A method, device, equipment and medium for optimizing injection and production parameters of oil reservoir gas drive

Abstract

The present application discloses a method, device, equipment and medium for optimizing gas-driven injection and production parameters in oil reservoirs, and relates to the field of gas-driven oil reservoirs. The method includes using a component numerical simulation model to determine the future oilfield production dynamics during each iterative optimization process using a target optimization algorithm, and using a gas-driven injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of each injection and production parameter combination scheme based on the future oilfield production dynamics; using the injection and production parameter combination schemes of all iterative optimizations performed as input parameters, and the optimization target values ​​obtained from all iterative optimizations performed as output parameters, establishing a sample set, continuously iteratively training a Transformer‑LSTM fusion model, and optimizing the injection and production parameters based on the trained Transformer‑LSTM fusion model. The present application can quickly, cost-effectively and accurately determine the injection and production parameters of each well in a gas-driven oil reservoir.

Description

Technical Field

[0001] The present application relates to the field of gas-driven oil reservoirs, and in particular to a method, device, equipment and medium for optimizing injection and production parameters of gas-driven oil reservoirs. Background Art

[0002] Gas injection to improve oil recovery has been widely used, especially in low permeability, tight oil and high water content reservoirs. How to design the injection and production parameters of gas drive reservoirs has a great impact on the recovery rate of the reservoir. However, the traditional determination of injection and production parameters of gas drive reservoirs is generally based on the design scheme of component numerical simulators. Due to the slow calculation speed of the actual reservoir component numerical simulation, it takes a lot of time to design the scheme each time, which makes it difficult to quickly guide the problems faced in the actual development of the field. At present, there are mainly the following methods for accelerating the optimization of gas drive injection and production parameters in oil reservoirs: (1) Improved intelligent optimization algorithm: Improve the current more mature intelligent optimization algorithm to improve the efficiency of gas drive injection and production parameter optimization in oil reservoirs. It requires strong computer programming skills. This method is more difficult for non-computer professionals. (2) Coarsening model: Coarsening the grid of the numerical simulation model in the plane or vertical direction to reduce the overall effective number of grids in the numerical model. This will speed up the calculation speed when using numerical simulation to calculate the objective function value, thereby improving the optimization efficiency. However, the accuracy of the roughened model will decrease when calculating the objective function value, which will lead to a decrease in the accuracy of the overall injection and production parameter optimization. (3) Analytical or semi-analytical calculation model: When using analytical or semi-analytical methods to calculate the objective function value, it is possible to avoid calculating large matrices, so the calculation speed will be faster, thereby improving the efficiency in the optimization process. However, the applicable conditions for using analytical or semi-analytical methods are limited, and they cannot be applied in all calculation situations. First, the influencing factors and calculation equations that need to be considered when calculating the development dynamics of actual reservoirs are very complex, and the analytical method cannot consider the relationship between the production dynamics and the time series during the reservoir development process. Secondly, the analytical method cannot consider the relationship between natural fractures and reservoir seepage mechanism in the process of calculating the objective function value. It can only calculate simple steady-state single-phase Darcy seepage problems, and the calculation error is large in actual reservoir seepage. (4) Large-scale computer cluster calculation: The time to calculate the objective function value based on the numerical simulator is closely related to the performance of the computer CPU (Central Processing Unit). Therefore, by using large-scale computer clusters to carry out numerical simulation calculations, the calculation time of each numerical simulation can be reduced to improve the overall optimization efficiency, but this will directly lead to an increase in calculation costs. (5) Establishing proxy model calculation: The core of this method is to first design a large number of injection and production schemes, use numerical simulators to calculate development dynamics, build a sample set, and use mathematical methods such as Kriging interpolation or deep learning methods to fit the mathematical relationship or deep learning model between the injection and production parameters and the objective function value. The obtained relationship or deep learning model is then used to replace the numerical simulation calculation process to predict the objective function value.Although this method can greatly improve the efficiency of optimization, it takes a lot of time to establish the sample set. In addition, the migration of the proxy model is poor for oil reservoirs in different reservoirs. The accuracy of the optimization results depends on the accuracy of the constructed proxy model.

[0003] It can be seen that the existing technology cannot quickly and cost-effectively determine the injection and production parameters of each well in the gas-driven oil reservoir, and it is also difficult to ensure accuracy. Summary of the invention

[0004] The purpose of the present application is to provide a method, device, equipment and medium for optimizing injection and production parameters of gas drive oil reservoirs, which can quickly, cost-effectively and accurately determine the injection and production parameters of each well in the gas drive oil reservoir.

[0005] To achieve the above objectives, this application provides the following solutions.

