0082] Step 1: Identify yield control factors.

[0083] Production control factors include porosity, permeability, net-to-gross ratio, reservoir thickness, formation inclination, reserve abundance, and well-controlled reserves, a total of 7 factors. Obtain data on production control factors for oil and gas reservoir well location matching.

[0084] Step 2: Data preprocessing.

[0085] Data preprocessing is performed on the yield control factors, and the missing values ​​are filled with the median of each yield control factor. The box method is used to perform outlier analysis based on the filled data. In order to avoid using different orders of magnitude between features, data standardization is required after outlier analysis. The standardization calculation formula is: Among them, x * represents the data after standardization, μ represents the mean of the data, x represents the data before standardization, and σ represents the standard deviation of the data. After standardization, the data conforms to the standard normal distribution, that is, the mean is 0 and the standard deviation is 1.

[0086] Figure 2B Schematic diagram of characteristic correlation between yield control factors provided according to the second embodiment of the present invention. Characteristic screening is performed on the standardized data, and the correlation between every two yield control factors is calculated using the Pearson correlation coefficient calculation formula, such as Figure 2B The figure shows the correlation between the characteristics of each yield control factor. A correlation coefficient value in the range of 0.8-1.0 indicates a strong correlation between the characteristics. A positive correlation coefficient indicates a positive correlation between the characteristics, and a negative correlation indicates a negative correlation. Yield control factors are screened based on the correlation coefficient. The Pearson correlation coefficient calculation formula is expressed as:

[0087]

[0088] Step 3: Use the recursive feature elimination method based on random forest to optimize the number of features and iteratively remove the production control factors with the smallest impact on the well production until the optimal number of production control factors is identified.

[0089] Each well location control factor is taken as a feature, and the selected k features are input as the initial feature subset into the random forest regression simulator. The importance of each feature is calculated, and the regression accuracy of the initial feature subset is obtained by cross-validation. The feature with the lowest feature importance is removed from the current feature subset to obtain a new feature subset, which is input into the random forest regression simulator again. The importance of each feature in the new feature subset is calculated, and the regression accuracy of the new feature subset is obtained by cross-validation. The feature removal process is repeated recursively until the feature subset is empty. A total of k feature subsets with different numbers of features are obtained, and the feature subset with the highest regression accuracy is selected as the optimal feature combination. Figure 2Cis a schematic diagram of the relationship between the number of production control factors and the mean square error according to the second embodiment of the present invention; in this embodiment, Figure 2C As shown, when 4 out of 7 features are retained, the model error is the smallest.

[0090] Step 4: Perform feature importance analysis based on the XGBoost algorithm.

[0091] Figure 2D Schematic diagram of the importance of production control factors to well production according to the second embodiment of the present invention. Figure 2D As shown in the figure, the feature importance analysis method based on the XGBoost algorithm obtained the importance of the input features after XGBoost training, and further explored the key factors affecting the well location decision.

[0092] Step 5: Regression prediction of cumulative oil production at different well locations based on XGBoost algorithm.

[0093] The cumulative oil production of different well locations is predicted by regression. The data set is randomly divided into training set and test set according to a certain ratio. During the model training process, the model parameters need to be adjusted. On the training set, the algorithm is optimized by 5-fold cross validation to avoid overfitting and obtain the optimal model. The genetic algorithm is used to train and evaluate the model under different parameter combinations, and the parameter combination with the smallest average cross validation error of the training set is selected. The trained XGBoost model is used to verify the new oil and gas reservoir production control factor data, that is, the test set. The RMSE of the training set is 0.001947, and the RMSE of the test set is 0.0722. The test results show that the model has good prediction ability, indicating that the model can better predict the cumulative oil production of different well locations.

[0094] This embodiment uses the XGBoost algorithm to optimize the well layout plan by preprocessing and analyzing various data in the oil and gas production process, effectively reducing the production cost and improving the production efficiency. The implementation of the method of the present invention can improve the efficiency and accuracy of oil and gas production, bringing new technological progress and development to the field of oil and gas production.

[0095] Embodiment 3

[0096] Figure 3 This is a schematic diagram of the structure of a well location deployment device for oil and gas reservoirs based on machine learning provided in Embodiment 3 of the present invention. Figure 3 As shown, the device comprises:

[0097] The control factor acquisition module 310 is used to acquire the production control factor matched with the well position to be predicted;

[0098] The prediction result determination module 320 is used to determine the well production prediction result of the well to be predicted according to the production control factors matched with the well to be predicted and the well production prediction model established in advance, so as to perform well deployment according to the well production prediction result;

[0099] The well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matching the production control factors.

