A method for predicting the productivity of tight gas wells
Through machine learning-based methods, a model that can quickly and efficiently predict tight gas well production capacity is established, which solves the problems of low accuracy and cumbersome work in traditional methods when dealing with heterogeneous gas reservoirs and complex gas-water relationships, and improves the real-time and accuracy of predictions.
Patent Information
- Application Number
- CN202411859672.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-17
- Publication Date
- 2025-04-08
Smart Images
Figure CN119783885A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of oil reservoir exploration and development, and in particular to a method for predicting the productivity of a tight gas well. Background Art
[0002] At present, oil and gas exploration and development has gradually shifted from traditional conventional gas reservoirs to unconventional gas reservoirs, among which the proportion of tight gas production has increased year by year.
[0003] In the early stages of gas well development, it is usually necessary to predict its production capacity. Traditional production capacity prediction methods mainly include mathematical analytical models, numerical simulation methods, and material balance methods. These physical models have achieved relatively practical results in traditional gas reservoir development, but they face some challenges when applied to tight gas wells. For example, mathematical analytical models are usually based on idealized assumptions and are difficult to truly reflect the heterogeneity of tight gas reservoirs; and although numerical simulation methods are accurate, they have difficulties in dealing with the complex gas-water relationship in tight gas reservoirs, and the process is cumbersome and time-consuming.
[0004] Therefore, researching and developing new production capacity prediction methods suitable for tight gas reservoirs has important theoretical significance and practical value, especially for improving the development efficiency and economy of tight gas reservoirs. Summary of the invention
[0005] The present invention provides a method for predicting the productivity of tight gas wells. Based on a machine learning method, the method is trained and analyzed through a large amount of data, and can quickly and efficiently complete the task of productivity prediction, showing strong adaptability.
[0006] To achieve the above object, the present invention adopts the following technical solutions:
[0007] In a first aspect, the present application provides a method for predicting the productivity of a tight gas well, comprising:
[0008] Data collection steps: collecting basic data of various preset parameters of tight gas wells as references, and preprocessing the basic data using statistical analysis methods to form a basic data set;
[0009] Correlation analysis step: based on a variety of preset correlation analysis methods, the input parameters of the prediction model to be established are screened from the basic data set;
[0010] Model building step: based on the data corresponding to the input parameters in the basic data set, the machine learning model of the preset type network architecture is trained, and the model with the best prediction effect is selected from the trained machine learning models as the prediction model;
[0011] Model application steps: for the tight gas well whose production capacity is to be predicted, obtain the data values of its input parameters and input them into the prediction model to obtain the predicted production capacity.
[0012] In one implementation, in the data collection step, the plurality of preset parameters include: gas well geological parameters, well logging parameters, lithology parameters, engineering parameters and production test data.
[0013] In one implementation, in the data collection step, the basic data is preprocessed using a statistical analysis method, including:
[0014] Determine data outliers in basic data based on preset indicators and perform correction processing on the data outliers; and
[0015] Fill in missing data values.
[0016] In one implementation, a calculation formula for determining a preset indicator of data outliers includes:
[0017]
[0018] Where X is the data point, μ is the mean of the underlying data, and σ is the standard deviation of the underlying data.
[0019] In one implementation, the calculation formula for filling in missing data values is:
[0020]
[0021] where y i are the known values of the K neighbors.
[0022] In one implementation, in the correlation analysis step, a threshold is set for each correlation analysis method to determine the strongly correlated influencing parameters of its corresponding sorting order; then, a preset number of parameters are jointly screened out from the strongly correlated influencing parameters determined by all correlation analysis methods as input parameters of the prediction model.
[0023] In one implementation, the correlation analysis method includes a Pearson correlation coefficient method and a mutual information method;
[0024] Pearson correlation coefficient calculation formula:
[0025]
[0026] Among them, r is the correlation coefficient value; x i are different values corresponding to the variable x; is the mean value of variable x; y i are different values corresponding to the variable y; is the average value of variable y; n is the number of variables;
[0027] The formula of the mutual information method is:
[0028]
[0029] Here, p(x,y) is the joint probability distribution, and p(x) and p(y) are the marginal probability distributions of X and Y respectively.
[0030] In one implementation, the determined input parameters include: gas well reservoir thickness, gas well production water-to-gas ratio, gas well reservoir gas saturation, average daily water production of the gas well, amount of liquid nitrogen injected for gas well fracturing, gas well reservoir porosity, actual amount of sand injected for gas well fracturing and gas well reservoir resistivity.