[0006] In a first aspect, the present application provides a method for optimizing gas-drive injection and production parameters in an oil reservoir, comprising: establishing a component numerical simulation model of a target gas-drive oil reservoir; constructing a gas-drive injection and production parameter collaborative optimization mathematical model; initializing the number of iterations of the target optimization algorithm, and setting e=1; e is a positive integer; using the target optimization algorithm to perform the e-th iteration optimization, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iteration optimization process, and using the gas-drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme based on the future oilfield production dynamics; if the value of e is less than the number of iterations, using all the injection and production parameter combination schemes that have been iteratively optimized as input parameters, and using all the optimization target values ​​obtained from the iterative optimization as output parameters, establishing a sample set, and using the sample set to train a Transformer-LSTM (Long Short-Term Memory, long short-term memory network) fusion model; determine whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions; if so, input the e-th injection and production parameter combination scheme into the trained Transformer-LSTM fusion model, output the predicted optimization target value, and replace the predicted optimization target value of the e-th injection and production parameter combination scheme with the predicted optimization target value, and increase the value of e by 1, and return to the step of "using the target optimization algorithm to perform the e-th iterative optimization, and use the component numerical simulation model to determine the future oilfield production dynamics in the e-th iterative optimization process, and according to the future oilfield production dynamics, using the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme"; if not satisfied, increase the value of e by 1, and return to the step of "using the target optimization algorithm to perform the e-th iterative optimization, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iterative optimization process, and according to the future oilfield production dynamics, using the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme"; if the value of e is equal to the number of iterations, the e-th injection and production parameter combination scheme is output as the optimal injection and production parameter combination scheme for the target gas drive reservoir.

[0007] Optionally, a mathematical model for collaborative optimization of gas drive injection and production parameters is constructed, specifically including: determining an optimization target; the optimization target is economic net present value, cumulative oil production, carbon dioxide storage capacity or recovery rate; determining an optimization objective function according to the optimization target; setting injection and production parameter optimization variables and constraints; and combining the optimization objective function, injection and production parameter optimization variables and constraints to form a mathematical model for collaborative optimization of gas drive injection and production parameters.

[0008] Optionally, when the optimization target is the economic net present value, an optimization objective function is established with the maximum economic net present value as the target; wherein the calculation formula of the economic net present value is:

[0009] ;

[0010] In the formula, is the economic net present value, , , Respectively The production well Annual oil production, gas production, and water production, For the The injection well is located in The total annual water injection volume, For the The gas injection well The total gas injection volume per year, is the selling price of gas, is the selling price of oil, is the water injection cost, is the cost of gas injection, is the cost of producing water treatment, is the discount rate, is the number of producing wells, is the number of injection wells, is the number of gas injection wells, Evaluation time.

[0011] Optionally, when the optimization target is the carbon dioxide storage capacity, an optimization objective function is established with the maximum carbon dioxide storage capacity as the target; wherein the calculation formula for the carbon dioxide storage capacity is: ; In the formula, is the carbon dioxide storage capacity, is the total amount of injected carbon dioxide, is the total amount of carbon dioxide produced.

[0012] Optionally, when the gas injection method of the target gas-drive oil reservoir is continuous gas injection, the injection and production parameter optimization variables of the injection well include gas injection timing, gas injection rate, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well include liquid production rate or bottom hole flow pressure; when the gas injection method of the target gas-drive oil reservoir is intermittent gas injection, the injection and production parameter optimization variables of the injection well are gas injection timing, gas injection rate, gas injection cycle, well shut-down time, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well are liquid production rate or bottom hole flow pressure; when the gas injection method of the target gas-drive oil reservoir is gas-water alternating gas injection, the injection and production parameter optimization variables of the injection well are gas injection timing, gas injection rate, gas injection cycle, gas-water ratio, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well are liquid production rate or bottom hole flow pressure of the production well.

[0013] Optionally, the constraints include: boundary constraints and equation constraints; boundary constraints include injection rate constraints of each gas injection well, injection timing constraints of each gas injection well, injection cycle constraints of each gas injection well, well shutoff time constraints of each gas injection well, injection rate constraints of each water injection well, liquid production rate constraints of each production well, gas-water ratio constraints, bottom hole flow pressure constraints of gas injection wells and bottom hole flow pressure constraints of production wells; equation constraints include total gas injection volume constraints, total water injection volume constraints and total liquid production volume constraints.

[0014] Optionally, the Transformer-LSTM fusion model includes: a Transformer layer, a first LSTM layer, a second LSTM layer and a fully connected layer connected in sequence.

[0015] In the second aspect, the present application provides an oil reservoir gas drive injection and production parameter optimization device, including: a simulation model building module, an optimization model construction module, an initialization module, an iterative optimization module, a training module, a judgment module, a replacement module, a return module and an output module.

[0016] A simulation model building module is used to build a component numerical simulation model of the target gas drive oil reservoir; an optimization model building module is used to build a gas drive injection and production parameter collaborative optimization mathematical model; an initialization module is used to initialize the number of iterations of the target optimization algorithm, and set e=1; e is a positive integer; an iterative optimization module is used to use the target optimization algorithm to perform the e-th iterative optimization, use the component numerical simulation model to determine the future oil field production dynamics in the e-th iterative optimization process, and use the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme based on the future oil field production dynamics; a training module is used to use the injection and production parameter combination schemes of all iterative optimizations that have been performed as input parameters, and the optimization target values ​​obtained from all iterative optimizations that have been performed as output parameters, if the value of e is less than the number of iterations. A sample set is established, and the Transformer-LSTM fusion model is trained by using the sample set; a judgment module is used to judge whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions; a replacement module is used to input the e-th injection-production parameter combination scheme into the trained Transformer-LSTM fusion model if the conditions are met, output the predicted optimization target value, and replace the predicted optimization target value with the optimization target value of the e-th injection-production parameter combination scheme, and increase the value of e by 1, and call the iterative optimization module; a return module is used to increase the value of e by 1, and call the iterative optimization module if the conditions are not met; an output module is used to output the e-th injection-production parameter combination scheme as the optimal injection-production parameter combination scheme for the target gas drive reservoir if the value of e is equal to the number of iterations.

