CO2-crude oil system minimum miscible pressure prediction method and device based on Fourier neural operator, equipment and medium

Abstract

The invention relates to the field of petroleum exploration and development, and discloses a method, a device, equipment and a medium for predicting the minimum miscible pressure of a CO2-crude oil system based on a Fourier neural operator, which can be used for obtaining the standardized molar content percentage of a plurality of crude oil components in the CO2-crude oil system and the standardized temperature value of the reservoir temperature. And a trained Fourier neural operator FNO model is acquired. And inputting the standardized molar content percentage and the standardized oil reservoir temperature value of each crude oil component into an input layer, a first full-connection layer, a feature processing layer, a second full-connection layer and an output layer of an FNO model to carry out minimum miscible pressure prediction so as to obtain the minimum miscible pressure of the CO2-crude oil system. The FNO model can efficiently capture multi-parameter global dynamic association among crude oil components, oil reservoir temperature and minimum miscible pressure in a CO2-crude oil system in a frequency domain, and the calculation efficiency and prediction precision can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of oil exploration and development, and particularly relates to a CO2-crude oil system minimum miscibility pressure prediction method, device, equipment and medium based on a Fourier neural operator. BACKGROUND

[0002] With the development of science and technology, oil exploration and development technology is continuously improved.

[0003] The minimum miscibility pressure (MMP) of a CO2-crude oil system refers to the lowest pressure required for CO2 and crude oil to reach dynamic miscibility through one or more contacts at a specific reservoir temperature, and is an important parameter in oil exploration.

[0004] The related art generally predicts the minimum miscibility pressure of a CO2-crude oil system through empirical formula regression analysis, simple numerical simulation, and slim tube experiments. However, the related art has low prediction accuracy for the minimum miscibility pressure of a CO2-crude oil system. SUMMARY

[0005] The present application provides a CO2-crude oil system minimum miscibility pressure prediction method, device, equipment and medium based on a Fourier neural operator to solve the problem of low prediction accuracy for the minimum miscibility pressure of a CO2-crude oil system in the related art and improve the prediction accuracy for the minimum miscibility pressure of a CO2-crude oil system.

[0006] In a first aspect, the present application provides a CO2-crude oil system minimum miscibility pressure prediction method based on a Fourier neural operator, comprising: obtaining the standardized mole content percentage of a plurality of crude oil components in a CO2-crude oil system and the standardized temperature value of a reservoir temperature, and obtaining a trained Fourier neural operator (FNO) model; wherein the FNO model is obtained by model training based on a sample CO2-crude oil system, a sample reservoir temperature value, and a corresponding minimum miscibility pressure, and the FNO model includes an input layer, a first fully connected layer, a feature processing layer, a second fully connected layer, and an output layer connected in turn; inputting the standardized mole content percentage of each of the crude oil components and the standardized reservoir temperature value into the input layer of the FNO model for feature conversion to obtain a corresponding low-dimensional feature tensor; inputting the low-dimensional feature tensor into the first fully connected layer for high-dimensional mapping to obtain a corresponding high-dimensional feature tensor; inputting the high-dimensional feature tensor into the feature processing layer, the second fully connected layer, and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0007] Optionally, the obtaining the normalized mole content percentage of a plurality of crude oil components in the CO2-crude oil system and the normalized temperature value of the reservoir temperature comprises: obtaining the original mole content percentage of each of the crude oil components in the CO2-crude oil system and the original temperature value of the reservoir temperature; for the original parameter value of any target parameter, obtaining the maximum parameter value and the minimum parameter value corresponding to the target parameter, and performing maximum-minimum normalization on the original parameter value of the target parameter according to the maximum parameter value, the minimum parameter value corresponding to the target parameter and the original parameter value of the target parameter, to obtain the normalized parameter value of the target parameter; wherein the target parameter is the crude oil component or the reservoir temperature.

