Heterogeneous reservoir flow-limiting sieve tube simulation method, system, equipment and medium

Abstract

The invention discloses a heterogeneous reservoir flow-limiting sieve tube simulation method, system, equipment and medium, and the method comprises the steps: after reservoir physical property parameters and sieve tube structure parameters are determined, utilizing a computational fluid dynamics model to simulate and analyze the flow-limiting sieve tube structure parameters and corresponding regulation and control indexes; constructing a neural network model of the acid injection process parameters and the regulation and control indexes, and calculating a fluid dynamic model to verify the reliability and optimize the model; and comparing the optimized model with engineering empirical parameters, enriching the training set according to a feedback result, and training again until the final model meets all convergence conditions. The structure parameters of the screen pipe are adaptively matched according to different reservoir conditions, the requirement for automatic design of the flow-limiting screen pipe in engineering practice is met, and the engineering design period and the uncertainty of manual judgment are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of acid injection in horizontal wells, and particularly to a simulation method, system, device and medium for a limited-entry screen pipe in a heterogeneous reservoir. Background Art

[0002] Limited-entry screen pipe completion is a method of acid distribution by lowering a perforated screen pipe to improve the production of low-permeability oil and gas reservoirs. Due to the "heel-toe effect" in the traditional general acid injection method, there is significant non-uniformity in the process of acid injection and reservoir transformation in horizontal wells, and the heterogeneity of the formation in the horizontal well section further increases the difficulty of engineering design. Therefore, the introduction of limited-entry screen pipe tools can effectively balance the acid injection difference between the heel end and the toe end, and at the same time, adaptively transform according to the non-uniform distribution characteristics of formation physical properties, realizing uniform acidification of each horizontal well section of the reservoir on the basis of general acid injection conditions.

[0003] In the prior art, empirical formulas are used for calculation, and its advantage is relatively convenient. However, due to the uncertainty and heterogeneity of reservoir physical property parameters, in order to ensure the reliability of calculation results, the design of the screen pipe needs to be calculated specifically in combination with specific conditions. It is difficult to meet the design accuracy requirements only through general empirical formulas. Therefore, a simulation method for the performance of limited-entry screen pipes that is fast, convenient, has good robustness, and strong generalization ability is needed. Summary of the Invention

[0004] Embodiments of the present invention provide a simulation method, system, device and medium for a limited-entry screen pipe in a heterogeneous reservoir, which solve the technical problem that in the prior art, accurate calculation cannot be performed based on reservoir physical property parameters with uncertainty and heterogeneity, and achieve the technical effects of reducing the empirical requirements for designers to judge on-site working conditions and conditions, reducing the engineering design cycle, and improving the accuracy and reliability of engineering design.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] In the first aspect, the present invention discloses a simulation method for a limited-entry screen pipe in a heterogeneous reservoir, including:

[0007] Based on the physical property parameters of the target reservoir and the screen pipe structure parameters, a computational fluid dynamics model is established; using the above computational fluid dynamics model, the fluid pressure distribution in the screen pipe and the relationship with the apparent flow velocity near the wellbore wall surface are simulated;

[0008] Taking specific boundary conditions as the convergence criterion, through the above computational fluid dynamics model for iterative calculation, a feedback relationship model between the hole structure parameters on the screen pipe and the flow field uniformity in the holes is obtained;

[0009] Based on the above feedback relationship model, taking the rated flow rate of acid injection and reservoir physical property parameters as input parameters, and taking the average apparent flow velocity and pore pressure drop of each section of the above screen pipe as output parameters, training samples are constructed;

[0010] Based on the above training samples, a radial basis function neural network is established, and the prediction results of the above radial basis function neural network are compared with the simulation results of the computational fluid dynamics model;

[0011] If the error exceeds the preset range, continue training; otherwise, compare the prediction results of the above radial basis function neural network with the threshold pressure. If each of the above pore pressures is between the lowest pressure threshold and the upper limit of the acid pump pressure, add the above prediction results to the training sample set to perform reinforcement training on the above radial basis function neural network until the target requirements are met.