[0017] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for optimizing injection and production parameters of oil reservoir gas drive.

[0018] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for optimizing injection and production parameters of oil reservoir gas drive.

[0019] According to the specific embodiments provided in this application, this application has the following technical effects.

[0020] The present application provides a method, device, equipment and medium for optimizing injection and production parameters of gas drive in oil reservoirs. It is not necessary to construct a large number of sample sets in advance. Instead, the numerical simulation scheme in the iterative optimization process is used as the sample set, and the Transformer-LSTM fusion model is continuously iterated and trained, which saves a lot of time for constructing sample sets; and the Transformer-LSTM fusion model that meets the preset conditions after training is used to calculate the optimization target value, and the injection and production parameter optimization is carried out. The fast calculation speed of the deep learning model is used to optimize the optimal solution, which effectively shortens the time spent on the optimization of the target optimization algorithm; at the same time, the Transformer in the Transformer-LSTM fusion model can effectively process long sequence data in parallel, and LSTM is good at capturing short-term and long-term time dependencies. Combining the advantages of the two, it is possible to capture time characteristics at different levels and provide more comprehensive time series modeling capabilities, thereby realizing fast, low-cost and accurate determination of the injection and production parameters of each well in the gas drive oil reservoir. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A schematic flow chart of a method for optimizing reservoir gas drive injection and production parameters provided in one embodiment of the present application.

[0023] Figure 2 A more detailed flow chart of a method for optimizing injection and production parameters of oil reservoir gas drive provided in one embodiment of the present application.

[0024] Figure 3 A schematic diagram of the Transformer-LSTM fusion model architecture provided for another embodiment of the present application.

[0025] Figure 4 Schematic diagram of the permeability distribution of H reservoir provided for the example application of this application.

[0026] Figure 5 Schematic diagram of H reservoir pressure distribution provided for the example application of this application.

[0027] Figure 6 Schematic diagram of the current oil saturation and well location distribution of the H reservoir provided for the example application of this application.

[0028] Figure 7 Schematic diagram of the Pareto front of MOPSO (Multi-Objective Particle Swarm Optimization)-Transformer-LSTM collaborative optimization provided for the example application of this application.

[0029] Figure 8 Schematic diagram of the Transformer-LSTM model evolution process provided for the example application of this application.

[0030] Fig. 9 Schematic diagram of the optimization results of the working system of the gas injection well provided for the example application of this application.

[0031] Fig.10 Schematic diagram of the optimization results of the production well working system provided for the example application of this application.

[0032] Fig.11 Schematic diagram of the optimization results of the water injection well working system provided for the example application of this application.

[0033] Fig.12 Schematic diagram of oil saturation distribution of the optimized H reservoir provided for the example application of this application.

[0034] Fig.13 Schematic diagram of oil saturation distribution of the H reservoir basic scheme provided for the example application of this application.

[0035] Fig.14 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0038] In an exemplary embodiment, Figure 1 As shown, a method for optimizing reservoir gas drive injection and production parameters is provided, including the following steps 101 to 109.

[0039] Step 101: Establish a component numerical simulation model of a target gas drive reservoir.

[0040] Step 102: Construct a mathematical model for collaborative optimization of gas drive injection and production parameters.

[0041] Step 103: Initialize the number of iterations of the target optimization algorithm, and set e=1; e is a positive integer.

[0042] Step 104: Use the target optimization algorithm to perform the e-th iterative optimization. During the e-th iterative optimization process, use the component numerical simulation model to determine the future oilfield production dynamics, and based on the future oilfield production dynamics, use the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme.

[0043] Step 105: If the value of e is less than the number of iterations, the injection-production parameter combination schemes optimized by all the iterative optimizations are used as input parameters, and the optimization target values ​​obtained by all the iterative optimizations are used as output parameters, a sample set is established, and the Transformer-LSTM fusion model is trained using the sample set.

[0044] Step 106: Determine whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions.

[0045] Step 107: If satisfied, the e-th injection-production parameter combination scheme is input into the trained Transformer-LSTM fusion model, the predicted optimization target value is output, and the predicted optimization target value replaces the optimization target value of the e-th injection-production parameter combination scheme, and the value of e is increased by 1, and the step of "using the target optimization algorithm to perform the e-th iterative optimization, using the component numerical simulation model to determine the future oil field production dynamics during the e-th iterative optimization process, and according to the future oil field production dynamics, using the gas drive injection-production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection-production parameter combination scheme" is returned.

[0046] Step 108: If not satisfied, increase the value of e by 1, and return to step "using the target optimization algorithm to perform the e-th iterative optimization, using the component numerical simulation model to determine the future oil field production dynamics during the e-th iterative optimization process, and according to the future oil field production dynamics, using the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme."

[0047] Step 109: If the value of e is equal to the number of iterations, the e-th injection-production parameter combination scheme is output as the optimal injection-production parameter combination scheme for the target gas drive reservoir.

