Intelligent prediction method for carbon dioxide output of CCUS-EOR production well based on KAN network
Through a KAN network-based method, combined with gated neural network and KAN network, the accuracy of long-term prediction of carbon dioxide output in oil fields is solved, and more accurate carbon dioxide output prediction and gas traversing prediction are achieved, which is suitable for the field of oil and natural gas engineering.
Patent Information
- Application Number
- CN202510643672.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-19
- Publication Date
- 2025-08-15
AI Technical Summary
The prior art is difficult to accurately predict oilfield carbon dioxide output, especially on long-term time scales, and fails to effectively capture the characteristic changes caused by stratigraphic changes.
Using a KAN network-based method, combined with gated neural network and Kolmogorov–Arnold Networks (KAN network), data preprocessing and model training are carried out on geological parameters, intelligent prediction model for carbon dioxide output is constructed, and time-sequential geological feature extraction is performed through the streamlined gated neural network sGRU, and the KAN network is cascaded to enhance the perception ability of the model.
It realizes more accurate and real-time prediction of carbon dioxide output, improves prediction capabilities and anti-interference capabilities, provides a basis for air raid prediction, and the system design is modular and lightweight, suitable for low-cost computing equipment.
Smart Images
Figure CN120494193A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of oil and gas engineering, and in particular to a method for intelligently predicting carbon dioxide output of CCUS-EOR production wells based on a KAN network. Background Art
[0002] Most current prediction methods are limited to predicting the overall oil production and volume of associated gas generated by an oil field. Since the composition of associated gas will continue to change as production progresses due to changes in the impact situation, they are unable to predict the output of carbon dioxide. In addition, most of the methods used on-site, although taking into account the timing, do not fully consider the injection and flow pressure conditions of the entire well group, and cannot capture the characteristic changes brought about by formation changes. The long-term prediction capability (more than 20 years) is relatively weak. Summary of the Invention
[0003] In view of this, the present invention provides a method for intelligently predicting carbon dioxide output in CCUS-EOR production wells based on a KAN network to solve the above problems.
[0004] The present invention provides a method for intelligently predicting carbon dioxide output in CCUS-EOR production wells based on a KAN network, comprising: constructing a block training data set according to the production dynamics and geological parameters of a research block; preprocessing the block training data in the block training data set; performing model training based on the preprocessed block training data to obtain an intelligent prediction model for carbon dioxide output, wherein the intelligent prediction model for carbon dioxide output includes a gated neural network and a KAN network; and inputting geological parameter data of the block to be predicted into the intelligent prediction model for prediction to obtain a prediction result for the carbon dioxide output of the block to be predicted.
[0005] In another implementation of the present invention, the preprocessing includes data cleaning, data standardization and normalization processing.
[0006] In another implementation of the present invention, the gated neural network is shown as follows:
[0007] h t =sGRU(p t-1 ,x t )
[0008] Among them, sGRU is a simplified gated neural network, p t-1 is the predicted value at the previous time state, x t is the input vector to the gated neural network at time t.
[0009] In another implementation of the present invention, the KAN network is represented as:
[0010]
[0011] Among them, KAN is the KAN network function, Layers is the number of network layers, and Φ is the B-spline function.
[0012] In another implementation of the present invention, the method further includes: drawing a learning error curve, using R 2 As an evaluation parameter, the hyperparameter optimization of the carbon dioxide output intelligent prediction model is performed.
[0013] In another implementation of the present invention, the geological parameter data of the block to be predicted includes time, corresponding bottom hole oil pressure, well group water injection volume, gas injection rate, oil production, liquid production, gas production and gas injection pressure.
[0014] In another implementation of the present invention, it further includes: making reasonable planning and decision-making according to the prediction result of the carbon dioxide output of the block to be predicted.