[0008] Optionally, the inputting the high-dimensional feature tensor into the feature processing layer, the second fully connected layer and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system comprises: inputting the high-dimensional feature tensor into the feature processing layer for processing to obtain target feature data; inputting the target feature data into the second fully connected layer for dimension reduction and focusing to screen out core feature data; inputting the core feature data into the output layer for feature mapping to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0009] Optionally, the feature processing layer comprises N sub-processing layers connected in sequence, N is an integer greater than 1, and each of the sub-processing layers comprises a Fourier layer and an activation layer.

[0010] Optionally, N is 4.

[0011] Optionally, the inputting the high-dimensional feature tensor into the feature processing layer for processing to obtain target feature data comprises: inputting the high-dimensional feature tensor into a first Fourier layer of a first sub-processing layer of the feature processing layer for global feature capture to obtain global feature data output by the first Fourier layer; inputting the global feature data output by the first Fourier layer into a first activation layer for local nonlinear activation to obtain corresponding first activated feature data; wherein the first activation layer is the activation layer in the first sub-processing layer of the feature processing layer; The first post-activation feature data is input into a second Fourier layer of a second sub-processing layer of the feature processing layer for global feature capturing until post-activation feature data output by the activation layer in a last sub-processing layer of the feature processing layer is obtained and taken as the target feature data.

[0012] Optionally, the plurality of crude oil components in the CO2-crude oil system include at least two of CO2, C1N2, C2, C3+, C7+, C24+, and C36+.

[0013] In a second aspect, the present application provides a CO2-crude oil system minimum miscibility pressure prediction device based on a Fourier neural operator, comprising: An acquisition unit is configured to acquire standardized mole content percentages of a plurality of crude oil components in a CO2-crude oil system and standardized temperature values of reservoir temperatures, and acquire a trained Fourier neural operator (FNO) model; wherein the FNO model is obtained by model training based on sample CO2-crude oil systems, sample reservoir temperature values, and corresponding minimum miscibility pressures, and the FNO model includes an input layer, a first full connection layer, a feature processing layer, a second full connection layer, and an output layer connected in sequence. A conversion unit is configured to input the standardized mole content percentages of each of the crude oil components and the standardized reservoir temperature values into the input layer of the FNO model for feature conversion to obtain a corresponding low-dimensional feature tensor; A mapping unit is configured to input the low-dimensional feature tensor into the first full connection layer for high-dimensional mapping to obtain a corresponding high-dimensional feature tensor; A prediction unit is configured to input the high-dimensional feature tensor into the feature processing layer, the second full connection layer, and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0014] In a third aspect, the present application provides a computer device, comprising a memory and a processor, which are communicatively connected with each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the CO2-crude oil system minimum miscibility pressure prediction method based on a Fourier neural operator of the first aspect or any of the corresponding embodiments thereof.

[0015] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer instructions for causing a computer to perform the CO2-crude oil system minimum miscibility pressure prediction method based on a Fourier neural operator of the first aspect or any of the corresponding embodiments thereof.