[0048] Existing methods for optimizing gas injection and production parameters in oil reservoirs are generally based on optimization theory and use component simulators to directly optimize injection and production parameters. This works well for smaller models or conceptual models, but it takes a lot of time to optimize larger actual models, and multiple optimizations and real-time optimization are difficult. The second method is to use a proxy model method. This method replaces the numerical simulation process by establishing a proxy model to accelerate optimization, but how to accurately predict the model is the key. The premise for building a proxy model is to design a large number of gas injection and production methods through Latin hypercube and construct a sample set through numerical simulation. The construction of the initial sample data set requires a lot of computing time, and a proxy model needs to be established for a specific reservoir. The proxy model has poor mobility for complex actual reservoirs and cannot guarantee the accuracy of the proxy model. By implementing the above steps 101 to 109, a Transformer-LSTM fusion model is first constructed, and then a gas drive injection and production parameter acceleration optimization method based on a deep learning technology-assisted intelligent optimization algorithm is constructed. The advantage of the fast calculation speed of the deep learning prediction model is used to accelerate the optimization process, effectively shortening the time spent on optimizing the optimization algorithm, and achieving simple, fast and accurate determination of gas drive injection and production parameters, significantly improving the optimization time of gas drive injection and production parameters, reducing related economic costs, and providing fast and scientific technical support for on-site development decisions.

[0049] In another exemplary embodiment of the present application, the above-mentioned step 101 organizes and optimizes the geological model of the target gas-drive reservoir and the historical working system data of the injection and production wells, uses reservoir numerical simulation software to establish a component numerical simulation model dat file, and carries out reservoir history fitting, so that the production dynamic fitting accuracy such as cumulative oil production, cumulative water production and water content meets the prediction requirements, and obtains the component numerical simulation model of the target gas-drive reservoir.

[0050] In another exemplary embodiment of the present application, the above step 102 mainly includes the following steps.

[0051] ① Determine the optimization objective function , specifically including: the optimization target can be determined according to actual needs, which can be economic net present value (NPV), cumulative oil production (FOPT), CO 2 Buried stock ( ) or recovery factor (Field Enhance Oil Recovery, FOE), etc., are calculated as shown in Equation (1) to Equation (3). FOPT and FOE can be calculated by a numerical simulator, and the specific optimization target can be selected according to actual needs.

[0052] (NPV, FOPT, C s , FOE) (1)

[0053] (2)

[0054] (3)

[0055] In the formula, is the economic net present value, , , Respectively The production well Annual oil production, gas production, water production, m 3 ·d -1 ; For the The injection well is located in Total water injection volume per year, m 3 ·d -1 ; For the The gas injection well Total gas injection volume per year, m 3 ·d -1 ; is the sales price of gas, RMB·m -3 ; is the sales price of oil, yuan·m -3 ; is the water injection cost, yuan·m -3 ; is the cost of gas injection, yuan·m -3 ; is the treatment cost of produced water, RMB·m -3 ; is the discount rate, %; is the number of producing wells, wells; is the number of water injection wells, ; is the number of gas injection wells, ; The evaluation time is year. is the carbon dioxide storage capacity, is the total amount of injected carbon dioxide, is the total amount of carbon dioxide produced.

[0056] ② The gas injection method can be continuous gas injection, gas-water alternation and intermittent gas injection. Different optimization variables correspond to different optimization gas injection methods. For the continuous gas injection method, the optimization variables of the injection well include the injection timing, gas injection rate, total gas injection volume per well and water injection rate. The optimization variables of the production well include the liquid production rate or bottom hole pressure. If the production well produces at a fixed liquid volume, the optimization variable is formula (4).

[0057] (4)

[0058] In the formula, Optimize variables for injection and production parameters, , , They are the first gas injection well, Gas injection well, No. Gas injection timing of gas injection wells, days; , , They are the first gas injection well, Gas injection well, No. Gas injection rate of gas injection wells, m 3 ·d -1 ; , , They are the first gas injection well, Gas injection well, No. Total gas injection volume of a single gas injection well, m 3 ; , , They are the first injection well, water injection well, No. The injection rate of the injection well, , , They are the first production well, Production well, Liquid production rate of production wells, m 3 ·d -1 .

[0059] For the intermittent gas injection method, the optimization variables of the injection well are the injection timing, injection rate, injection cycle, well shut-down time, total gas injection volume per well and water injection rate. The optimization variables of the production well are the liquid production rate or bottom hole pressure. The production well produces at a fixed liquid volume, and the optimization variable is formula (5).

[0060] (5)

[0061] In the formula, , , They are the first gas injection well, Gas injection well, No. Gas injection cycle of gas injection wells, ; , , They are the first gas injection well, Gas injection well, No. The sealing time of the gas injection well is days.

[0062] For the gas-water alternating injection method, the optimization variables of the injection well are the injection timing, injection rate, injection cycle, gas-water ratio, total gas injection volume per well, and water injection rate. The optimization variables of the production well are the liquid production rate or the bottom flow pressure of the production well. The production well produces at a fixed liquid volume, and the optimization variable is formula (6).

[0063] (6)

[0064] In the formula, , , They are the first gas injection well, Gas injection well, No. Gas-water ratio of gas injection wells.