[0012] Optionally, the steps of establishing the computational fluid dynamics model described above specifically include:

[0013] Based on the drilling size, horizontal wellbore length, diameter of the limited-entry screen pipe, hole size and distribution data, a computational fluid dynamics model of the near-well and in-pipe flow fields in the screen completion section is constructed;

[0014] Set the material parameters of the computational fluid dynamics model according to the acid type and formation parameters.

[0015] Optionally, the steps of simulating the fluid pressure distribution in the above screen pipe and the relationship with the apparent flow velocity near the wellbore wall specifically include:

[0016] Select the inlet flow rate of the above screen pipe as the inlet boundary, and the formation physical property parameters of each section of the target area as the flow field outlet boundary conditions, and use the above computational fluid dynamics model to calculate the pore fluid pressure distribution and the apparent flow velocity of each section of the reservoir under the completion conditions of different structural limited-entry screen pipes.

[0017] Optionally, the above feedback relationship model includes:

[0018]

[0019]

[0020] Among them, p i and p a are the outlet pressure and inlet pressure of the i-th hole respectively, Q1 is the flow rate of the first section of the screen pipe, l is the segmented length of the limited-entry screen pipe, d is the hole diameter, C d is the orifice flow coefficient without cause, ρ is the density of the fluid passing through the hole, L is the total length of the screen pipe, and f is the Fanning friction factor.

[0021] Optionally, the steps of establishing a radial basis function neural network based on the above training samples specifically include:

[0022] After the above-mentioned radial basis function neural network selects the training parameters, it is trained using the Bayesian regularization algorithm until the convergence condition is met.

[0023] Optionally, the above-mentioned radial basis function neural network adopts a three-layer network structure. The number of neurons in the input layer of the above-mentioned radial basis function neural network is consistent with the conditional parameters. The number of neurons in the output layer of the above-mentioned radial basis function neural network is consistent with the control parameters. The hidden neurons of the above-mentioned radial basis function neural network are determined according to the empirical formula.

[0024] Optionally, the above-mentioned step of constructing the training sample further includes:

[0025] Classify according to the average permeability distribution of the reservoir data, and use the stratified sampling method to extract the above-mentioned reservoir data samples as training samples.

[0026] In a second aspect, the present invention discloses a heterogeneous reservoir limited flow screen simulation system, including:

[0027] A CFD model establishment module, based on the physical property parameters of the target reservoir and the screen structure parameters, establishes a computational fluid dynamics model; uses the above-mentioned computational fluid dynamics model to simulate the fluid pressure distribution in the above-mentioned screen and the relationship with the apparent flow velocity near the wellbore wall surface;

[0028] A feedback module establishment module, based on the above-mentioned computational fluid dynamics model, establishes a feedback relationship model between the pore structure parameters on the above-mentioned screen and the flow field uniformity in the pores under specific boundary conditions;

[0029] A training sample construction module, used to construct training samples on the above-mentioned feedback relationship model, with the rated acid injection flow rate and reservoir physical property parameters as input parameters, and the average apparent flow velocity and pore pressure drop of each section of the above-mentioned screen as output parameters;

[0030] A neural network training module, based on the above-mentioned training samples, establishes a radial basis function neural network, and compares the prediction results of the above-mentioned radial basis function neural network with the simulation results of the computational fluid dynamics model;

[0031] A strengthening module, used to continue training if the error exceeds the preset range; otherwise, compare the prediction results of the above-mentioned radial basis function neural network with the threshold pressure. If the pore pressures are between the lowest pressure threshold and the upper limit of the acid pump pressure, add the above-mentioned prediction results to the training sample set to perform strengthening training on the above-mentioned radial basis function neural network until the target requirements are met.

[0032] In a third aspect, the present invention discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps corresponding to the method in the first aspect are implemented.

[0033] In a fourth aspect, the present invention discloses a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps corresponding to the method in the first aspect are implemented.