[0100] In this solution, optionally, the device further includes a regression training module, and the regression training module includes:

[0101] A target feature set determination unit is used to obtain production control factors of deployed wells in oil and gas reservoirs, and generate a target feature set according to the production control factors; wherein the production control factors include at least one of porosity, permeability, net-to-gross ratio, oil layer thickness, formation dip, reserve abundance and well-controlled reserves;

[0102] A target feature subset determination unit, used to perform recursive feature elimination on the target yield control factors in the target feature set to determine a target feature subset; wherein the target feature subset includes at least two target yield control factors;

[0103] The prediction model determination unit is used to perform at least one regression training on the decision tree model according to the target feature subset and the well production matched by each production control factor in the target feature subset to obtain a well production prediction model.

[0104] Based on the above solution, optionally, the target feature set determination unit is specifically used to:

[0105] If the number of yield control factors is greater than 1, the feature correlation between every two yield control factors is determined, and based on the feature correlation between every two groups of yield control factors, a target feature set is determined; wherein the target feature set includes at least two target yield control factors determined in each yield control factor.

[0106] In this embodiment, optionally, the target feature subset determination unit is specifically configured to:

[0107] Inputting each target yield control factor in the target feature set into the random forest regression simulator in turn to determine the first importance evaluation result of each target control factor;

[0108] According to the first importance evaluation result, at least one target control factor in the target feature set that does not meet the preset evaluation condition is removed to obtain the target feature set for this recursive matching, and the regression accuracy of the target feature set is determined;

[0109] If the target feature set of this recursive matching is not empty, then return to execute and input the yield control factors of each target group in the target feature set into the random forest regression simulator in turn;

[0110] If the target feature set of this recursive matching is empty, the target feature sets of all previous recursive matching are output, and the target feature subset is determined according to the regression accuracy of each target feature set.

[0111] In a feasible solution, the prediction model determination unit is specifically used to:

[0112] The target production control factors in the target feature subset are used as inputs of the decision tree model to obtain a prediction result of the well production, and a decision coefficient is calculated according to the prediction result and the well production matched with the pre-acquired target production control factors;

[0113] According to the decision coefficient, at least one regression training is performed to obtain a well production prediction model.

[0114] In another feasible solution, the prediction model determination unit is specifically used to:

[0115] Generate a parameter population based on a genetic algorithm within a preset model parameter range; wherein the parameter population includes at least two model parameter combinations; each model parameter combination includes at least two model parameter items;

[0116] The model parameter combinations in the parameter population are sequentially used as the parameters of the decision tree model, and the decision tree model is subjected to at least one regression training according to the target feature subset and the well location production matched by each production control factor in the target feature subset, so as to determine the candidate well location production prediction model matched by each model parameter combination and the error of each candidate well location production prediction model;

[0117] According to the candidate well location production prediction models matched by each model parameter combination and the errors of each candidate well location production prediction model, the parameter population is genetically iterated at least once to determine the target model parameter combination, and the candidate well location production prediction model matched by the target model parameter combination is used as the well location production prediction model.

[0118] In a preferred embodiment, the device further comprises:

[0119] A data cleaning module is used to perform data cleaning on each production control factor after obtaining at least one production control factor of a deployed well position in an oil and gas reservoir and before generating a target feature set based on the production control factor; wherein the data cleaning includes at least one of duplicate value processing, missing value processing, outlier processing, standardization processing and normalization processing.

[0120] The oil and gas reservoir well location deployment device based on machine learning provided in an embodiment of the present invention can execute the oil and gas reservoir well location deployment method based on machine learning provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0121] Embodiment 4

[0122] Figure 4 A schematic diagram of an electronic device 410 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0123] like Figure 4 As shown, the electronic device 410 includes at least one processor 411, and a memory connected to the at least one processor 411 in communication, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 411 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 to the random access memory (RAM) 413. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0124] Multiple components in the electronic device 410 are connected to the I / O interface 415, including: an input unit 416, such as a keyboard, a mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a disk, an optical disk, etc.; and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0125] The processor 411 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 411 executes the various methods and processes described above, such as a method for deploying wells in oil and gas reservoirs based on machine learning.

[0126] In some embodiments, the method for well location deployment of oil and gas reservoirs based on machine learning may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the method for well location deployment of oil and gas reservoirs based on machine learning described above may be performed. Alternatively, in other embodiments, the processor 411 may be configured to execute the method for well location deployment of oil and gas reservoirs based on machine learning in any other appropriate manner (e.g., by means of firmware).