[0065] ③ The optimization problem of gas drive injection and production is mainly divided into boundary constraints and equation constraints. The boundary constraints include the injection rate constraints of each gas injection well, the injection timing constraints of each gas injection well, the injection cycle constraints of each gas injection well, the well shut-down time constraints of each gas injection well, the injection rate constraints of each water injection well, the liquid production rate constraints of each production well, the gas-water ratio constraints, the bottom flow pressure constraints of the gas injection well, and the bottom flow pressure constraints of the production well. The boundary constraints are shown in formula (7). The equation constraints include the total gas injection amount constraints, the total water injection amount constraints, and the total liquid production amount constraints. The equation constraints are shown in formula (8).

[0066] (7)

[0067] (8)

[0068] In the formula, , They are the minimum gas injection timing and the maximum gas injection timing of the gas injection well respectively; , are the minimum injection rate and maximum injection rate of the gas injection well, respectively; , are the minimum injection rate and maximum injection rate of the water injection well respectively; , are the minimum and maximum liquid production rates of the production wells, respectively; , are the minimum injection period and maximum injection period of the gas injection well respectively; , They are the minimum well blocking time and the maximum well blocking time of the gas injection well respectively; , are the minimum gas-water ratio and the maximum gas-water ratio respectively; P Pi For the i Bottom hole flowing pressure of production well, MPa; , are the minimum and maximum bottom hole flowing pressures of the production well, MPa; For the Bottom hole flowing pressure of gas injection well, MPa; , are the minimum and maximum bottom hole flowing pressures of the gas injection well, MPa; is the minimum gas injection volume of the gas injection well, m 3 ; For the Gas injection volume of gas injection wells, m 3 ; is the total gas injection volume of all gas injection wells in the reservoir, m 3 ; is the maximum gas injection volume of the gas injection well, m 3 ; For the j Water injection volume of injection wells, m 3 ; is the total water injection volume of all water injection wells in the reservoir, m 3 ; For the i Liquid production of production wells, m 3 ; is the total liquid production of all production wells in the reservoir, m 3 .

[0069] In another exemplary embodiment of the present application, referring to Figure 2 A more detailed implementation process of steps 103 to 109 may be steps 201 to 206 as follows.

[0070] Step 201: Initialize the intelligent optimization algorithm, set the optimization initial value X, the number of optimization iterations M and the number of populations N, and start the intelligent optimization system; where e=0, 1, 2…, M. The intelligent optimization algorithm can select a single-objective optimization algorithm as needed: particle swarm optimization algorithm, Bayesian optimization algorithm, adaptive differential evolution algorithm and deep learning model construction, etc. The multi-objective optimization algorithm can be selected: multi-objective artificial hummingbird algorithm, multi-objective particle swarm optimization algorithm and multi-objective genetic algorithm, etc. The optimization algorithm described in this application includes the algorithms listed above but is not limited to the algorithms listed.

[0071] Step 202: The intelligent optimization system automatically iterates and optimizes, calls the component simulator to calculate the e-th iteration optimization target value, and performs the optimization evaluation and optimization process. The component simulator calls the fitted component numerical simulation model to calculate the future oilfield production dynamics, calculates the e-th iteration optimization target value, and performs the optimization evaluation and optimization process.

[0072] Step 203: When the e-th iteration optimization is completed, determine whether the termination condition is met. If the optimization termination condition is met, proceed to step 206; if the termination condition is not met, proceed to step 204.

[0073] Step 204: Evolutionary agent model construction: collect e×N component simulation calculation schemes, establish a sample set, pre-process the sample set, and construct a Transformer-LSTM fusion model to predict the optimization target value; step 204 specifically includes the following steps.

[0074] ① Select the reservoir numerical simulation results of e×N injection-production parameter combination schemes generated during the e-th iteration of the optimization algorithm, construct a sample set for the deep learning model and perform preprocessing, including eliminating abnormal schemes and normalization. The input parameters of the deep learning model are the variables of the injection-production optimization mathematical model of each well, and the output parameters are the objective function values ​​(NPV, FOE, FOPT, etc.).

[0075] ②Build and design deep learning models to predict and optimize target values, such as Figure 3 As shown, this application constructs a Transformer-LSTM fusion model. Transformer can effectively process long sequence data in parallel, and LSTM is good at capturing short-term and long-term time dependencies. Combining the advantages of these two to build a Transformer-LSTM model can capture time features at different levels and provide more comprehensive time series modeling capabilities.

[0076] The Transformer-LSTM fusion model includes: a Transformer layer, a first LSTM layer, a second LSTM layer, and a fully connected layer connected in sequence. The input of the Transformer-LSTM fusion model is the gas injection and production parameters, and the output parameter is the objective function value of the optimization algorithm. The parameter changes of the Transformer layer, the first LSTM layer, the second LSTM layer, and the fully connected layer are determined by the hyperparameter training of the fusion model. The input of the Transformer layer is the gas injection and production parameters, and the output is the NPV and CO at each prediction time step. 2 Buried stock ( ). The input to the first LSTM layer is the NPV and CO at each prediction time step. 2 Buried stock ( ). The output of the first LSTM layer, the input and output of the second LSTM layer, and the input of the fully connected layer are determined by their respective parameters, and the changes in their respective parameters are determined by the hyperparameter training of the fusion model. The output of the fully connected layer is the objective function value of the optimization algorithm.