[0016] The application provides a CO2-crude oil system minimum miscibility pressure prediction method, device, equipment and medium based on a Fourier neural operator, which can obtain the standardized mole content percentage of a plurality of crude oil components in a CO2-crude oil system and the standardized temperature value of a reservoir temperature, and a trained Fourier neural operator FNO model; wherein the FNO model is obtained by model training based on a sample CO2-crude oil system, a sample reservoir temperature value and a corresponding minimum miscibility pressure, and the FNO model comprises an input layer, a first full connection layer, a feature processing layer, a second full connection layer and an output layer connected in sequence. The standardized mole content percentage and the standardized reservoir temperature value of each crude oil component are input into the input layer of the FNO model for feature conversion, and a corresponding low-dimensional feature tensor is obtained. The low-dimensional feature tensor is input into the first full connection layer for high-dimensional mapping, and a corresponding high-dimensional feature tensor is obtained. The high-dimensional feature tensor is input into the feature processing layer, the second full connection layer and the output layer for minimum miscibility pressure prediction, and the minimum miscibility pressure of the CO2-crude oil system is obtained. The FNO model can efficiently capture the multi-parameter global dynamic correlation among the crude oil components, the reservoir temperature and the minimum miscibility pressure in the CO2-crude oil system in the frequency domain, and can effectively improve the calculation efficiency and the prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present application or related art, the following will briefly introduce the drawings needed to be used in the embodiments or related art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0018] Figure 1 A flowchart of a CO2-crude oil system minimum miscibility pressure prediction method based on a Fourier neural operator provided by the embodiment of the present application; Figure 2 A sample data schematic diagram for training a model provided by the embodiment of the present application; Figure 3 A sample data schematic diagram after standardization provided by the embodiment of the present application; Figure 4 A training flowchart of a FNO model provided by the embodiment of the present application; Figure 5 A model architecture schematic diagram of a FNO model provided by the embodiment of the present application; Figure 6 A model training and verification loss result diagram of a FNO model provided by the embodiment of the present application; Figure 7 A prediction result analysis diagram of a FNO model provided by the embodiment of the present application; Figure 8 A sensitivity analysis result graph of a predicted result provided for an embodiment of the present application; Figure 9 A structural schematic diagram of a CO2-crude oil system minimum miscibility pressure prediction device based on a Fourier neural operator provided for an embodiment of the present application; Figure 10 A structural schematic diagram of a computer device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described below in conjunction with the drawings in the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0020] In related technologies, the prediction methods of the CO2-crude oil system minimum miscibility pressure, such as the slim tube experiment and the empirical formula regression, have the problems of long cycle, dependence on static parameters, poor dynamic adaptability, disconnection between the prediction result and the field demand, and low prediction accuracy.

[0021] The embodiments of the present application will be described below in conjunction with Figures 1-8 The CO2-crude oil system minimum miscibility pressure prediction method based on the Fourier neural operator of the present application is described.

[0022] As shown in the figure, the present embodiment proposes a first CO2-crude oil system minimum miscibility pressure prediction method based on the Fourier neural operator, which can include the following steps: Figure 1 S101, obtaining the standardized mole content percentage of a plurality of crude oil components in a CO2-crude oil system and the standardized temperature value of a reservoir temperature, and obtaining a trained Fourier neural operator FNO model; wherein the FNO model is obtained by model training based on a sample CO2-crude oil system, a sample reservoir temperature value and a corresponding minimum miscibility pressure, and the FNO model includes an input layer, a first full connection layer, a feature processing layer, a second full connection layer and an output layer connected in turn. Optionally, the plurality of crude oil components in the CO2-crude oil system include at least two of CO2, C1N2, C2, C3+, C7+, C24+ and C36+.

[0023] Specifically, the plurality of crude oil components in the CO2-crude oil system can simultaneously include CO2, C1N2, C2, C3+, C7+, C24+ and C36+.

[0024]

[0025] ​Optionally, the obtaining the normalized mole content percentage of each crude oil component in the CO2-crude oil system and the normalized temperature value of the reservoir temperature comprises: obtaining the original mole content percentage of each crude oil component in the CO2-crude oil system and the original temperature value of the reservoir temperature; For the original parameter value of any target parameter, the maximum parameter value and the minimum parameter value corresponding to the target parameter are obtained, and the original parameter value of the target parameter is maximally and minimally standardized according to the maximum parameter value, the minimum parameter value corresponding to the target parameter and the original parameter value of the target parameter, to obtain the standardized parameter value of the target parameter; wherein the target parameter is a crude oil component or a reservoir temperature.

[0026] Specifically, when the target parameter is a crude oil component, the original parameter value of the target parameter is the original mole content percentage of the crude oil component. When the target parameter is a reservoir temperature, the original parameter value of the target parameter is the original temperature value of the reservoir temperature.

[0027] Specifically, the present embodiment can use the created formula to perform minimax standardization processing on the original parameter value of the target parameter. The formula is: .

[0028] wherein, is the standardized parameter value of the target parameter, is the original parameter value of the target parameter, and is the maximum parameter value and the minimum parameter value corresponding to the target parameter.