[0034] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0035] In the technical solution of the present invention, after determining the reservoir physical property parameters and the screen pipe structure parameters, a computational fluid dynamics model is used to simulate and analyze the restricted flow screen pipe structure parameters and the corresponding control indexes; a neural network model of the acid injection process parameters and the control indexes is constructed, and its reliability is verified and the model is optimized through the computational fluid dynamics model. The training set is enriched and retrained according to the feedback results until the final model meets all convergence conditions. It realizes the self-adaptive matching of the screen pipe structure parameters according to different reservoir conditions, meets the needs of the automated design of the restricted flow screen pipe in engineering practice, and reduces the engineering design cycle and the uncertainty of human judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a flowchart of a method for simulating a restricted flow screen pipe in a heterogeneous reservoir provided by the present invention;

[0038] Figure 2 It is a schematic structural diagram of the screen pipe structure and the spatial relationship with the reservoir provided by the present invention;

[0039] Figure 3 It is a schematic structural diagram of a system for simulating a restricted flow screen pipe in a heterogeneous reservoir provided by the present invention.

[0040] Reference numerals: 1, screen pipe; 2, reservoir space; 3, perforation. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Components of the embodiments of the present invention generally described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0042] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0043] It should be noted that like reference numerals and letters denote like items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0044] It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solutions of this application, rather than limitations on the technical solutions of this application. Without conflict, the technical features in the embodiments of this application and the embodiments can be combined with each other.

[0045] In the embodiments of the present invention, there is provided a method for simulating a heterogeneous reservoir restricted flow screen pipe as shown in Figure 1 and Figure 2 The method includes steps S101 to S105:

[0046] Step S101: Based on the physical property parameters of the target reservoir and the structural parameters of the screen pipe 1, establish a computational fluid dynamics model; use the computational fluid dynamics model to simulate the fluid pressure distribution inside the screen pipe 1 and the relationship with the apparent flow velocity near the wellbore wall surface.

[0047] Among them, the computational fluid dynamics model is mainly the fluid calculation domain composed of the screen pipe 1 and the drilling wellbore. The structure of the screen pipe 1 and the relationship with the reservoir space 2 are as shown in Figure 2As shown. Specifically, based on the drilling size, horizontal wellbore length, diameter of the restricted flow screen pipe 1, size and distribution data of the perforations 3, a computational fluid dynamics model of the near-wellbore and in-pipe flow fields in the completion section of the screen pipe 1 is constructed; the material parameters of the computational fluid dynamics model are set according to the acid type and formation parameters. Thus, the computational fluid dynamics model can predict to a certain extent the near-wellbore and in-pipe flow fields in the completion section of the screen pipe 1, thereby optimizing the design of the screen pipe 1 and improving the reservoir stimulation effect. This effect includes optimizing the reservoir stimulation effect by precisely controlling the fluid flow in the completion section of the screen pipe 1, thereby increasing the oil and gas production capacity. By precisely controlling the fluid flow in the completion section of the screen pipe 1, the reservoir stimulation effect is optimized, thereby increasing the oil and gas production capacity. By reducing unnecessary drilling and completion engineering workload, the cost is reduced. By precisely controlling the fluid flow in the completion section of the screen pipe 1, damage to the formation is reduced and the formation is protected.

[0048] The specific simulation of the fluid pressure distribution in the screen pipe 1 and the relationship with the apparent flow velocity near the wellbore wall is as follows: The inlet flow rate of the screen pipe 1 is selected as the inlet boundary, and the formation physical property parameters of each section in the target area are used as the flow field outlet boundary conditions. The computational fluid dynamics model is used to calculate the fluid pressure distribution of the perforations 3 and the apparent flow velocity of each section of the reservoir under the completion conditions of different structured restricted flow screen pipes 1.

[0049] Specifically, the Fluent solver is used for the solution calculation. The wellhead of the screen pipe 1 is the flow rate inlet boundary, and the annulus is set as the pressure outlet. Steady-state calculation is adopted to improve the calculation efficiency and the relaxation factor is appropriately adjusted. The turbulence model is the standard k-ε turbulence model. The acid concentration is set to 15%, the acid density is 1.15 g / cm3, the acid viscosity is taken as 0.7 mPa·s, and the pressure drop of each perforation 3 is monitored.

[0050] Step S102: Taking specific boundary conditions as the convergence criterion, perform cyclic iterative calculations through the computational fluid dynamics model to obtain a feedback relationship model between the structural parameters of the perforations 3 on the screen pipe 1 and the uniformity of the flow field in the perforations 3.