[0015] Another aspect of the present invention provides a KAN network-based intelligent prediction system for carbon dioxide output in CCUS-EOR production wells, comprising: a data processing module: constructing a block training data set based on the production dynamics and geological parameters of the study block; preprocessing the block training data in the block training data set; a model training module: performing model training based on the preprocessed block training data to obtain an intelligent prediction model for carbon dioxide output, wherein the intelligent prediction model for carbon dioxide output includes a gated neural network and a KAN network; and a result prediction module: inputting the geological parameter data of the block to be predicted into the intelligent prediction model for carbon dioxide output to perform prediction, thereby obtaining a prediction result for carbon dioxide output of the block to be predicted.
[0016] In another aspect of the present invention, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for intelligently predicting carbon dioxide output in CCUS-EOR production wells based on a KAN network as described above are implemented.
[0017] Another aspect of the present invention provides a computer storage medium, characterized in that a computer program is stored on the computer storage medium, and when the computer program is executed by a processor, the steps of the intelligent prediction method of carbon dioxide output of CCUS-EOR production wells based on the KAN network as described in any of the above are implemented.
[0018] The KAN network-based intelligent prediction method for carbon dioxide output in CCUS-EOR production wells of the present invention designs a streamlined gated neural network sGRU to extract time-series geological features from the collected data. The cascaded KAN network enhances the model's perception of time series and geological fuzzy information to achieve the purpose of efficient numerical prediction. Compared with other production prediction technologies, this solution has significant advantages in prediction ability and accuracy, anti-interference ability, modular design, real-time learning, and model lightweighting. It can predict the carbon dioxide output of production wells in a more comprehensive, real-time, and accurate manner, and can also provide a basis for the prediction of gas channeling. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. By reading the detailed description of the embodiments below, the advantages and benefits of the solutions will become clear to those skilled in the art. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. In the drawings:
[0020] Figure 1 The figure is a flow chart of a method for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network according to an embodiment of the present invention.
[0021] Figure 2 Schematic diagram of a gated neural network according to an embodiment of the present invention.
[0022] Figure 3 FIG. 4 is a schematic diagram of a KAN network according to an embodiment of the present invention.
[0023] Figure 4 FIG. 1 is a schematic diagram of a block training data set according to an embodiment of the present invention.
[0024] Figure 5 Schematic diagram of a model learning curve according to an embodiment of the present invention.
[0025] Figure 6 Schematic diagram for comparing effects of an embodiment of the present invention. DETAILED DESCRIPTION
[0026] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and detailedly described below in conjunction with the accompanying 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 them. All other embodiments obtained by those skilled in the art based on the embodiments in the embodiments of the present invention should fall within the scope of protection of the embodiments of the present invention.
[0027] Figure 1 A schematic flow chart of a method for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network is provided in an embodiment of the present invention. Figure 1 As shown, this embodiment mainly includes:
[0028] S101. Construct a block training data set based on the production dynamics and geological parameters of the research block.
[0029] S102: Preprocess the block training data in the block training data set.
[0030] S103 , performing model training based on the pre-processed block training data to obtain a carbon dioxide output intelligent prediction model, wherein the carbon dioxide output intelligent prediction model includes a gated neural network and a KAN network (Kolmogorov–Arnold Networks).
[0031] S104: Input the geological parameter data of the block to be predicted into the carbon dioxide output intelligent prediction model to perform prediction, and obtain the carbon dioxide output prediction result of the block to be predicted.
[0032] The KAN network-based intelligent prediction method for carbon dioxide output in CCUS-EOR production wells of the present invention designs a streamlined gated neural network sGRU to extract time-series geological features from the collected data. The cascaded KAN network enhances the model's perception of time series and geological fuzzy information to achieve the purpose of efficient numerical prediction. Compared with other production prediction technologies, this solution has significant advantages in prediction ability and accuracy, anti-interference ability, modular design, real-time learning, and model lightweighting. It can predict the carbon dioxide output of production wells in a more comprehensive, real-time, and accurate manner, and can also provide a basis for the prediction of gas channeling.
[0033] In another implementation of the present invention, the preprocessing includes data cleaning, data standardization and normalization processing.