[0029] Specifically, the present embodiment can collect sample CO2-crude oil systems, sample reservoir temperature values and corresponding minimum miscibility pressures, and use the sample CO2-crude oil systems, sample reservoir temperature values and corresponding minimum miscibility pressures to train a Fourier Neural Operator (FNO) model to be trained to obtain a trained FNO model.

[0030] As shown in Figure 2 , part of the sample CO2-crude oil systems, sample reservoir temperature values and corresponding minimum miscibility pressures collected by the present embodiment. As shown in Figure 3 , the present embodiment can perform standardization processing on the collected sample data to obtain corresponding standardization processing results.

[0031] S102, input the normalized mole content percentage of each crude oil component and the normalized reservoir temperature value into the input layer of the FNO model for feature conversion to obtain a corresponding low-dimensional feature tensor.

[0032] Specifically, the embodiment can input the normalized mole content percentage and the normalized reservoir temperature value of each crude oil component in the CO2-crude oil system into the input layer of the FNO model for feature conversion, to obtain a low-dimensional feature tensor output by the input layer.

[0033] S103, inputting the low-dimensional feature tensor into the first full connection layer for high-dimensional mapping, to obtain a corresponding high-dimensional feature tensor.

[0034] Specifically, the embodiment can input the low-dimensional feature tensor output by the input layer into the first full connection layer for high-dimensional mapping, to obtain a high-dimensional feature tensor output by the first full connection layer.

[0035] S104, inputting the high-dimensional feature tensor into the feature processing layer, the second full connection layer and the output layer for minimum miscibility pressure prediction, to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0036] Specifically, the embodiment can input the high-dimensional feature tensor output by the first full connection layer into the feature processing layer, the second full connection layer and the output layer for forward propagation, to predict the minimum miscibility pressure of the CO2-crude oil system, to obtain the minimum miscibility pressure of the CO2-crude oil system output by the output layer.

[0037] Optionally, in other CO2-crude oil system minimum miscibility pressure prediction methods based on the Fourier neural operator proposed in the embodiment, step S104 includes: inputting the high-dimensional feature tensor into the feature processing layer for processing, to obtain target feature data; inputting the target feature data into the second full connection layer for dimension reduction and focusing, to filter out core feature data; inputting the core feature data into the output layer for feature mapping, to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0038] Optionally, the feature processing layer includes N sub-processing layers connected in sequence, N is an integer greater than 1, and each sub-processing layer includes a Fourier layer and an activation layer.

[0039] Optionally, N is 4.

[0040] Optionally, the above processing of the high-dimensional feature tensor into the feature processing layer to obtain the target feature data includes: inputting the high-dimensional feature tensor into a first Fourier layer of a first sub-processing layer of the feature processing layer for global feature capture, to obtain global feature data output by the first Fourier layer; The global feature data output by the first Fourier layer is input into the first activation layer for local nonlinear activation to obtain corresponding first post-activation feature data; wherein, the first activation layer is an activation layer in the first sub-processing layer of the feature processing layer; The first post-activation feature data is input into the second Fourier layer of the second sub-processing layer of the feature processing layer for global feature capture, until the post-activation feature data output by the activation layer in the last sub-processing layer of the feature processing layer is obtained and serves as the target feature data.

[0041] The CO2-crude oil system minimum miscibility pressure prediction method based on the Fourier neural operator proposed in this embodiment can obtain the standardized mole content percentage of multiple crude oil components in the CO2-crude oil system and the standardized temperature value of the reservoir temperature, and obtain the trained Fourier neural operator FNO model; wherein, the FNO model is obtained by model training based on the sample CO2-crude oil system, the sample reservoir temperature value and the corresponding minimum miscibility pressure, and the FNO model includes an input layer, a first full connection layer, a feature processing layer, a second full connection layer and an output layer connected in turn. The standardized mole content percentage of each crude oil component and the standardized reservoir temperature value are input into the input layer of the FNO model for feature conversion to obtain corresponding low-dimensional feature tensors. The low-dimensional feature tensors are input into the first full connection layer for high-dimensional mapping to obtain corresponding high-dimensional feature tensors. The high-dimensional feature tensors are input into the feature processing layer, the second full connection layer and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system. The FNO model in this embodiment can efficiently capture the multi-parameter global dynamic correlation among the crude oil components, the reservoir temperature and the minimum miscibility pressure in the CO2-crude oil system in the frequency domain, and can effectively improve the calculation efficiency and the prediction accuracy.

