Optimization method and device for target well flow data

Abstract

The invention provides a target well flow data optimization method and device. The method comprises the steps that initial weight scaling parameters are input into a parameter selection equation, when parameter selection termination conditions are met, screening weight scaling parameters are output, the initial weight scaling parameters are determined according to flow field diagnosis index data of wells in the oil reservoir, and the flow field diagnosis index data are obtained according to injection-production data of the wells in the oil reservoir; the parameter selection equation is constructed according to a preset constraint and an initial weight scaling parameter response value; judging whether the number of times of optimizing the injection-production optimization model reaches an optimization termination condition or not; determining a target weight scaling parameter based on the target injection-production optimization model when the flow data of the target well is reached, so as to optimize the flow data of the target well; and according to the type of the target well in the oil reservoir, whether secondary optimization processing needs to be carried out on the flow data subjected to primary optimization or not is judged, and target flow data of the target well in the oil reservoir is obtained when the secondary optimization processing needs to be carried out. The method can effectively optimize the flow data of the target well in the oil reservoir.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas field development, and particularly to an optimization method and device for target well flow rate data. Background Art

[0002] There can be multiple wells in a reservoir (for example, multiple injection wells exist, or multiple production wells exist, or multiple injection wells and multiple production wells exist simultaneously). The multiple wells can generate corresponding flow rate data during injection and production. Accurately and efficiently optimizing the flow rate data generated by multiple wells in the reservoir is of great significance for improving the accuracy and efficiency of reservoir development regulation and formulating long-term development strategies.

[0003] Existing flow rate data optimization methods mainly include reservoir engineering methods, numerical simulation methods, simplified physical model methods, machine learning surrogate modeling optimization methods, streamline simulation-based flow rate update criterion methods, etc. Among them, reservoir engineering methods are highly empirical and have low accuracy; numerical simulation methods have high data requirements, high computational costs, and are time-consuming and laborious; simplified physical models (such as CRM, INSIM, and FNM, etc.) have strong multi-solution problems, weak long-term prediction capabilities, and low reliability; machine learning surrogate modeling optimization methods still take a long time to build surrogates, and the optimization efficiency for high-dimensional problems is relatively low; the streamline simulation-based flow rate update criterion still needs to be optimized, etc. That is, existing flow rate data optimization methods have a large amount of calculation, require a large amount of computing resources (such as memory, CPU, etc.), and thus have problems such as low optimization efficiency and accuracy, high computational costs, being time-consuming and laborious, and low reliability.

[0004] In view of the above problems, no effective solution has been proposed yet. Summary of the Invention

[0005] This specification provides an optimization method and device for target well flow rate data to reduce the occupation of computing resources and solve the problem that the prior art cannot accurately, efficiently, and cost-effectively optimize the flow rate data of different types of wells in a reservoir.

[0006] In a first aspect, an embodiment of this specification provides an optimization method for target well flow rate data, and the method includes:

[0007] Input an initial weight scaling parameter into a parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output a screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of wells in the reservoir. The flow field diagnosis index data is obtained from the injection-production data of wells in the reservoir. The parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to an injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model;

[0008] Determine whether the number of times of optimizing the injection-production optimization model by screening the weight scaling parameter reaches the optimization termination condition;

[0009] When the optimization termination condition is reached, determine the target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target wells in the reservoir based on the target weight scaling parameter to obtain the flow rate data after one optimization;

[0010] According to the type to which the target wells in the reservoir belong, determine whether it is necessary to perform secondary optimization processing on the flow rate data after one optimization. When secondary optimization processing is required, obtain the target flow rate data of the target wells in the reservoir.

[0011] In some embodiments, the flow field diagnosis index data includes at least one of the following: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency.

[0012] In some embodiments, the parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, and includes:

[0013] Construct a comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter;

[0014] Construct a parameter selection equation according to the comprehensive weight evaluation equation and the preset constraints of the initial weight scaling parameter according to the following formula:

[0015]

[0016] where L B is the parameter selection equation; x is the initial weight scaling parameter; μ is the multiplier vector corresponding to the equality constraint in the preset constraint; λ is the multiplier vector corresponding to the inequality constraint condition in the preset constraint; σ is the penalty factor; f merit is the comprehensive weight evaluation equation; l is the number of equality constraint conditions in the preset constraint; m is the number of inequality constraint conditions in the preset constraint; h i (x) is the equality constraint in the preset constraint; g j (x) is the inequality constraint in the preset constraint.

