Machine learning modeling method and device for predicting shale gas productivity, equipment and medium
By pre-training offline data sets and Bayesian optimization methods, combining integrated selection methods, data processing and hyperparameter search are automatically processed, and small and medium sample sizes and model complexity problems are solved, achieving efficient and accurate shale Gas capacity prediction model is constructed, suitable for business personnel who lack artificial intelligence background.
Patent Information
- Application Number
- CN202310871298.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-07-17
- Publication Date
- 2025-01-17
AI Technical Summary
The existing technology has a small sample size in the prediction of shale gas production capacity, resulting in overfitting, complex model establishment, time-consuming and labor-intensive, and it is difficult to quickly implement and apply it. Especially for business personnel who lack artificial intelligence background, it is difficult to achieve high-quality. Machine learning model construction.
By pre-training offline data sets, an offline data warehouse is established, and a Bayesian optimization method based on random forest agent functions is used to search for hyperparameters. Combining the integrated selection method, multiple machine learning models are combined, data preprocessing and feature engineering are automated, and humans are reduced. Intervention to improve modeling efficiency.
It has achieved rapid, automatic and accurate establishment of high-quality shale gas production capacity prediction models with very little human intervention, which has improved modeling efficiency and prediction accuracy, reduced learning costs, and improved the applicability and accuracy of the model.
Smart Images
Figure CN119323274A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of upstream shale gas development, and specifically to a machine learning modeling method, device, equipment and medium for predicting shale gas production capacity. Background Art
[0002] Shale gas is an important strategic resource with abundant reserves worldwide. The development and production of shale gas reservoirs are affected by complex factors such as natural fractures, artificial fractures, connectivity, and reservoir stress changes. The flow mechanism is very complex and production capacity prediction is difficult. Traditional production capacity prediction methods have limited accuracy and are gradually unable to meet today's production needs. With the widespread application of machine learning algorithms in various industries, the use of machine learning modeling to accurately predict shale gas production capacity has been favored by the majority of energy industry practitioners in recent years. However, the construction of machine learning models involves multiple complex links such as feature engineering, model optimization, and hyperparameter tuning. Business personnel without a professional background in artificial intelligence often find it difficult to quickly and easily establish high-quality prediction models.
[0003] When establishing a shale gas production capacity prediction model, the following problems are usually encountered: (1) In the early stages of development, machine learning models are easily affected by small sample sizes and overfit, which greatly reduces accuracy and applicability. In order to reduce overfitting, the quality of the data set and the number of features must be strictly controlled, so a large amount of data preprocessing, feature engineering, and other work must be carried out. (2) The optimization of model architecture and hyperparameters requires a large number of repeated experiments, and the process of establishing a high-quality model consumes a lot of time and computing power. (3) The inherent mechanism of shale gas production capacity changes is very complex, and the prediction ability and accuracy of a single model are relatively limited. It is often necessary to examine the different ways in which multiple machine learning models handle bias and variance, combining their respective advantages to provide more accurate prediction results. At the same time, the complexity of the model and the difficulty of modeling also increase dramatically.
[0004] In summary, in order to establish a high-quality shale gas capacity prediction machine learning model, it is necessary to examine many complex algorithms and hyperparameter combinations, and the modeling process requires a lot of manpower and computing power to carry out repetitive experiments. For most shale gas production sites and business personnel, in the absence of artificial intelligence knowledge background, machine learning capacity prediction models are often difficult to quickly implement and promote due to their complex modeling process and high learning costs. Summary of the invention
[0005] In view of this, in order to improve the modeling efficiency of the shale gas production capacity prediction model, this application proposes a machine learning modeling method, device, equipment and medium for predicting shale gas production capacity, which intelligently and rationally allocates computing time and computing power to automatically, efficiently and accurately establish a high-quality shale gas production capacity prediction model with minimal human intervention.
[0006] In a first aspect, an embodiment of the present application provides a machine learning modeling method for predicting shale gas production capacity, comprising:
[0007] Collect and process target data and organize them into target data sets;
[0008] Pre-train a large number of offline data sets and obtain the optimal machine learning model to establish an offline data warehouse;
[0009] According to the target data set, an initialization model is selected from the offline data warehouse, and an optimal model is established by fusion to perform capacity prediction;
[0010] The main controlling factors of the production capacity of the optimal model are analyzed.
[0011] In a possible implementation, the collecting and processing of target data to form a target data set includes: collecting parameters related to shale gas exploration, development and production as target data; and processing the target data to obtain a target data set.
[0012] In a possible implementation, the parameters include: geological parameters, engineering parameters and production capacity characterization data.