[0042] Based on Figure 1 As Figure 4 shown, in other CO2-crude oil system minimum miscibility pressure prediction methods based on the Fourier neural operator proposed in this embodiment, this embodiment can collect and use sample CO2-crude oil systems, sample reservoir temperature values and corresponding minimum miscibility pressures to train the FNO model to be trained to obtain the trained FNO model. Specifically, this embodiment can perform the following processes to train the model: (1) Data collection and preprocessing: collect CO2-crude oil system minimum miscibility pressure measured samples and sample data obtained from other channels under different crude oil components and formation temperature conditions, and clean the collected sample data, eliminate samples with obvious errors and missing key information, and standardize the crude oil component and formation temperature data, so that the crude oil component data or formation temperature data of different orders of magnitude (CO2, C1N2, C2, C3+, C7+, C24+, C36+ and other 7 types of components) are in the same numerical range.

[0043] (2) Sample division: The effective sample data of minimum miscibility pressure of the CO2-crude oil system after pretreatment are divided into training set, validation set and test set according to the ratio of 7:2:1. The training set and validation set are used for FNO model training and validation, respectively, and the test set is used for final evaluation of model prediction performance.

[0044] (3) FNO model architecture construction. For example... Figure 5 As shown, the FNO model includes an input layer, a fully connected layer A, four alternating Fourier layers and activation layers, a fully connected layer B, and an output layer. Figure 5 In the above, fully connected layer A and fully connected layer B are the first fully connected layer and the second fully connected layer, respectively. v ( x () represents the feature data input to the Fourier layer; F This is a Fourier transform operation; R This is a frequency domain feature transformation / modulation module; F -1 This is the inverse Fourier transform operation; W This is a linear transformation matrix; + represents the addition operation. σ This is the LeakyReLU activation function.

[0045] The input layer converts the preprocessed feature data into a tensor form that the model can process. The first fully connected layer maps the low-dimensional features of the input to a higher dimension, providing sufficient feature base for the Fourier layers to support frequency domain transformation and global feature capture. Four Fourier-activation layers alternately contain four Fourier layers, each followed by a LeakyReLU activation layer. The alternating Fourier-activation layers capture global features through Fourier layers and introduce nonlinearity through activation layers, achieving progressive learning of complex global correlations and nonlinear relationships in the data. The second fully connected layer reduces and focuses the high-dimensional features after Fourier layer processing, filters core information, and prepares for mapping MMP predictions to the output layer. The output layer maps the processed features to MMP predictions.

[0046] (4) FNO model training and validation. For example... Figure 6 As shown, the training set is input into the constructed FNO model, the mean squared error loss (MSE) is set, and the model parameters are iteratively updated through the backpropagation algorithm. During the training process, the loss change of the validation set is monitored in real time until the training loss and validation loss reach the optimal. Figure 6 In this context, Epoch represents the training round.

[0047] (5) FNO model prediction: input the test set into the optimized FNO model to obtain the MMP prediction value. The prediction accuracy and stability of the model are evaluated by calculating the mean absolute error (MAE), root mean square error (RMSE), and R-squared (R²) of the prediction value and the measured value.

[0048] (6) FNO model prediction result analysis: import the prediction set of different original sample sets into the trained and debugged FNO model to obtain different prediction results. The prediction analysis results are shown in Figure 7 .

[0049] As shown in Figure 8 , the present embodiment can utilize the trained Fourier neural operator FNO model to obtain MMP data that adapt to the dynamic changes of multiple parameters and reveal the phase state correlation rule of CO2 and crude oil in tight reservoirs, and determine the sensitivity between different parameters and MMP.