[0031] In one implementation, in the model building step, the machine learning models of the preset type network architecture include: long short-term memory network, random forest regression model.
[0032] In one implementation, the model building step further includes:
[0033] Optimize the trained model based on random search method;
[0034] Evaluate the models based on the root mean square error and coefficient of determination to select the best prediction model; and
[0035] Based on the prediction model, physical interpretability verification is carried out from the perspective of basic theory.
[0036] The present invention adopts the above technical solution, which has the following advantages:
[0037] 1. The present invention can solve the problem of tight gas reservoir production capacity prediction in the early stage of development. The existing empirical method has low accuracy. This method starts from basic actual data to conduct tight gas well production capacity prediction research. According to the linear and nonlinear relationship between influencing factors and production capacity, the weights of different evaluation methods are calculated to determine the main controlling factors of production capacity.
[0038] 2. The present invention breaks through the problems of harsh application conditions and cumbersome work of traditional prediction models, and greatly improves the real-time and accuracy of tight gas reservoir production capacity prediction in the early stage of development.
[0039] 3. The present invention has universality and extensibility, and can provide reference for the application of machine learning methods in the prediction of gas well productivity in different blocks.
[0040] 4. The present invention can solve the problem that current machine learning models lack physical meaning in exploration and development, and provide a basis for the application of machine learning models in exploration and development.
[0041] In summary, the present invention can be widely used in the technical field of oil reservoir exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a schematic diagram of a method flow provided by an embodiment of the present invention;
[0043] Figure 2 is a schematic diagram of correlation analysis provided by an embodiment of the present invention;
[0044] Figure 3 It is a schematic diagram of training results after optimization of a machine learning model provided by an embodiment of the present invention;
[0045] Figure 4 It is a schematic diagram of prediction results of a machine learning model test set provided by an embodiment of the present invention;
[0046] Figure 5 It is a schematic diagram of explaining the LIME model provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.
[0048] It should be understood that the terms used herein are only for the purpose of describing specific example embodiments and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "include", "comprise", "contain", and "have" are inclusive, and therefore specify the existence of stated features, steps, operations, elements and / or parts, but do not exclude the existence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not interpreted as necessarily requiring them to be performed in the specific order described or illustrated, unless the execution order is clearly indicated. It should also be understood that additional or alternative steps may be used.
[0049] Although the terms first, second, third, etc. can be used in the text to describe multiple elements, components, regions, layers and / or sections, these elements, components, regions, layers and / or sections should not be limited by these terms. These terms can only be used to distinguish an element, component, region, layer or section from another region, layer or section. Unless the context clearly indicates, terms such as "first", "second" and other numerical terms do not imply order or sequence when used in the text. Therefore, the first element, component, region, layer or section discussed below can be referred to as the second element, component, region, layer or section without departing from the teaching of the example embodiments.
[0050] The method and system for predicting the production capacity of tight gas wells provided in the embodiments of the present invention make full use of the actual production data of the gas wells to establish a machine learning production capacity prediction model, which overcomes the problems of harsh application conditions and cumbersome work of traditional prediction models, and greatly improves the real-time and accuracy of on-site tight gas well production capacity prediction.
[0051] In view of the defects and problems of existing technologies, such as Figure 1 The present application provides a method for predicting the productivity of a tight gas well, comprising:
[0052] (1) Data collection step: collecting basic data of various preset parameters of tight gas wells as reference, and preprocessing the basic data using statistical analysis methods to form a basic data set;
[0053] Specifically, the production data of the production wells in the target block are collected to obtain a production well database of the target block, and data preprocessing is completed.
[0054] Among them, there are a variety of preset parameters, including: gas well geological parameters, logging parameters, lithology parameters, engineering parameters and production test data.
[0055] For the target block, we collected gas well logging, geological and production data, and used the open flow rate as the target value, and the rest were parameters affecting production capacity. We classified the gas wells based on the open flow rate. We used the Z-score method to detect data outliers.
[0056] The formula of the Z-score method is:
[0057]
[0058] Where X is the data point, μ is the mean of the data set, and σ is the standard deviation of the data set.
[0059] Preprocessing also includes processing missing values.
[0060] The K-Nearest Neighbors method is used to handle missing values, and the formula is:
[0061]
[0062] where y i are the known values of the K neighbors.