[0013] In a possible implementation, the pre-training of a large number of offline data sets and obtaining an optimal machine learning model and establishing an offline data warehouse includes:
[0014] Performing meta-feature extraction and machine learning model training on each offline data set in a large number of offline data sets to obtain an optimal machine learning model for each offline data set;
[0015] The meta-features of each offline data set and its corresponding optimal machine learning model are stored in the data warehouse to establish an offline data warehouse.
[0016] In a possible implementation, selecting an initialization model from the offline data warehouse according to the target data set, fusing and establishing an optimal model, and performing capacity forecasting includes:
[0017] Extracting meta-features of the target data set, respectively calculating the L1 distances between the meta-features of the target data set and the meta-features of each offline data set, and selecting optimal machine learning models corresponding to several groups of offline data sets with the smallest L1 distances as initialization models;
[0018] The Bayesian optimization method based on the random forest proxy function is used to search for the hyperparameters of each initialization model and obtain the optimal hyperparameters of each initialization model;
[0019] The integrated selection method is used to fuse the initialization models, establish an integrated machine learning model, perform capacity forecasting, and output the final capacity forecasting results.
[0020] In a possible implementation, the calculation formula of the L1 distance is:
[0021]
[0022] Where N represents the number of meta-features of the target dataset and the offline dataset; x m represents the mth meta-feature of the offline dataset; f m Represents the mth meta-feature of the target dataset.
[0023] In a possible implementation, analyzing the main control factors of the production capacity of the optimal model includes:
[0024] Fine-tune the values of various geological engineering parameters to obtain a fine-tuning target data set;
[0025] Calculate the model errors ε and ε corresponding to the original target dataset and the fine-tuned target dataset respectively p ;
[0026] According to (ε p -ε) to analyze the impact of each parameter on production capacity and rank them by importance.
[0027] In a second aspect, an embodiment of the present application provides a machine learning modeling device for predicting shale gas production capacity, characterized in that it includes:
[0028] A target data set generation module is used to collect and process target data and organize them into a target data set;
[0029] Offline data warehouse establishment module, used to pre-train a large number of offline data sets and obtain the optimal machine learning model to establish an offline data warehouse;
[0030] A capacity prediction model is used to select an initialization model from the offline data warehouse according to the target data set, fuse and establish an optimal model, and perform capacity prediction;
[0031] The model interpretation module is used to analyze the main control factors of the production capacity of the optimal model.
[0032] In a third aspect, an embodiment of the present application provides an electronic device, including:
[0033] processor;
[0034] Memory;
[0035] And a computer program, wherein the computer program is stored in the memory, and the computer program includes instructions, and when the instructions are executed by the processor, the electronic device executes any one of the methods described in the first aspect.
[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described in the first aspect.
[0037] This application is based on relevant geological, engineering and production data in shale gas exploration and development, and realizes automated data preprocessing, feature preprocessing, model selection and hyperparameter search, assisting shale gas production sites and business personnel to achieve "one-stop" machine learning modeling, and improving the feasibility of the application and promotion of machine learning methods in the field of shale gas production capacity prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0039] Figure 1 A schematic diagram of a flow chart of a machine learning modeling method for predicting shale gas production capacity provided in an embodiment of the present application;
[0040] Figure 2 A schematic diagram showing the variation of the accuracy of a single optimal model and an integrated model with the operation time during the application of a machine learning modeling method for predicting shale gas production capacity provided in an embodiment of the present application;
[0041] Figure 3 A schematic diagram showing the comparison of the final capacity prediction value and the true value of the model generated by a machine learning modeling method for predicting shale gas capacity provided in an embodiment of the present application on a training set and a test set respectively;
[0042] Figure 4 A schematic diagram of the ranking of the influence of geological engineering parameters based on the analysis of the machine learning model during the application of a machine learning modeling method for predicting shale gas production capacity provided in an embodiment of the present application;
[0043] Figure 5 A structural block diagram of a machine learning modeling device for predicting shale gas production capacity provided in an embodiment of the present application;
[0044] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0045] In order to better understand the technical solution of the present application, the embodiments of the present application are described in detail below with reference to the accompanying drawings.
[0046] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.
[0047] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.
[0048] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.
[0049] See also Figure 1 , is a flow chart of a machine learning modeling method for predicting shale gas production capacity provided in an embodiment of the present application. Figure 1 As shown, it mainly includes the following steps.