[0050] The CO2-crude oil system minimum miscibility pressure prediction method based on the Fourier neural operator proposed in the present embodiment can perform model training through data collection and preprocessing, sample division, FNO model architecture building, and FNO model training and verification, etc., to ensure the reliability of the FNO model, and further ensure the prediction accuracy of the CO2-crude oil system minimum miscibility pressure.

[0051] As shown in Figure 9 , the present embodiment proposes a CO2-crude oil system minimum miscibility pressure prediction device based on the Fourier neural operator, which can include: An acquisition unit 101 is configured to acquire the standardized mole content percentage of a plurality of crude oil components in a CO2-crude oil system and the standardized temperature value of reservoir temperature, and acquire a trained Fourier neural operator FNO model. The FNO model is obtained by model training based on a sample CO2-crude oil system, a sample reservoir temperature value, and a corresponding minimum miscibility pressure. The FNO model includes an input layer, a first full connection layer, a feature processing layer, a second full connection layer, and an output layer connected in sequence. A conversion unit 102 is configured to input the standardized mole content percentage of each crude oil component and the standardized reservoir temperature value into the input layer of the FNO model for feature conversion to obtain a corresponding low-dimensional feature tensor. A mapping unit 103 is configured to input the low-dimensional feature tensor into the first full connection layer for high-dimensional mapping to obtain a corresponding high-dimensional feature tensor. A prediction unit 104 is configured to input the high-dimensional feature tensor into the feature processing layer, the second full connection layer, and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0052] It should be noted that the processing procedures of the acquisition unit 101, the conversion unit 102, the mapping unit 103 and the prediction unit 104 and the beneficial effects brought by the same can be referred to steps S101-S104 in the method 1000 respectively, and will not be repeated here. Figure 1

[0053] Optionally, the acquisition unit 101 is further configured to: acquire the original molar content percentage of each crude oil component in the CO2-crude oil system and the original temperature value of the reservoir temperature; For the original parameter value of any target parameter, acquire the maximum parameter value and the minimum parameter value corresponding to the target parameter, and perform maximum-minimum standardization on the original parameter value of the target parameter according to the maximum parameter value, the minimum parameter value corresponding to the target parameter and the original parameter value of the target parameter, to obtain the standardized parameter value of the target parameter; wherein the target parameter is a crude oil component or a reservoir temperature.

[0054] Optionally, the prediction unit 104 is further configured to: input the high-dimensional feature tensor into the feature processing layer for processing to obtain target feature data; input the target feature data into the second fully connected layer for dimension reduction and focusing to filter out core feature data; input the core feature data into the output layer for feature mapping to obtain the minimum miscibility pressure of the CO2-crude oil system.

[0055] Optionally, the feature processing layer includes N sub-processing layers connected in sequence, N is an integer greater than 1, and each sub-processing layer includes a Fourier layer and an activation layer.

[0056] Optionally, N is 4.

[0057] Optionally, the prediction unit 104 is further configured to: input the high-dimensional feature tensor into the first Fourier layer of the first sub-processing layer of the feature processing layer for global feature capture to obtain global feature data output by the first Fourier layer; input the global feature data output by the first Fourier layer into the first activation layer for local nonlinear activation to obtain corresponding first activated feature data; wherein the first activation layer is the activation layer in the first sub-processing layer of the feature processing layer; input the first activated feature data into the second Fourier layer of the second sub-processing layer of the feature processing layer for global feature capture, until the activated feature data output by the activation layer in the last sub-processing layer of the feature processing layer is obtained and used as target feature data.

[0058] ​Optionally, the plurality of crude oil components in the CO2-crude oil system includes at least two of CO2, C1N2, C2, C3+, C7+, C24+, and C36+.