[0051] It should be noted that the purpose of the cyclic calculation in this step is to obtain a feedback relationship model between the structural parameters of the perforations 3 on the screen pipe 1 and the uniformity of the flow field in the perforations 3 to optimize the performance of the screen pipe 1. By taking specific boundary conditions as the convergence criterion, the structural parameters of the perforations 3 on the screen pipe 1 are optimized to improve the uniformity of the flow field in the perforations 3. As an important index of the performance of the screen pipe 1, the flow field uniformity directly affects the separation effect and efficiency of the screen pipe 1. Therefore, this embodiment provides theoretical support and practical guidance for the optimization design of the screen pipe 1. Specifically, through cyclic iterative calculations, the structural parameters of the perforations 3 on the screen pipe 1 can be gradually adjusted to obtain the best flow field uniformity. This optimization design method can significantly improve the separation effect and efficiency of the screen pipe 1 and can reduce energy consumption and costs.

[0052] Specifically, in this embodiment, the inhomogeneity evaluation index includes the number of sections N and the apparent flow velocity v, and the optimization objective is the standard deviation S of the apparent flow velocity after reservoir stimulation in each section v and the pressure difference threshold ΔP at the outlet end of the perforation 3 of the screen pipe 1, where:

[0053]

[0054] In the formula, ε is the reservoir porosity and n is the number of sections.

[0055] The set optimization objective is:

[0056]

[0057] The specific boundary conditions are:

[0058]

[0059] After establishing the geometric model of the restricted flow screen pipe 1, in order to achieve uniform acid distribution in the horizontal section, the modeling of the perforation 3 arrangement of the restricted flow screen pipe 1 in each reservoir section is as follows:

[0060]

[0061]

[0062] where p i and p a are the outlet pressure and inlet pressure of the i-th perforation 3 respectively, Q1 is the flow rate of the first section of the screen pipe 1, l is the segmented length of the restricted flow screen pipe 1, d is the diameter of the perforation 3, C d is the flow coefficient of the perforation 3 without cause, ρ is the fluid density passing through the perforation 3, L is the total length of the screen pipe 1, and f is the Fanning friction factor.

[0063] In step S103, on the feedback relationship model, with the rated acid injection flow rate and reservoir physical property parameters as input parameters and the average apparent flow velocity of each section of the screen pipe 1 and the pressure drop of the perforation 3 as output parameters, a training sample is constructed. Among them, for the construction of the training sample, it is mainly classified according to the average permeability distribution of the reservoir data, and the reservoir data samples are extracted as training samples by the stratified sampling method. The advantage of stratified sampling is that it can extract samples targeted according to the characteristics of different layers, thereby improving the representativeness of the samples and making the samples closer to the overall population; at the same time, stratified sampling can divide the overall population into several layers and conduct random sampling for each layer, thereby obtaining a more accurate estimated value.

[0064] Step S104: Establish a radial basis function neural network based on training samples, and compare the prediction results of the radial basis function neural network with the simulation results of the computational fluid dynamics model. Specifically, when establishing the radial basis function neural network, after selecting the training parameters for the radial basis function neural network, the Bayesian regularization algorithm is used for training until the convergence condition is met. The radial basis function neural network adopts a three-layer network structure. The number of neurons in the input layer of the radial basis function neural network is the same as the conditional parameters, the number of neurons in the output layer of the radial basis function neural network is the same as the control parameters, and the hidden neurons of the radial basis function neural network are determined according to the empirical formula.

[0065] In addition, Computational Fluid Dynamics (CFD) is used to analyze the selective acidification distribution under the completion conditions of the specific structure restricted flow screen pipe 1, aiming to evaluate the production pressure difference balance of different sections of the reservoir after transformation. However, due to the complexity of the CFD method and the long calculation cycle, the cost of multi-round iterative optimization is relatively high, and the final optimization effect also depends more on the design experience of engineers. Therefore, it is particularly necessary to combine CFD with machine learning to develop a performance simulation method for the restricted flow screen pipe 1 that is fast, convenient, has good robustness, and strong generalization ability. Therefore, the prediction results of the radial basis function neural network are compared with the simulation results of the computational fluid dynamics model.