[0050] Step S1: Collect and process target data and organize them into a target data set. Specifically, it includes:
[0051] Step S1.1: Collect target data. Collect parameters related to shale gas exploration, development and production, including but not limited to geological parameters such as gas layer depth, formation pressure, porosity, gas content, engineering parameters such as gas test section length, number of sections, number of clusters, sand addition intensity, liquid intensity, and capacity characterization data such as open flow, technically recoverable reserves, and cumulative production.
[0052] Step S1.2: Process the collected target data. The automatic modeling process covers some data preprocessing and feature processing functions, including one-hot encoding, missing value interpolation, feature aggregation, independent component analysis (ICA), principal component analysis (PCA), etc. However, the original data still needs to be preprocessed or manually adjusted in combination with expert experience, especially when the amount of data is small, the data quality has a great impact on model building. Therefore, it is necessary to combine machine learning and professional knowledge for data quality control and feature processing, including outlier processing, missing value processing, feature selection, etc.
[0053] Step S2: Pre-train a large number of offline data sets and obtain the optimal machine learning model to establish an offline data warehouse. Including:
[0054] Step S2.1: Perform meta-feature extraction and machine learning model training on each offline data set in a large number of offline data sets to obtain the optimal machine learning model for each offline data set.
[0055] In this embodiment, meta-feature extraction and machine learning model training are performed on 140 sets of offline data sets in the OpenML data warehouse. 38 meta-features including data quantity, feature quantity, data skewness, etc. are extracted from each set of data, and the sequential model algorithm configuration (SMAC) method is used to search for model hyperparameters.
[0056] The SMAC method is based on the Bayesian optimization algorithm, uses the random forest algorithm to construct a proxy function, uses the existing observation data (model, hyperparameter combination and fit) to fit the random forest model, and obtains the error prediction value and prediction value covariance of the model corresponding to the unknown hyperparameters based on the model. In the next iteration, the observation point with the smallest expected error value and the largest covariance is selected.
[0057] Step S2.2: Establish an offline data warehouse. Store the meta-features of each offline data set and its corresponding optimal machine learning model in the data warehouse, and establish an offline data warehouse for subsequent use.
[0058] Step S3: According to the target data set, select an initialization model from the offline data warehouse, fuse and establish an optimal model, and perform capacity forecasting. Specifically, it includes:
[0059] Step S3.1: Select an initialization model. Extract the meta-features of the target dataset, calculate the L1 distances between the meta-features of the target dataset and the meta-features of each offline dataset based on the k-nearest neighbor method, and select the optimal machine learning model corresponding to several groups of offline datasets with the smallest L1 distance as the initialization model. The L1 distance is calculated as follows:
[0060]
[0061] Where N represents the number of meta-features of the target dataset and the offline dataset; x m represents the mth meta-feature of the offline dataset; f m Represents the mth meta-feature of the target dataset.
[0062] In this embodiment, the L1 distance is calculated as follows:
[0063]
[0064] Step S3.2: Search the hyperparameters of each initialization model to obtain the optimal hyperparameters of each initialization model. Use the Bayesian optimization method based on the random forest surrogate function to search the hyperparameters of each initialization model.
[0065] Bayesian optimization has great advantages in solving black-box optimization problems with expensive computational costs. It mainly uses Bayesian theorem to guide the search for hyperparameters to minimize the error of machine learning models. In each iteration, the historical information (prior knowledge) observed before is used to perform the next calculation and optimization. The following work is performed for each initialization model:
[0066] i. First, select a small number of hyperparameter combinations and calculate their corresponding model errors, and initialize the proxy function that maps the hyperparameter θ to the model error ε;
[0067] ii. Calculate the acquisition function α(θ), select the hyperparameters that maximize the acquisition function α(θ), and balance the mining nature (searching near the optimal area) and exploration nature (searching in the area with the greatest uncertainty) of the hyperparameter search;
[0068] iii. Calculate the model test error ε corresponding to the hyperparameter θ;
[0069] iv. Use the new data (θ, ε) to update the surrogate function as the prior distribution for the next step;
[0070] v. Repeat steps ii to iv until the maximum number of iterations is reached and the optimal hyperparameters are output.
[0071] Step S3.3: Fuse each initialization model to establish an integrated machine learning model for capacity forecasting. Save the trained algorithm, use the ensemble selection method, repeatedly search and gradually add the initialization model that improves the model accuracy the most to the integrated model, and give each initialization model different weights according to its prediction accuracy, establish an integrated machine learning model, and output the final capacity forecast result.