[0017] In some embodiments, the method further includes:

[0018] When the parameter selection termination condition of the parameter selection equation is not satisfied, re-obtain the initial weight scaling parameter and input it into the parameter selection equation, and continuously update the screening weight scaling parameter until the parameter selection termination condition is satisfied to obtain the updated screening weight scaling parameter;

[0019] Among them, the parameter selection termination condition includes that the total error of the equality constraint in the preset constraint and the inequality constraint in the preset constraint is less than the preset error threshold and the comprehensive weight evaluation value is less than the preset comprehensive weight evaluation threshold, and the comprehensive weight evaluation value is determined according to the comprehensive weight evaluation equation.

[0020] In some embodiments, the method further includes:

[0021] When the optimization termination condition is not reached, input the screening weight scaling parameter into the parameter selection equation, and when the parameter selection termination condition of the parameter selection equation is satisfied, output the second screening weight scaling parameter, where the second screening weight scaling parameter is different from the screening weight scaling parameter;

[0022] Obtain the second objective function value of the second screening weight scaling parameter, and update the data set where the screening weight scaling parameter is located based on the second screening weight scaling parameter and the second objective function value to obtain a second data set, where the data set where the screening weight scaling parameter is located includes the screening weight scaling parameter and the first objective function value of the screening weight scaling parameter;

[0023] Fit each screening weight scaling parameter in the second data set and the objective function value corresponding to each screening weight scaling parameter to obtain a second injection-production optimization model;

[0024] Judge again whether the optimization termination condition is reached. When the optimization termination condition is reached, use the second injection-production optimization model as the optimized target injection-production optimization model.

[0025] In some embodiments, obtaining the second objective function value of the second screening weight scaling parameter includes:

[0026] Construct an objective function according to the following formula:

[0027]

[0028] Among them, NPV represents the net present value (objective function); u represents the number of the time step; N t represents the total number of time steps; N pro is the total number of effective production wells; r o (t u ) represents the oil price at time t u ; t u represents the termination time corresponding to the time step u; q o,j (t u ) represents the oil production of production well j at time t u ; r w (t u ) represents the water treatment cost at time t u ; q w,j (t u ) represents tu Water production of production well j at time t; r wi (t u ) represents t u Water injection cost at time t; N inj Is the total number of effective injection wells; w i (t u ) represents t u Injection volume of injection well i at time t; t u-1 Represents the end time corresponding to time step u - 1; b represents the annual discount rate; t0 represents the initial optimization time;

[0029] Input the second screening weight scaling parameter into the objective function, and use the preset streamline simulation method to output the second objective function value.

[0030] In some embodiments, when secondary optimization processing is required, obtaining the target flow rate data of the target wells in the reservoir includes:

[0031] Perform secondary optimization processing on the flow rate data of the primary optimization according to the following formula:

[0032]

[0033]

[0034]

[0035] Among them, Is the injection volume of injection well i in the target flow rate data; Is the updated value of the lost injection volume of injection well i; Is the flow rate at injection well i on well pair (i, j) in the flow rate data of the primary optimization; Is the liquid production of production well j in the target flow rate data; Is the updated value of the injection volume of the aquifer into production well j; Is the flow rate at production well j on well pair (i, j) in the flow rate data of the primary optimization; AQ i Is the injection volume lost by injection well i due to injection into the aquifer; r AQ Is the proportion of the reduction in the lost injection volume.

[0036] In some embodiments, the method further includes:

[0037] Judge whether there is a flow rate upper bound constraint on the target flow rate data;

[0038] When there is a flow rate upper bound constraint, perform scaling processing on the target flow rate data to obtain the second target flow rate data, and the scaling processing is performed according to the following formula:

[0039]

[0040]

[0041] Among them, is the injection volume of injection well i in the second target flow rate data; N inj is the total number of effective injection wells; is the injection volume of injection well i in the target flow rate data; w sum is the target or upper bound of the total injection volume of all injection wells; is the liquid production volume of production well j in the second target flow rate data; N pro is the total number of effective production wells; is the liquid production volume of production well j in the target flow rate data; q sum is the target or upper bound of the total liquid production volume of all production wells.