[0077] Step 205: Accelerated optimization based on evolutionary agent model: Determine the prediction accuracy R of deep learning model training 2 Whether it is greater than 0.95; if it is satisfied, replace the component numerical simulator in step 202 to calculate the optimization target value, and continue to loop step 202-step 206; if it is not satisfied, continue to loop step 202-step 206.

[0078] Step 206: until the termination condition in step 203 is met, end and output The optimal solution and optimization target value.

[0079] Example application: Taking the actual H reservoir CO of a certain oil field as an example 2 Taking the drive as an example, the method flow of the present application is illustrated.

[0080] (1) Collect and organize the basic parameters of the H reservoir development data of a certain oil field: porosity, permeability, saturation, historical working system of injection wells and production wells, etc. Figure 4 The permeability distribution shown is Figure 5 The pressure distribution shown, Figure 6 As shown in the saturation distribution and well location distribution, the target reservoir has 21 production wells, 12 water wells and 4 gas injection wells, and is developed by continuous gas injection. Figure 6 P1, P2, ..., P21 represent 21 oil wells, I2, I3, I4, I6, I7, I9, I10, I11, I13, I14, I15, I16 represent water wells, and I1, I5, I8, I12 represent gas injection wells.

[0081] (2) The objective function selects the economic net present value, as shown in formula (9). The objective function value is calculated based on the component numerical simulator. The optimization objectives are NPV and CO 2 Buried stock.

[0082] (9)

[0083] (3) Initialize the intelligent optimization algorithm and select the multi-objective particle swarm algorithm to carry out injection and production parameter optimization. Set the maximum number of iterations of the multi-objective particle swarm algorithm to 20, the population size to 50, and the maximum number of particles to 1000. The upper and lower limits of the optimization variables are set according to the actual situation on site as shown in Table 1. The relevant economic parameters are shown in Table 2.

[0084] Table 1 Upper and lower limits of optimization variables

[0085]

[0086] Table 2 Related economic parameters

[0087]

[0088] (4) The intelligent optimization system automatically iterates and searches for the optimal solution. MOPSO calls the component simulator to calculate the objective function for the e-th iteration and performs the optimization evaluation and search process.

[0089] (5) When the optimization is completed for the eth iteration, determine whether the optimization algorithm has reached the termination condition. If the optimization termination condition is met, proceed to step (8); if the termination condition is not met, proceed to step (6).

[0090] (6) In the construction of evolutionary agent model: the sample set is constructed using the formed injection and production scheme, and the Transformer-LSTM fusion model is used to predict the dynamics of reservoir development. The model hyperparameters are intelligently optimized using the Bayesian optimization method. The trained model is used to predict NPV and optimized using the MOPSO optimization algorithm. MOPSO-Transformer-LSTM collaborative optimization of the Pareto frontier is as follows: Figure 7 shown. Figure 7 Max Cs in CO 2 Max NPV represents the maximum value of the buried storage volume, and Max NPV represents the maximum economic net present value.

[0091] (7) Evolutionary agent model acceleration optimization: judging the prediction accuracy of deep learning model training R 2Is it greater than 0.95? If it is, replace the component numerical simulator in step (4) to calculate the objective function value, and continue to cycle the optimization system steps (4) to (8); if it is not satisfied, continue to cycle the optimization system steps (4) to (8). The relationship between the evolutionary training process and the number of iterations of the Transformer-LSTM fusion model is as follows: Figure 8 As shown in the figure, when MOPSO calls the component numerical simulator to optimize to the 9th generation, the prediction accuracy of the Transformer-LSTM fusion model reaches 0.95. This shows that the component numerical simulator can be used to continue the optimization, and the optimization efficiency can be improved by about 70%.

[0092] (8) Until the termination condition in step (5) is met, the optimization system ends and the output is The optimal solution and objective function value of MOPSO-Transformer-LSTM optimization H reservoir collaborative optimization results are shown in Table 3. The optimization results of gas injection wells are shown in Table 3. Fig. 9 As shown in Figure 2, the optimization results of the production well working system are as follows: Fig.10 As shown in Figure 2, the optimization results of the water injection working system are as follows: Fig.11 As shown in Figure 2, the oil saturation of the optimized scheme is compared with that of the basic scheme. Fig.12 and Fig.13 shown. Fig. 9 , Fig.10 and Fig.11 Δt1, Δt2, Δt3, Δt4, Δt5, Δt6, Δt7, Δt8, and Δt9 respectively represent 9 dynamic optimization stages of injection-production parameters divided according to the predicted production duration. Fig.13 So in it is the abbreviation of Oil Saturation, which stands for oil saturation.