[0072] Step S4: Use the sensitivity analysis method to analyze the main control factors of the production capacity of the optimal model. Specifically including:
[0073] Step S4.1: fine-tune the values of various geo-engineering parameters to obtain a fine-tuning target data set;
[0074] Step S4.2: Calculate the model errors ε and ε corresponding to the original target dataset and the fine-tuned target dataset respectively p ;
[0075] Step S4.3: According to (ε p -ε) to analyze the impact of each parameter on production capacity and rank them by importance.
[0076] In this embodiment, taking a shale gas block as an example, based on the exploration, development and production data of the block, a daily gas production prediction is carried out during the early stage of single well production through casing production.
[0077] According to a machine learning modeling method for predicting shale gas production capacity disclosed in the embodiment of this application, first collect the data of each shale gas horizontal well in the block, including geological parameters such as gas layer depth and TOC, engineering parameters such as gas test length, liquid intensity, sand addition intensity, and production data such as open flow, technically recoverable reserves, and cumulative production. After eliminating missing values and outliers, 70 single wells and 24 geological engineering parameters were finally sorted out to form a data set, some of which are shown in Table 1.
[0078] Table 1: Horizontal well geological, engineering and production data sets collected and collated for a shale gas block (partial display)
[0079]
[0080] Using a machine learning modeling method for predicting shale gas production capacity disclosed in an embodiment of the present application, 245 sets of models and hyperparameter combinations were quickly searched within 30 minutes, and the final prediction model integrated 4 machine learning algorithms. Figure 2 The accuracy of a single optimal model and an integrated model changes with the computation time. The training accuracy of a single optimal model first increases and then decreases, and the overfitting of the model is avoided based on cross-validation. Figure 2 It can be seen that the accuracy of the integrated model on the test set is much higher than that of a single model on the test set, and the fusion model has a greater advantage. Figure 3 The comparison between the final capacity prediction value and the actual value on the training set and the test set is shown. The R 2 The fit is 0.71, and the R 2 The goodness of fit is 0.51. Figure 4 The influence ranking of geological engineering parameters based on machine learning model analysis is demonstrated. Sensitivity analysis is mainly used to randomly fine-tune the values of each parameter and calculate the impact of the adjustment on the model accuracy, thereby guiding the analysis of the main controlling factors of production capacity.
[0081] In summary, this application can be used to efficiently and automatically establish a shale gas production capacity prediction machine learning model based on shale gas exploration, development and production data by combining offline learning with online learning. This application can be applied to the shale gas development and production process, efficiently predict and analyze shale gas production capacity, and quickly adjust the iterative model based on the updated data on site, greatly improving the efficiency of prediction and analysis, and effectively guiding the shale gas development and production practice.
[0082] This application is based on advanced data-driven and automatic machine learning modeling methods. Compared with traditional shale gas production capacity prediction modeling and general machine learning methods, this application does not require too much human intervention in data and feature preprocessing, model selection and hyperparameter search processes, and can automatically integrate multiple machine learning algorithms, while improving modeling efficiency and prediction accuracy, and has higher feasibility of implementation and promotion and application.
[0083] Corresponding to the above embodiments, the present application also provides a machine learning modeling device for predicting shale gas production capacity.
[0084] See also Figure 5 , is a structural block diagram of a machine learning modeling device for predicting shale gas production capacity provided in an embodiment of the present application. Figure 5 As shown, it mainly includes the following modules.
[0085] The target data set generation module 501 is used to collect and process the target data and organize it into a target data set;
[0086] An offline data warehouse establishment module 502 is used to pre-train a large number of offline data sets and obtain an optimal machine learning model to establish an offline data warehouse;
[0087] The capacity prediction model 503 is used to select an initialization model from the offline data warehouse according to the target data set, fuse and establish an optimal model, and perform capacity prediction;
[0088] The model interpretation module 504 is used to analyze the main control factors of the production capacity of the optimal model.
[0089] It should be pointed out that the specific contents involved in the embodiments of the present application can be found in the description of the above method embodiments, and for the sake of brevity, they will not be repeated here.
[0090] Corresponding to the above embodiment, an embodiment of the present application further provides an electronic device.
[0091] See also Figure 6 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 As shown, the electronic device 600 may include: a processor 601, a memory 602 and a communication unit 603. These components communicate via one or more buses. Those skilled in the art will appreciate that the electronic device structure shown in the figure does not limit the embodiments of the present application, and it may be a bus structure or a star structure, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently.
[0092] The communication unit 603 is used to establish a communication channel so that the electronic device can communicate with other devices.
[0093] The processor 601 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It runs or executes software programs and / or modules stored in the memory 602, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of multiple packaged ICs with the same or different functions. For example, the processor 601 can include only a central processing unit (CPU). In the embodiment of the present application, the CPU can be a single computing core or multiple computing cores.