[0063] (2) Correlation analysis step: Based on a plurality of preset correlation analysis methods, the input parameters of the prediction model to be established are screened from the basic data set;
[0064] Specifically, the dynamic and static data of tight gas well production are analyzed, the correlation between the parameters affecting gas well productivity in the target block and the productivity of the production wells is calculated, and the main controlling factors of gas well productivity are screened.
[0065] First, the data set is standardized to avoid interference caused by different dimensions of the main controlling factors.
[0066] The normalization formula is:
[0067]
[0068] Among them, X1 is the standardized data; X is the original data; X min is the minimum value in the original data; X max is the maximum value of the original data.
[0069] Furthermore, based on the standardized tight gas well data set, the Pearson correlation coefficient, Spearman correlation coefficient, and Kendall correlation coefficient methods are used to evaluate the main control parameters of tight gas well productivity from linear and nonlinear perspectives, respectively. The main control parameters are determined in combination with actual production, such as Figure 2 Indicates.
[0070] Pearson correlation coefficient calculation formula:
[0071]
[0072] Among them, r is the correlation coefficient value; x i are different values corresponding to the variable x; is the mean value of variable x; y i are different values corresponding to the variable y; is the mean value of variable y; n is the number of variables.
[0073] Furthermore, the mutual information method formula is:
[0074]
[0075] Here, p(x,y) is the joint probability distribution, and p(x) and p(y) are the marginal probability distributions of X and Y respectively.
[0076] (3) Model construction step: Based on the data corresponding to the input parameters in the basic data set, the machine learning model of the preset type network architecture is trained, and the model with the best prediction effect is selected from the trained machine learning models as the prediction model;
[0077] Specifically, based on the determined main controlling factors of gas well production capacity, a machine learning production capacity prediction model group for tight gas wells is established, and the model is optimized. The prediction results of different models are compared and evaluated to determine the optimal machine learning production capacity prediction model.
[0078] The machine learning capacity prediction model group includes long short-term memory network and random forest regression model.
[0079] Enter the main control factors and target values of production capacity to complete model training, such as Figure 3 shown.
[0080] Furthermore, the root mean square error (RMSE) and the coefficient of determination (R 2 ) is the evaluation index,
[0081] The root mean square error formula is:
[0082]
[0083] Where m is the sample size; yi is the actual value; is the predicted value.
[0084] The formula for the coefficient of determination is:
[0085]
[0086] Where m is the number of samples; y i is the actual value; is the predicted value; is the average value.
[0087] This step also includes optimizing the model based on the random search method to improve the training accuracy, such as Figure 4 shown.
[0088] In this step, from the perspective of machine learning, the test set is used to predict new wells that have not participated in training, and the results are compared. Figure 5 shown.
[0089] (4) Model application step: For the tight gas well whose production capacity is to be predicted, obtain the data value of its input parameter and input it into the prediction model to obtain the predicted production capacity.
[0090] This step can also be further based on the best machine learning capacity prediction model. For the main controlling factors of the input, the LIME method can be used to analyze the contribution of each input feature in the selected best machine learning capacity prediction model to each prediction result. This can provide a clear explanation for each sample (or each prediction), showing which features have a positive (or negative) impact on the prediction results, and the size of their impact, and compare them with the previous correlation analysis results.
[0091] In another aspect of the embodiment of the present application, a system for predicting the productivity of a tight gas well is provided, comprising:
[0092] The tight gas well production data analysis module is used to collect and pre-process tight gas well data and analyze and determine the main controlling factors of tight gas well production capacity under different conditions.
[0093] The tight gas well machine learning capacity prediction model module is used to compare and analyze the prediction accuracy and practical application significance of various machine learning models under different conditions, and select the best model to complete the unobstructed flow prediction.
[0094] The physical verification module of the tight gas well machine learning capacity prediction model is used to solve the problem that the current machine learning model lacks physical meaning in oil and gas field development, and provide a strong basis for the application of machine learning models.
[0095] On the other hand, an embodiment of the present application provides a processing device corresponding to a method for predicting the production capacity of a tight gas well. The processing device can be applicable to a client processing device, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the method of the aforementioned embodiment.
[0096] The processing device includes a processor, a memory, a communication interface and a bus, and the processor, the memory and the communication interface are connected through the bus to complete mutual communication. The memory stores a computer program that can be run on the processing device, and the processing device executes the method for predicting the production capacity of a tight gas well provided in the embodiment of the present application when running the computer program.