[0093] Table 3 Results of collaborative optimization of H reservoir optimized by MOPSO-Transformer-LSTM

[0094]

[0095] The key technology of this application is to establish a method for optimizing gas-driven injection and production parameters in oil reservoirs based on deep learning technology and intelligent optimization algorithms, so as to quickly, economically and accurately determine the injection and production parameters of each well in the gas-driven oil reservoir. The entire accelerated optimization method is an innovation. (1) Construction of a mathematical model for collaborative optimization of injection and production parameters in gas-driven oil reservoirs: For different gas injection methods, mathematical models for collaborative optimization of injection and production parameters in different gas-driven oil reservoirs are constructed; (2) Accelerated optimization method for gas-driven injection and production parameters based on deep learning technology assisted intelligent optimization algorithm: The numerical simulation scheme in the optimization process is used as a sample set, a deep learning model is constructed for training, and the trained qualified model (Transformer-LSTM fusion model) is used to calculate the objective function. The injection and production parameters are optimized based on the deep learning model, and the fast calculation speed of the deep learning model is used to optimize the optimal solution. The key to this application is that it does not need to use the Latin hypercube design method to construct a large number of sample sets in advance, but uses the evaluation scheme in the optimization iteration process to construct the sample set. By continuously iterating and evolving the proxy model, a lot of time for constructing the sample set can be saved. (3) Transformer-LSTM fusion model architecture: Construct and design a deep learning model to predict F min This application constructs a Transformer-LSTM fusion model. Transformer can effectively process long sequence data in parallel, and LSTM is good at capturing short-term and long-term time dependencies. Combining the advantages of these two, the Transformer-LSTM fusion model can capture time characteristics at different levels and provide more comprehensive time series modeling capabilities. This application can quickly, economically and accurately determine the injection and production parameters of each well in the gas-driven oil reservoir.

[0096] Based on the same inventive concept, the embodiment of the present application also provides a reservoir gas drive injection and production parameter optimization device for implementing the reservoir gas drive injection and production parameter optimization method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more reservoir gas drive injection and production parameter optimization device embodiments provided below can refer to the limitations of the reservoir gas drive injection and production parameter optimization method above, and will not be repeated here.

[0097] In an exemplary embodiment, a reservoir gas drive injection and production parameter optimization device is provided, comprising: a simulation model building module, an optimization model construction module, an initialization module, an iterative optimization module, a training module, a judgment module, a replacement module, a return module and an output module.

[0098] A simulation model building module is used to build a component numerical simulation model of the target gas drive oil reservoir; an optimization model building module is used to build a gas drive injection and production parameter collaborative optimization mathematical model; an initialization module is used to initialize the number of iterations of the target optimization algorithm, and set e=1; e is a positive integer; an iterative optimization module is used to use the target optimization algorithm to perform the e-th iterative optimization, use the component numerical simulation model to determine the future oil field production dynamics in the e-th iterative optimization process, and use the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme based on the future oil field production dynamics; a training module is used to use the injection and production parameter combination schemes of all iterative optimizations that have been performed as input parameters, and the optimization target values ​​obtained from all iterative optimizations that have been performed as output parameters, if the value of e is less than the number of iterations. A sample set is established, and the Transformer-LSTM fusion model is trained by using the sample set; a judgment module is used to judge whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions; a replacement module is used to input the e-th injection-production parameter combination scheme into the trained Transformer-LSTM fusion model if the conditions are met, output the predicted optimization target value, and replace the predicted optimization target value with the optimization target value of the e-th injection-production parameter combination scheme, and increase the value of e by 1, and call the iterative optimization module; a return module is used to increase the value of e by 1, and call the iterative optimization module if the conditions are not met; an output module is used to output the e-th injection-production parameter combination scheme as the optimal injection-production parameter combination scheme for the target gas drive reservoir if the value of e is equal to the number of iterations.

[0099] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig.14 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the optimal injection and production parameter combination scheme of the target gas drive oil reservoir. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for optimizing gas drive injection and production parameters in an oil reservoir is implemented.

[0100] Those skilled in the art will understand that Fig.14 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0101] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0102] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0103] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0104] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for optimizing injection and production parameters of oil reservoir gas drive, characterized in that: include: Establish a component numerical simulation model of the target gas drive reservoir; Construct a mathematical model for collaborative optimization of gas drive injection and production parameters; Initialize the number of iterations of the target optimization algorithm and set e=1; e is a positive integer; Using a target optimization algorithm to perform an e-th iteration optimization, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iteration optimization process, and using the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme based on the future oilfield production dynamics; If the value of e is less than the number of iterations, the injection-production parameter combination schemes optimized by all iterations are used as input parameters, and the optimization target values ​​obtained by all iterations are used as output parameters to establish a sample set, and the Transformer-LSTM fusion model is trained using the sample set; Determine whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions; If satisfied, the e-th injection-production parameter combination scheme is input into the trained Transformer-LSTM fusion model, the predicted optimization target value is output, and the predicted optimization target value replaces the optimization target value of the e-th injection-production parameter combination scheme, and the value of e is increased by 1, and the step of "using the target optimization algorithm to perform the e-th iterative optimization, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iterative optimization process, and according to the future oilfield production dynamics, using the gas drive injection-production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection-production parameter combination scheme" is returned; If not satisfied, the value of e is increased by 1, and the process returns to step "using the target optimization algorithm to perform the e-th iterative optimization, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iterative optimization process, and according to the future oilfield production dynamics, using the gas drive injection and production parameter collaborative optimization mathematical model to obtain the optimization target value of the e-th injection and production parameter combination scheme"; If the value of e is equal to the number of iterations, the e-th injection-production parameter combination scheme is output as the optimal injection-production parameter combination scheme for the target gas drive reservoir.