[0094] Memory 602 is used to store execution instructions of processor 601. Memory 602 can be implemented by any type of volatile or non-volatile storage device or a combination of them, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.
[0095] When the execution instructions in the memory 602 are executed by the processor 601 , the electronic device 600 is enabled to execute part or all of the steps in the above method embodiment.
[0096] Corresponding to the above embodiment, the embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium may store a program, wherein when the program is running, the device where the computer-readable storage medium is located may be controlled to perform some or all of the steps in the above method embodiment. In a specific implementation, the computer-readable storage medium may be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).
[0097] Corresponding to the above-mentioned embodiment, the embodiment of the present application further provides a computer program product, which includes executable instructions. When the executable instructions are executed on a computer, the computer executes some or all of the steps in the above-mentioned method embodiment.
[0098] In the embodiments of the present application, "at least one" refers to one or more, and "plurality" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.
[0099] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0100] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0101] In several embodiments provided in the present application, any function can be stored in a computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., and other media that can store program codes.
[0102] The above is only a specific implementation of the present application. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. The protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A machine learning modeling method for predicting shale gas production capacity, characterized in that: include: Collect and process target data and organize them into target data sets; Pre-train a large number of offline data sets and obtain the optimal machine learning model to establish an offline data warehouse; According to the target data set, an initialization model is selected from the offline data warehouse, and an optimal model is established by fusion to perform capacity prediction; The main controlling factors of the production capacity of the optimal model are analyzed.
2. A machine learning modeling method for predicting shale gas production capacity according to claim 1, characterized in that: The collecting and processing of target data to form a target data set includes: collecting parameters related to shale gas exploration, development and production as target data; and processing the target data to obtain the target data set.
3. A machine learning modeling method for predicting shale gas production capacity according to claim 2, characterized in that: The parameters include: geological parameters, engineering parameters and production capacity characterization data.
4. The machine learning modeling method for predicting shale gas production capacity according to claim 1, characterized in that: The method of pre-training a large number of offline data sets and obtaining an optimal machine learning model and establishing an offline data warehouse includes: Performing meta-feature extraction and machine learning model training on each offline data set in a large number of offline data sets to obtain an optimal machine learning model for each offline data set; The meta-features of each offline data set and its corresponding optimal machine learning model are stored in the data warehouse to establish an offline data warehouse.
5. The machine learning modeling method for predicting shale gas production capacity according to claim 1, characterized in that: The method of selecting an initialization model from the offline data warehouse according to the target data set, fusing and establishing an optimal model, and performing capacity prediction includes: Extracting meta-features of the target data set, respectively calculating the L1 distances between the meta-features of the target data set and the meta-features of each offline data set, and selecting optimal machine learning models corresponding to several groups of offline data sets with the smallest L1 distances as initialization models; The Bayesian optimization method based on the random forest proxy function is used to search for the hyperparameters of each initialization model and obtain the optimal hyperparameters of each initialization model; The integrated selection method is used to fuse the initialization models, establish an integrated machine learning model, perform capacity forecasting, and output the final capacity forecasting results.
6. A machine learning modeling method for predicting shale gas production capacity according to claim 5, characterized in that: The calculation formula of the L1 distance is: Where N represents the number of meta-features of the target dataset and the offline dataset; x m represents the mth meta-feature of the offline dataset; f m Represents the mth meta-feature of the target dataset.
7. The machine learning modeling method for predicting shale gas production capacity according to claim 3, characterized in that: The analysis of the main control factors of the production capacity of the optimal model includes: Fine-tune the values of various geological engineering parameters to obtain a fine-tuning target data set; Calculate the model errors ε and ε corresponding to the original target dataset and the fine-tuned target dataset respectively p ; According to (ε p -ε) to analyze the impact of each parameter on production capacity and rank them by importance.
8. A machine learning modeling device for predicting shale gas production capacity, characterized in that: include: A target data set generation module is used to collect and process target data and organize them into a target data set; Offline data warehouse establishment module, used to pre-train a large number of offline data sets and obtain the optimal machine learning model to establish an offline data warehouse; A capacity prediction model is used to select an initialization model from the offline data warehouse according to the target data set, fuse and establish an optimal model, and perform capacity prediction; The model interpretation module is used to analyze the main control factors of the production capacity of the optimal model.
9. An electronic device, characterized in that: include: processor; Memory; And a computer program, wherein the computer program is stored in the memory, and the computer program includes instructions, and when the instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.