[0034] For example, Figure 4 As shown, a block training dataset is created based on the production dynamics and geological parameters of the study block, and the collected data is preprocessed. Data cleaning includes checking the dataset for missing values and, if so, filling them with the mean. Data standardization involves using a normalizer to convert each column of features to a mean of 0 and a standard deviation of 1 to prevent differences in feature scales from adversely affecting the model. Normalization involves using a min-max scaler to normalize the target variable y, converting it to the range [0, 1]. Even in situations where the input data is distorted or missing, the model can still achieve good prediction accuracy, ensuring the reliability of the prediction.
[0035] In another implementation of the present invention, the gated neural network is shown as follows:
[0036] h t =sGRU(p t-1 ,x t )
[0037] Among them, sGRU is a simplified gated neural network, p t-1 is the predicted value at the previous time state, x t is the input vector to the gated neural network at time t.
[0038] For example, Figure 2 As shown, the simplified gated neural network is used to extract the time series features in the data, which consists of two basic gate structures: updating the gate structure z t and candidate hidden layer gating states The gating structure z will be updated t As weights, the specific structures of sGRU and two gating structures are expressed as follows:
[0039]
[0040] Among them, h t is the hidden state at time t; h t-1 is the hidden state at time t-1; x t is the input vector at time t; z t It is the update gate, whose value range is between 0 and 1, and determines the fusion ratio of new feature information and old feature information; is the candidate hidden layer gating state, which represents the possible new hidden layer state at time t; ⊙ represents the Hadamard product; σ is the Sigmoid activation function, where b is the bias, which is 0 by default in the invention and is used to compress the input to between 0 and 1; tanh is the hyperbolic tangent function, which is used to compress the input to between -1 and 1 and activate the candidate state; L is the fully connected layer, which converts the input data x t Mapped to the output space through matrix multiplication.
[0041] This invention streamlines traditional gated networks, removing unnecessary cascade layers to facilitate long-term predictions. Each network unit (streamlined gated neural network, KAN network) and input and output interfaces (water production, oil production, gas production, CO2 production from the associated gas skid-mounted processing unit, and wellhead pressure) utilizes a modular, independent parameter design, eliminating anomalies caused by missing parameters and facilitating diverse data maintenance.
[0042] In another implementation of the present invention, the KAN network is represented as:
[0043]
[0044] Among them, KAN is the KAN network function, Layers is the number of network layers, and Φ is the B-spline function.
[0045] For example, Figure 3 As shown, the KAN network is based on the Kolmogorov-Arnold representation theorem, which states that a real-valued, smooth, and continuous multivariate function between 0 and 1 can be approximated by the superposition of a finite number of single-variable functions. Specifically, it can be expressed as:
[0046]
[0047] Where Φ is a real function, which is a B-spline function in the KAN network, and φ is a real-valued, smooth, and continuous multivariate function ranging between 0 and 1. In the KAN network, each KAN layer is defined by a matrix consisting of a series of univariate functions ranging from 1 to n, indicating the number of input variables, and j ranging from 1 to N, indicating the number of output variables. Each function is a trainable spline function.
[0048] The KAN network extracts geological characteristic information based on the time series characteristic information after the simplified gated neural network, and outputs the predicted carbon dioxide output prediction matrix P = [p1, ..., p t In the present invention, the KAN network is based on the Kolmogorov-Arnold representation theorem to summarize and express the relationship between geological and production data to form an auxiliary model.
[0049] In another implementation of the present invention, the method further includes: drawing a learning error curve, using R 2 As an evaluation parameter, the hyperparameter optimization of the carbon dioxide output intelligent prediction model is performed.
[0050] For example, we use learning curve and hyperparameter optimization to optimize: First, we draw the learning error curve and use R 2 As evaluation parameters, such as Figure 5 As shown in the figure, it shows the training score and validation score of the model under different training set sizes, calculates and plots the error of the training set and cross-validation set, then performs hyperparameter optimization and prints the optimal parameters.
[0051] In another implementation of the present invention, the geological parameter data of the block to be predicted includes time, corresponding bottom hole oil pressure, well group water injection volume, gas injection rate, oil production, liquid production, gas production and gas injection pressure.