[0059] The Fourier neural operator-based CO2-crude oil system minimum miscibility pressure prediction device provided in the embodiment can obtain the standardized mole content percentage of the plurality of crude oil components in the CO2-crude oil system and the standardized temperature value of the reservoir temperature, and obtain the trained Fourier neural operator FNO model; wherein the FNO model is obtained by model training based on the sample CO2-crude oil system, the sample reservoir temperature value, and the corresponding minimum miscibility pressure, and the FNO model includes an input layer, a first full connection layer, a feature processing layer, a second full connection layer, and an output layer connected in sequence. The standardized mole content percentage and the standardized reservoir temperature value of each crude oil component are input into the input layer of the FNO model for feature conversion to obtain a corresponding low-dimensional feature tensor. The low-dimensional feature tensor is input into the first full connection layer for high-dimensional mapping to obtain a corresponding high-dimensional feature tensor. The high-dimensional feature tensor is input into the feature processing layer, the second full connection layer, and the output layer for minimum miscibility pressure prediction to obtain the minimum miscibility pressure of the CO2-crude oil system. The FNO model in the embodiment can efficiently capture the multi-parameter global dynamic correlation among the crude oil components, the reservoir temperature, and the minimum miscibility pressure in the CO2-crude oil system in the frequency domain, and can effectively improve the calculation efficiency and the prediction accuracy.

[0060] The Fourier neural operator-based CO2-crude oil system minimum miscibility pressure prediction device in the embodiment is presented in the form of a functional unit. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory executing one or more software or fixed programs, and / or other devices that can provide the above functions.

[0061] The embodiment of the present application also provides a computer device having the Fourier neural operator-based CO2-crude oil system minimum miscibility pressure prediction device shown in the above Figure 9

[0062] Please refer to Figure 10 ​The computer device provided by the optional embodiment of the present application comprises one or more processors 10, a memory 20, and an interface for connecting various components, including a high-speed interface and a low-speed interface. The various components are communicatively connected with each other by different buses, and can be installed on a common mainboard or in other manners as required. The processor can process instructions executed in the computer device, including instructions stored in the memory or on the memory to display graphical information of a GUI on an external input / output device, such as a display device coupled to the interface. In some optional embodiments, multiple processors and / or multiple buses can be used together with multiple memories if required. Similarly, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 10 The processor 10 is taken as an example.

[0063] The processor 10 can be a central processor, a network processor, or a combination thereof. The processor 10 can further comprise a hardware chip. The hardware chip can be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device can be a complex programmable logic device, a field programmable logic gate array, a generic array logic, or any combination thereof.

[0064] The memory 20 stores instructions executable by the at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0065] The memory 20 can comprise a program storage area and a data storage area. The program storage area can store an operating system and application programs required by at least one function. The data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can comprise a high-speed random access memory, and can further comprise a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some optional embodiments, the memory 20 can optionally comprise a memory remotely arranged with respect to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0066] The memory 20 can comprise a volatile memory, such as a random access memory. The memory can also comprise a non-volatile memory, such as a flash memory, a hard disk, or a solid-state disk. The memory 20 can further comprise a combination of the above-mentioned kinds of memories.

[0067] The computer device further comprises a communication interface 30 for communication between the computer device and other devices or communication networks.

[0068] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded through a network and stored in a local storage medium, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. Wherein, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that the computer, processor, microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code, when the software or computer code is accessed and executed by the computer, processor or hardware, the method shown in the above embodiments is implemented.

[0069] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting the minimum miscibility pressure of a CO2-crude oil system based on Fourier neural operators, characterized in that, include: The standardized molar percentages of multiple crude oil components and the standardized temperature values ​​of the reservoir in the CO2-crude oil system are obtained, as well as a trained Fourier neural operator (FNO) model is obtained. The FNO model is trained based on the sample CO2-crude oil system, the sample reservoir temperature values, and the corresponding minimum miscibility pressure. The FNO model includes an input layer, a first fully connected layer, a feature processing layer, a second fully connected layer, and an output layer connected in sequence. The standardized molar percentage of each crude oil component and the standardized reservoir temperature value are input into the input layer of the FNO model for feature transformation to obtain the corresponding low-dimensional feature tensor. The low-dimensional feature tensor is input into the first fully connected layer for high-dimensional mapping to obtain the corresponding high-dimensional feature tensor. The high-dimensional feature tensor is input into the feature processing layer, the second fully connected layer, and the output layer to predict the minimum miscibility pressure, thereby obtaining the minimum miscibility pressure of the CO2-crude oil system.