[0097] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.
[0098] In some other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors of various types, which are not limited herein.
[0099] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes 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 invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0100] Those skilled in the art will understand that the structure of the above-mentioned computing device is only a partial structure related to the scheme of the present application, and does not constitute a limitation on the computing device to which the scheme of the present application is applied. The specific computing device may include more or fewer components, or combine certain components, or have a different arrangement of components.
[0101] An embodiment of the present application also provides a computer program product for tight gas well productivity prediction. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the tight gas well productivity prediction method described in Example 1 are loaded.
[0102] Computer readable storage media can be tangible devices that hold and store instructions used by instruction execution devices. Computer readable storage media can be, for example, but not limited to, electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any combination thereof.
[0103] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.
[0104] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present application. It should be understood that the flowcharts and / or block diagrams may be implemented by computer program instructions. Each flow The combination of the processes and / or blocks in the flowchart and / or block diagram can be provided. Supplying these computer program instructions to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce the instructions for implementing the process Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.
[0105] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 Steps for specifying functions in a box or boxes
[0107] In the several embodiments provided by the present invention, it should be understood that the disclosed method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0108] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for predicting the productivity of a tight gas well, characterized in that: include: Data collection steps: collecting basic data of various preset parameters of tight gas wells as references, and preprocessing the basic data using statistical analysis methods to form a basic data set; Correlation analysis step: based on a variety of preset correlation analysis methods, the input parameters of the prediction model to be established are screened from the basic data set; Model building step: based on the data corresponding to the input parameters in the basic data set, the machine learning model of the preset type network architecture is trained, and the model with the best prediction effect is selected from the trained machine learning models as the prediction model; Model application steps: for the tight gas well whose production capacity is to be predicted, obtain the data values of its input parameters and input them into the prediction model to obtain the predicted production capacity.
2. The method for predicting the productivity of a tight gas well according to claim 1, characterized in that: In the data acquisition step, the plurality of preset parameters include: gas well geological parameters, logging parameters, lithology parameters, engineering parameters and production test data.
3. The method for predicting the productivity of a tight gas well according to claim 1, characterized in that: In the data collection step, statistical analysis methods are used to pre-process the basic data, including: Determine data outliers in basic data based on preset indicators and perform correction processing on the data outliers; and Fill in missing data values.
4. The method for predicting the productivity of a tight gas well according to claim 3, characterized in that: The calculation formula for the preset indicator to determine the data outlier value includes: Where X is the data point, μ is the mean of the underlying data, and σ is the standard deviation of the underlying data.
5. The method for predicting the productivity of a tight gas well according to claim 3, characterized in that: The calculation formula for filling missing values is: where y i are the known values of the K neighbors.
6. The method for predicting the productivity of a tight gas well according to claim 1, characterized in that: In the correlation analysis step, for each correlation analysis method, a threshold is set to determine the strongly correlated influencing parameters of the corresponding sorting order; then, from the strongly correlated influencing parameters determined by all correlation analysis methods, a preset number of parameters are jointly screened out as input parameters of the prediction model.
7. The method for predicting the productivity of a tight gas well according to claim 6, characterized in that: The correlation analysis method includes Pearson correlation coefficient method and mutual information method; Pearson correlation coefficient calculation formula: Among them, r is the correlation coefficient value; x i are different values corresponding to the variable x; is the mean value of variable x; y i are different values corresponding to the variable y; is the average value of variable y; n is the number of variables; The formula of the mutual information method is: Here, p(x,y) is the joint probability distribution, and p(x) and p(y) are the marginal probability distributions of X and Y respectively.
8. The method for predicting the productivity of a tight gas well according to claim 7, characterized in that: The determined input parameters include: gas well reservoir thickness, gas well production water-gas ratio, gas well reservoir gas saturation, gas well average daily water production, gas well fracturing injection liquid nitrogen volume, gas well reservoir porosity, gas well actual fracturing injection sand volume and gas well reservoir resistivity.
9. The method for predicting the productivity of a tight gas well according to claim 1, characterized in that: In the model building step, the machine learning models of the preset types of network architectures include: long short-term memory network, random forest regression model.
10. The method for predicting the productivity of a tight gas well according to claim 9, characterized in that: The model building step also includes: Optimize the trained model based on random search method; Evaluate the models based on the root mean square error and coefficient of determination to select the best prediction model; and Based on the prediction model, physical interpretability verification is carried out from the perspective of basic theory.