2. The method for optimizing reservoir gas flooding injection and production parameters according to claim 1, characterized in that: Construct a mathematical model for collaborative optimization of gas drive injection and production parameters, including: Determining an optimization target; the optimization target is economic net present value, cumulative oil production, carbon dioxide storage capacity or recovery factor; According to the optimization goal, determine the optimization objective function; Set injection and production parameter optimization variables and constraints; The optimization objective function, injection-production parameter optimization variables and constraint conditions together constitute a mathematical model for collaborative optimization of gas drive injection-production parameters.

3. The method for optimizing reservoir gas flooding injection and production parameters according to claim 2, characterized in that: When the optimization target is the economic net present value, an optimization target function is established with the maximum economic net present value as the target; The calculation formula of economic net present value is: ； In the formula, is the economic net present value, , , Respectively The production well Annual oil production, gas production, and water production, For the The injection well is located in The total annual water injection volume, For the The gas injection well The total gas injection volume per year, is the selling price of gas, is the selling price of oil, is the water injection cost, is the cost of gas injection, is the cost of producing water treatment, is the discount rate, is the number of producing wells, is the number of injection wells, is the number of gas injection wells, For evaluation time.

4. The method for optimizing reservoir gas flooding injection and production parameters according to claim 2, characterized in that: When the optimization target is the carbon dioxide storage capacity, an optimization objective function is established with the maximum carbon dioxide storage capacity as the target; The calculation formula for carbon dioxide storage is: ； In the formula, is the carbon dioxide storage capacity, is the total amount of injected carbon dioxide, is the total amount of carbon dioxide produced.

5. The method for optimizing reservoir gas flooding injection and production parameters according to claim 2, characterized in that: When the gas injection method of the target gas drive reservoir is continuous gas injection, the injection and production parameter optimization variables of the injection well include gas injection timing, gas injection rate, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well include liquid production rate or bottom hole flow pressure; When the gas injection method of the target gas drive reservoir is intermittent gas injection, the injection and production parameter optimization variables of the injection well are gas injection timing, gas injection rate, gas injection cycle, well shut-down time, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well are liquid production rate or bottom hole flow pressure; When the gas injection method of the target gas drive reservoir is the gas-water alternating gas injection method, the injection and production parameter optimization variables of the injection well are the gas injection timing, gas injection rate, gas injection cycle, gas-water ratio, total gas injection volume of a single well and water injection rate, and the injection and production parameter optimization variables of the production well are the liquid production rate or the bottom hole flowing pressure of the production well.

6. The method for optimizing reservoir gas flooding injection and production parameters according to claim 2, characterized in that: The constraints include: boundary constraints and equality constraints; Boundary constraints include injection rate constraints of each gas injection well, injection timing constraints of each gas injection well, injection cycle constraints of each gas injection well, well blocking time constraints of each gas injection well, injection rate constraints of each water injection well, liquid production rate constraints of each production well, gas-water ratio constraints, bottom hole flow pressure constraints of gas injection wells and bottom hole flow pressure constraints of production wells; The equality constraints include total gas injection amount constraint, total water injection amount constraint and total liquid production amount constraint.

7. The method for optimizing reservoir gas flooding injection and production parameters according to claim 1, characterized in that: The Transformer-LSTM fusion model includes: a Transformer layer, a first LSTM layer, a second LSTM layer and a full-link layer connected in sequence.

8. An oil reservoir gas drive injection and production parameter optimization device, characterized in that: The oil reservoir gas drive injection and production parameter optimization device comprises: A simulation model building module is used to build a component numerical simulation model of the target gas drive reservoir; Optimization model building module, used to build a mathematical model for collaborative optimization of gas drive injection and production parameters; The initialization module is used to initialize the number of iterations of the target optimization algorithm and set e=1; e is a positive integer; An iterative optimization module, used for performing the e-th iterative optimization using a target optimization algorithm, using the component numerical simulation model to determine the future oilfield production dynamics during the e-th iterative optimization process, and obtaining the optimization target value of the e-th injection-production parameter combination scheme based on the future oilfield production dynamics using the gas drive injection-production parameter collaborative optimization mathematical model; A training module is used to establish a sample set if the value of e is less than the number of iterations, and to use the injection-production parameter combination schemes obtained through all the iterations as input parameters and the optimization target values ​​obtained through all the iterations as output parameters, and to train the Transformer-LSTM fusion model using the sample set; The judgment module is used to judge whether the prediction accuracy of the trained Transformer-LSTM fusion model meets the preset conditions; The replacement module is used to input the e-th injection and production parameter combination scheme into the trained Transformer-LSTM fusion model if it is satisfied, output the predicted optimization target value, and replace the predicted optimization target value with the optimization target value of the e-th injection and production parameter combination scheme, and increase the value of e by 1, and call the iterative optimization module; Return module, if not satisfied, then increase the value of e by 1 and call iterative optimization module; The output module is used to output the e-th injection-production parameter combination scheme as the optimal injection-production parameter combination scheme for the target gas drive reservoir if the value of e is equal to the number of iterations.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for optimizing injection and production parameters of oil reservoir gas drive according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for optimizing reservoir gas drive injection and production parameters according to any one of claims 1 to 7 is implemented.

Images