2. The method according to claim 1, characterized in that, The process of obtaining the standardized molar percentage of multiple crude oil components and the standardized temperature value of the reservoir temperature in the CO2-crude oil system includes: Obtain the original molar percentage of each crude oil component in the CO2-crude oil system and the original temperature value of the reservoir. For any target parameter's original parameter value, obtain the maximum and minimum parameter values ​​corresponding to the target parameter. Based on the maximum and minimum parameter values ​​corresponding to the target parameter and the original parameter value of the target parameter, perform maximum-minimum standardization on the original parameter value of the target parameter to obtain the standardized parameter value of the target parameter; wherein, the target parameter is the crude oil composition or reservoir temperature.

3. The method according to claim 1, characterized in that, The step of inputting the high-dimensional feature tensor into the feature processing layer, the second fully connected layer, and the output layer to predict the minimum miscibility pressure of the CO2-crude oil system includes: The high-dimensional feature tensor is input into the feature processing layer for processing to obtain the target feature data; The target feature data is input into the second fully connected layer for dimensionality reduction and focusing to filter out the core feature data; The core feature data is input into the output layer for feature mapping to obtain the minimum miscibility pressure of the CO2-crude oil system.

4. The method according to claim 3, characterized in that, The feature processing layer includes N sub-processing layers connected in sequence, where N is an integer greater than 1. Each sub-processing layer includes a Fourier layer and an activation layer.

5. The method according to claim 4, characterized in that, N is 4.

6. The method according to claim 5, characterized in that, The step of inputting the high-dimensional feature tensor into the feature processing layer for processing to obtain target feature data includes: The high-dimensional feature tensor is input into the first Fourier layer of the first sub-processing layer of the feature processing layer for global feature capture, and the global feature data output by the first Fourier layer is obtained. The global feature data output from the first Fourier layer is input into the first activation layer for local nonlinear activation to obtain the corresponding first activated feature data; wherein, the first activation layer is the activation layer in the first sub-processing layer of the feature processing layer; The first activated feature data is input into the second Fourier layer of the second sub-processing layer of the feature processing layer for global feature capture until the activated feature data output by the activation layer in the last sub-processing layer of the feature processing layer is obtained, and is used as the target feature data.

7. The method according to any one of claims 1 to 6, characterized in that, The multiple crude oil components in the CO2-crude oil system include at least two of CO2, C1N2, C2, C3+, C7+, C24+, and C36+.

8. A device for predicting the minimum miscibility pressure of a CO2-crude oil system based on Fourier neural operators, characterized in that, include: The acquisition unit is used to acquire the standardized molar percentage of multiple crude oil components and the standardized temperature value of the reservoir in the CO2-crude oil system, as well as to acquire the trained Fourier neural operator (FNO) model; wherein, the FNO model is obtained by training the model based on the sample CO2-crude oil system, the sample reservoir temperature value and the corresponding minimum miscibility pressure, and the FNO model includes an input layer, a first fully connected layer, a feature processing layer, a second fully connected layer and an output layer connected in sequence; The transformation unit is used to input the standardized molar percentage of each crude oil component and the standardized reservoir temperature value into the input layer of the FNO model for feature transformation to obtain the corresponding low-dimensional feature tensor. The mapping unit is used to input the low-dimensional feature tensor into the first fully connected layer for high-dimensional mapping to obtain the corresponding high-dimensional feature tensor. The prediction unit is used to input the high-dimensional feature tensor into the feature processing layer, the second fully connected layer and the output layer to predict the minimum miscibility pressure and obtain the minimum miscibility pressure of the CO2-crude oil system.

9. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the minimum miscibility pressure prediction method for the CO2-crude oil system based on Fourier neural operators as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the method for predicting the minimum miscibility pressure of the CO2-crude oil system based on Fourier neural operators as described in any one of claims 1 to 7.

Images