[0052] For example, the present invention uses the water production, oil production, gas production, carbon dioxide production of the associated gas skid-mounted processing device, and wellhead pressure data of the production wells in the X block of the eastern A oil reservoir, and preprocesses them to obtain prediction data, whose characteristic matrix is represented by X=[x1,…,x t ] represents. Use the function to load the previously trained model, input the data feature matrix X into the loaded model to predict the predicted CO2 output at the predicted time.
[0053] This system collects daily parameter data for each production unit, including water, oil, and gas production, CO2 production from associated gas skid-mounted processing units, and wellhead pressure. The model quickly analyzes changes in formation information based on historical data, enabling more accurate subsequent forecasts. The system's modular and streamlined design reduces hardware computing requirements, allowing for flexible selection of low-cost computing modules and compatibility with older computing devices.
[0054] In another implementation of the present invention, it further includes: making reasonable planning and decision-making according to the prediction result of the carbon dioxide output of the block to be predicted.
[0055] In another implementation of the present invention, Figure 6 As shown, experimental comparison data of the intelligent prediction method for carbon dioxide output of CCUS-EOR production wells of the present invention and the prior art are provided.
[0056] Table 1. Comparison of performance evaluation of common methods
[0057]
[0058] By comparing Table 1, it can be seen that the four different deep learning algorithms are preferred for predicting the multi-component gas flooding front. Among them, the prediction results of this method are the best, and it can accurately predict the specific content of carbon dioxide produced by production wells in CCUS-EOR projects.
[0059] Another aspect of the present invention provides a KAN network-based intelligent prediction system for carbon dioxide output from CCUS-EOR production wells, comprising:
[0060] Data processing module: constructs a block training data set based on the production dynamics and geological parameters of the research block; and preprocesses the block training data in the block training data set.
[0061] Model training module: Model training is performed based on the preprocessed block training data to obtain a carbon dioxide output intelligent prediction model, which includes a gated neural network and a KAN network.
[0062] Result prediction module: inputs the geological parameter data of the block to be predicted into the carbon dioxide output intelligent prediction model for prediction, and obtains the carbon dioxide output prediction result of the block to be predicted.
[0063] The KAN network-based intelligent prediction system for carbon dioxide output in CCUS-EOR production wells of the present invention designs a streamlined gated neural network sGRU to extract time-series geological features from the collected data. The cascaded KAN network enhances the model's perception of time series and geological fuzzy information to achieve the purpose of efficient numerical prediction. Compared with other production prediction technologies, this solution has significant advantages in prediction ability and accuracy, anti-interference ability, modular design, real-time learning, and model lightweighting. It can predict the carbon dioxide output of production wells in a more comprehensive, real-time, and accurate manner, and can also provide a basis for the prediction of gas channeling.
[0064] In another aspect of the present invention, an electronic device includes a processor, a memory, a communication bus, and a communication interface.
[0065] in:
[0066] The processor, memory and communication interface communicate with each other through a communication bus.
[0067] Communication interface, used to communicate with other electronic devices or servers.
[0068] The processor is configured to execute a program, specifically executing the steps of any one of the KAN network-based CCUS-EOR production well carbon dioxide output intelligent prediction methods in the above embodiments.
[0069] Specifically, the program may include program codes including computer operation instructions.
[0070] The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; or different types of processors, such as one or more CPUs and one or more ASICs.
[0071] Memory is used to store programs. The memory may include high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk storage.
[0072] The program can be specifically configured to cause a processor to execute the steps of any of the methods for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network as described in the embodiments. The specific implementation of each step in the program can be found in the corresponding descriptions of the steps and units executed in any of the aforementioned methods for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network, and is not further described here. Those skilled in the art will clearly understand that, for ease and brevity of description, the specific operating processes of the devices and modules described above can be referenced to the corresponding process descriptions in the aforementioned method embodiments.
[0073] The exemplary embodiments of the present application further provide a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the methods of the various embodiments of the present application.