[0127] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0128] The computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable machine learning-based oil and gas reservoir well location deployment device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0129] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0130] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0131] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0132] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0133] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0134] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for well location deployment in oil and gas reservoirs based on machine learning, characterized in that: The method comprises: Obtain production control factors matching the well positions to be predicted; Determine the well production prediction result of the well to be predicted according to the production control factors matched with the well to be predicted and the well production prediction model established in advance, so as to deploy the well according to the well production prediction result; The well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matching the production control factors.

2. The method according to claim 1, characterized in that: The regression training process of the well location production prediction model includes: Obtaining production control factors of deployed wells in oil and gas reservoirs, and generating a target feature set based on the production control factors; wherein the production control factors include at least one of porosity, permeability, net-to-gross ratio, oil layer thickness, formation dip, reserve abundance, and well-controlled reserves; Recursively eliminating the target yield control factors in the target feature set to determine a target feature subset; wherein the target feature subset includes at least two target yield control factors; According to the target feature subset and the well production matching each production control factor in the target feature subset, the decision tree model is trained for at least one regression to obtain a well production prediction model.

3. The method according to claim 2, characterized in that The target feature set is generated according to the yield control factor data, including: If the number of yield control factors is greater than 1, the feature correlation between every two yield control factors is determined, and based on the feature correlation between every two groups of yield control factors, a target feature set is determined; wherein the target feature set includes at least two target yield control factors determined in each yield control factor.

4. The method according to claim 2, characterized in that: The recursive feature elimination of the target yield control factors in the target feature set to determine the target feature subset includes: Inputting each target yield control factor in the target feature set into the random forest regression simulator in turn to determine the first importance evaluation result of each target control factor; According to the first importance evaluation result, at least one target control factor in the target feature set that does not meet the preset evaluation condition is removed to obtain the target feature set for this recursive matching, and the regression accuracy of the target feature set is determined; If the target feature set of this recursive matching is not empty, then return to execute and input the yield control factors of each target group in the target feature set into the random forest regression simulator in turn; If the target feature set of this recursive matching is empty, the target feature sets of all previous recursive matching are output, and the target feature subset is determined according to the regression accuracy of each target feature set.

5. The method according to claim 2, characterized in that: According to the target feature subset and the well production matched by each production control factor in the target feature subset, the decision tree model is subjected to at least one regression training to obtain a well production prediction model, including: The target production control factors in the target feature subset are used as inputs of the decision tree model to obtain a prediction result of the well production, and a decision coefficient is calculated according to the prediction result and the well production matched with the pre-acquired target production control factors; According to the decision coefficient, at least one regression training is performed to obtain a well production prediction model.

6. The method according to claim 2, characterized in that According to the target feature subset and the well production matched by each production control factor in the target feature subset, the decision tree model is subjected to at least one regression training to obtain a well production prediction model, and the method further comprises: Generate a parameter population based on a genetic algorithm within a preset model parameter range; wherein the parameter population includes at least two model parameter combinations; each model parameter combination includes at least two model parameter items; The model parameter combinations in the parameter population are sequentially used as the parameters of the decision tree model, and the decision tree model is subjected to at least one regression training according to the target feature subset and the well location production matched by each production control factor in the target feature subset, so as to determine the candidate well location production prediction model matched by each model parameter combination and the error of each candidate well location production prediction model; According to the candidate well location production prediction models matched by each model parameter combination and the errors of each candidate well location production prediction model, the parameter population is genetically iterated at least once to determine the target model parameter combination, and the candidate well location production prediction model matched by the target model parameter combination is used as the well location production prediction model.

7. The method according to claim 2, characterized in that After obtaining at least one production control factor of a deployed well position of an oil and gas reservoir well, and before generating a target feature set according to the production control factor, the method further includes: Data cleaning is performed on each yield control factor; wherein the data cleaning includes at least one of duplicate value processing, missing value processing, outlier processing, standardization processing and normalization processing.

8. A device for deploying oil and gas reservoir wells based on machine learning, characterized in that: include: A control factor acquisition module is used to obtain production control factors matching the well position to be predicted; A prediction result determination module is used to determine the well production prediction result of the well to be predicted according to the production control factors matched with the well to be predicted and the well production prediction model established in advance, so as to deploy the well according to the well production prediction result; The well location production prediction model is a decision tree model obtained by regression training based on the production control factors of the deployed well locations of the oil and gas reservoir wells and the well location production matching the production control factors.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the oil and gas reservoir well location deployment method based on machine learning as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the oil and gas reservoir well location deployment method based on machine learning as described in any one of claims 1-7 when executed.

Images