[0066] Step S105: If the error exceeds the preset range, continue training; otherwise, compare the prediction results of the radial basis function neural network with the threshold pressure. If the pressure of each hole 3 is between the minimum pressure threshold and the upper limit of the acid pump pressure, add the prediction results to the training sample set to strengthen the training of the radial basis function neural network until the target requirements are met.

[0067] Specifically, use the trained radial basis function neural network model to replace the computational fluid dynamics model to simulate the working characteristics of the restricted flow screen pipe 1 under random reservoir conditions. Test the radial basis function neural network model through the test set, input the reservoir physical property parameters and the pumping acid flow rate, calculate whether the pressure drop of the holes 3 of the screen pipe 1 meets the condition of being greater than 0.15 MPa, and at the same time calculate whether the apparent flow velocity fluctuation in different reservoir sections is within the range required by the experiment. When the optimization goal of the test results is achieved, the structure of the screen pipe 1 is the reasonable structure predicted by the model, and the computational fluid dynamics method is used to test and analyze the reliability of the prediction results.

[0068] It should be noted that the acid pump refers to the acid pump used in the restricted flow screen pipe 1. It is used to drive the acid liquid to pass through the restricted flow screen pipe 1 for filtration and separation. The restricted flow screen pipe 1 usually consists of multiple holes 3, and these holes 3 can control the flow of the fluid. The acid pump sucks the acid liquid from one side and then presses it through the holes 3 of the restricted flow screen pipe 1 to the other side, thereby realizing filtration and separation.

[0069] Based on the same inventive concept, an embodiment of the present invention provides a simulation system for a heterogeneous reservoir limited flow screen, as Figure 3 shown, including:

[0070] A CFD model building module, which builds a computational fluid dynamics model based on the physical property parameters of the target reservoir and the structural parameters of the screen 1; and uses the computational fluid dynamics model to simulate the fluid pressure distribution inside the screen 1 and the relationship with the apparent flow velocity near the wellbore wall surface;

[0071] A feedback module building module, which builds a feedback relationship model between the structural parameters of the holes 3 on the screen 1 and the flow field uniformity inside the holes 3 under specific boundary conditions based on the computational fluid dynamics model;

[0072] A training sample construction module, which is used to construct training samples on the feedback relationship model, with the rated acid injection flow rate and the reservoir physical property parameters as input parameters, and the average apparent flow velocity of each section of the screen 1 and the pressure drop of the holes 3 as output parameters;

[0073] A neural network training module, which builds a radial basis function neural network based on the training samples, and compares the prediction results of the radial basis function neural network with the simulation results of the computational fluid dynamics model;

[0074] A strengthening module, which is used to continue training if the error exceeds the preset range; otherwise, compare the prediction results of the radial basis function neural network with the threshold pressure. If the pressure of each hole 3 is between the minimum pressure threshold and the upper limit of the acid pump pressure, add the prediction results to the training sample set to strengthen the training of the radial basis function neural network until the target requirements are met.

[0075] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the simulation method for a heterogeneous reservoir limited flow screen.

[0076] Based on the same inventive concept, this embodiment provides a computer-readable storage medium, on which a computer program is stored. The program is characterized in that when it is executed by a processor, it implements the simulation method for a heterogeneous reservoir limited flow screen.

[0077] The present invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each flow and / or block in the flowchart illustrations and / or block diagrams, and combinations of flows and / or blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing apparatus create means for implementing the functions specified in the flow Figure 1 one or more flows and / or blocks Figure 1 or means for implementing the functions specified in one or more blocks.

[0078] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0079] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A simulation method for a heterogeneous reservoir restricted flow screen pipe, characterized in that, The method includes: Based on the physical property parameters of the target reservoir and the screen pipe structure parameters, a computational fluid dynamics model is established; using the computational fluid dynamics model, the fluid pressure distribution inside the screen pipe and the relationship with the apparent flow velocity near the wellbore wall surface are simulated; Taking specific boundary conditions as the convergence criterion, through the computational fluid dynamics model for iterative calculation, a feedback relationship model between the hole structure parameters on the screen pipe and the flow field uniformity inside the holes is obtained; On the feedback relationship model, taking the rated acid injection flow rate and reservoir physical property parameters as input parameters, and taking the average apparent flow velocity of each section of the screen pipe and the hole pressure drop as output parameters, a training sample is constructed; Based on the training sample, a radial basis function neural network is established, and the prediction result of the radial basis function neural network is compared with the simulation result of the computational fluid dynamics model; If the error exceeds the preset range, continue training; otherwise, compare the prediction result of the radial basis function neural network with the threshold pressure. If the pressure of each hole is between the minimum pressure threshold and the upper limit of the acid pump pressure, add the prediction result to the training sample set to strengthen the training of the radial basis function neural network until the target requirements are met.