[0074] The method according to the embodiment of the present invention described above can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk or magneto-optical disk), or as computer code that is originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded via a network and will be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor or programmable or dedicated hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown here, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown here.
[0075] Thus far, specific embodiments of the present invention have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. Additionally, the processes depicted in the accompanying drawings do not necessarily require the specific order shown, or sequential order, to achieve the desired results.
[0076] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, back, etc.) are only used to explain the relative position relationship between the components in a certain specific order (as shown in the accompanying drawings). If the specific order changes, the directional indication will also change accordingly.
[0077] In the description of the present invention, the terms "first" and "second" are used solely to facilitate description of different components or names and should not be construed as indicating or implying a sequential relationship, relative importance, or implicitly specifying the quantity of the technical features being described. Therefore, features specified as "first" or "second" may explicitly or implicitly include at least one of such features.
[0078] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.
[0079] It should be noted that although the specific embodiments of the present invention are described in detail in conjunction with the accompanying drawings, this should not be construed as limiting the scope of protection of the present invention. Within the scope described by the claims, various modifications and variations that can be made by those skilled in the art without creative effort still fall within the scope of protection of the present invention.
[0080] The examples of the embodiments of the present invention are intended to briefly illustrate the technical features of the embodiments of the present invention so that those skilled in the art can intuitively understand the technical features of the embodiments of the present invention, and are not intended to improperly limit the embodiments of the present invention.
[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.
Claims
1. A method for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network, characterized in that: include: Construct a block training dataset based on the production performance and geological parameters of the research block; Preprocessing the block training data in the block training data set; Performing model training based on the preprocessed block training data to obtain a carbon dioxide output intelligent prediction model, wherein the carbon dioxide output intelligent prediction model includes a gated neural network and a KAN network; The geological parameter data of the block to be predicted is input into the carbon dioxide output intelligent prediction model for prediction to obtain the carbon dioxide output prediction result of the block to be predicted.
2. The method according to claim 1, characterized in that The preprocessing includes data cleaning, data standardization and normalization processing.
3. The method according to claim 1, characterized in that The gated neural network is shown in the following formula: h t =sGRU(p t-1 ,x t ) Among them, sGRU is a simplified gated neural network, p t-1 is the predicted value at the previous time state, x t is the input vector to the gated neural network at time t.
4. The method according to claim 3, characterized in that The KAN network is expressed as: Among them, KAN is the KAN network function, Layers is the number of network layers, and Φ is the B-spline function.
5. The method according to claim 1, wherein Also includes: Plot the learning error curve using R 2 As an evaluation parameter, the hyperparameter optimization of the carbon dioxide output intelligent prediction model is performed.
6. The method according to claim 1, characterized in that The geological parameter data of the block to be predicted include time, corresponding bottom hole oil pressure, well group water injection volume, gas injection rate, oil production, liquid production, gas production and gas injection pressure.
7. The method according to claim 5, characterized in that Also includes: Make reasonable plans and decisions based on the carbon dioxide output prediction results of the block to be predicted.
8. A CCUS-EOR production well carbon dioxide output intelligent prediction system based on KAN network, characterized by: include: Data processing module: Constructs block training data set based on the production dynamics and geological parameters of the research block; Preprocessing the block training data in the block training data set; Model training module: performs model training based on the pre-processed block training data to obtain a carbon dioxide output intelligent prediction model, which includes a gated neural network and a KAN network; Result prediction module: inputs the geological parameter data of the block to be predicted into the carbon dioxide output intelligent prediction model for prediction, and obtains the carbon dioxide output prediction result of the block to be predicted.
9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of a method for intelligently predicting carbon dioxide output from CCUS-EOR production wells based on a KAN network are implemented as described in any one of claims 1 to 7.
10. A computer storage medium, characterized in that The computer storage medium stores a computer program, which, when executed by a processor, implements the steps of the intelligent prediction method for carbon dioxide output of CCUS-EOR production wells based on a KAN network according to any one of claims 1 to 7.