2. The method according to claim 1, wherein The steps of establishing the computational fluid dynamics model specifically include: Based on the drilling size, horizontal wellbore length, limited flow screen pipe diameter, hole size and distribution data, a computational fluid dynamics model of the near-well and in-pipe flow fields in the screen pipe completion section is constructed; Set the material parameters of the computational fluid dynamics model according to the acid type and formation parameters.

3. The method according to claim 1, characterized in that The steps of simulating the fluid pressure distribution inside the screen pipe and the relationship with the apparent flow velocity near the wellbore wall surface specifically include: Select the inlet flow rate of the screen pipe as the inlet boundary, and the formation physical property parameters of each section of the target area as the flow field outlet boundary conditions, and use the computational fluid dynamics model to calculate the hole fluid pressure distribution and the apparent flow velocity of each section of the reservoir under the completion conditions of different structure limited flow screen pipes.

4. The method according to claim 3, wherein The feedback relationship model includes: where p i and p a are the outlet pressure and the inlet pressure of the i-th perforation respectively, Q1 is the flow rate of the first-stage screen pipe, l is the sectional length of the restricted-flow screen pipe, d is the perforation diameter, C d is the flow coefficient of the perforation without considering other factors, ρ is the density of the fluid flowing through the perforation, L is the total length of the screen pipe, and f is the Fanning friction factor.

5. The method according to claim 1, characterized in that, The steps of establishing the radial basis function neural network based on the training sample specifically include: After the radial basis function neural network selects the training parameters, it is trained using the Bayesian regularization algorithm until the convergence condition is met.

6. The method according to claim 5, wherein The radial basis function neural network adopts a three-layer network structure. The number of neurons in the input layer of the radial basis function neural network is the same as the condition parameters, the number of neurons in the output layer of the radial basis function neural network is the same as the control parameters, and the hidden neurons of the radial basis function neural network are determined according to the empirical formula.

7. The method according to any one of claims 1 to 6, characterized in that The steps of constructing the training sample further include: Classify according to the average permeability distribution of the reservoir data, and use the stratified sampling method to extract the reservoir data samples as training samples.

8. A heterogeneous reservoir restricted flow screen simulation system, characterized in that, The system includes: A CFD model establishment module, which based on the physical property parameters of the target reservoir and the screen pipe structure parameters, establishes a computational fluid dynamics model; using the computational fluid dynamics model, simulates the fluid pressure distribution inside the screen pipe and the relationship with the apparent flow velocity near the wellbore wall surface; A feedback module establishment module, based on the computational fluid dynamics model, establishes a feedback relationship model between the hole structure parameters on the screen pipe and the flow field uniformity in the holes under specific boundary conditions; A training sample construction module is used to construct training samples on the feedback relationship model, with the rated flow rate of acid injection and reservoir physical property parameters as input parameters and the average apparent flow velocity and hole pressure drop of each section of the screen pipe as output parameters; A neural network training module establishes a radial basis function neural network based on the training samples, and compares the prediction results of the radial basis function neural network with the simulation results of the computational fluid dynamics model; A reinforcement module is used to continue training if the error exceeds the preset range; otherwise, the prediction results of the radial basis function neural network are compared with the threshold pressure. If the pressures of the holes are between the minimum pressure threshold and the upper limit of the acid pump pressure, the prediction results are added to the training sample set to perform reinforcement training on the radial basis function neural network until the target requirements are met.

9. An electronic device, characterized in that, The electronic device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method steps described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the steps corresponding to the method described in any one of claims 1 to 7 are implemented.

Images