[0042] In a second aspect, an embodiment of the present specification further provides an optimization device for target well flow rate data, and the device includes:

[0043] A screening weight scaling parameter output module, configured to input an initial weight scaling parameter into a parameter selection equation, and output a screened weight scaling parameter when a parameter selection termination condition of the parameter selection equation is satisfied. The initial weight scaling parameter is determined according to the flow field diagnosis index data of wells in the reservoir, the flow field diagnosis index data is obtained according to the injection-production data of wells in the reservoir, the parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, the response value of the initial weight scaling parameter is determined according to an injection-production optimization model, and the screened weight scaling parameter is used to optimize the injection-production optimization model;

[0044] A judgment module, configured to judge whether the number of times of optimizing the injection-production optimization model by the screened weight scaling parameter reaches an optimization termination condition;

[0045] A primary optimization module, configured to, when the optimization termination condition is reached, determine a target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of target wells in the reservoir based on the target weight scaling parameter to obtain once-optimized flow rate data;

[0046] A secondary optimization module, configured to judge whether secondary optimization processing needs to be performed on the once-optimized flow rate data according to the type of the target wells in the reservoir, and obtain the target flow rate data of the target wells in the reservoir when secondary optimization processing is required.

[0047] In a third aspect, an embodiment of the present specification further provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the optimization method for target well flow rate data in the above embodiments is implemented.

[0048] The embodiment of this specification provides a method and device for optimizing the flow rate data of a target well. First, input the initial weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir. The flow field diagnosis index data is obtained from the injection-production data of the wells in the reservoir. The parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to the injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model. Secondly, judge whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition. Then, when the optimization termination condition is reached, determine the target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target well in the reservoir based on the target weight scaling parameter to obtain the flow rate data optimized once. Finally, according to the type of the target well in the reservoir, judge whether it is necessary to perform secondary optimization processing on the flow rate data optimized once. When secondary optimization processing is required, obtain the target flow rate data of the target well in the reservoir. In the embodiment of this specification, by inputting the initial weight scaling parameter into the parameter selection equation, the screened weight scaling parameter can be quickly obtained, so that the optimization efficiency of the injection-production optimization model can be improved, and then the target weight scaling parameter can be determined in time, and the optimization efficiency of the flow rate data of the target well in the reservoir can be improved. By comprehensively constructing the parameter selection equation based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, the accuracy of obtaining the screened weight scaling parameter can be improved, so as to improve the optimization accuracy of the injection-production optimization model, and then improve the optimization accuracy of the flow rate data of the target well in the reservoir. By judging whether secondary optimization processing is required according to the type of the target well in the reservoir, the optimization direction of different types of wells can be clarified, and the accurate and efficient optimization of the flow rate data of different types of wells can be realized, so as to reduce the calculation amount and the occupancy rate of resources. Description of the Drawings

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. In the drawings:

[0050] Figure 1 is a schematic flow chart of a method for optimizing the flow rate data of a target well provided by the embodiment of this specification;

[0051] Figure 2 It is the injection destination diagram of the injection well in a certain 5-injection 4-production model provided by the embodiments of this specification;

[0052] Figure 3 It is the liquid production source diagram of the production well in a certain 5-injection 4-production model provided by the embodiments of this specification;

[0053] Figure 4 It is a schematic diagram of the relationship curve between W and β in the unbounded weight criterion provided by the embodiments of this specification;

[0054] Figure 5 It is a schematic diagram of the relationship curve between W and β in the bounded weight criterion provided by the embodiments of this specification;

[0055] Figure 6 It is a schematic diagram of the flow of the surrogate model optimization algorithm provided by the embodiments of this specification;

[0056] Figure 7 It is a schematic diagram of the calculation process of the GA_AL algorithm for the non-linear constraint optimization example provided by the embodiments of this specification;

[0057] Figure 8 It is a schematic diagram of the calculation process of the PSO_AL algorithm for the non-linear constraint optimization example provided by the embodiments of this specification;

[0058] Figure 9 It is a schematic diagram of the flow of an embodiment of the optimization method based on the target well flow rate data in a specific scenario example provided by this specification;

[0059] Figure 10 It is a schematic diagram of the flow of an embodiment of the optimization method based on the target well flow rate data in a specific scenario example provided by this specification;

[0060] Figure 11 It is a schematic diagram of the structural composition of an optimization device for target well flow rate data provided by this specification;

[0061] Figure 12 It is a schematic diagram of the structural composition of the electronic device provided by the embodiments of this specification. Detailed implementation manners

[0062] In order to enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification with reference to the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this specification without making creative efforts shall fall within the protection scope of this specification.

[0063] There can be multiple wells in a reservoir (e.g., multiple injection wells, multiple production wells, or both multiple injection wells and multiple production wells). When injecting and producing, these multiple wells can generate corresponding flow rate data. Optimizing the flow rate data generated by multiple wells in the reservoir accurately and efficiently is of great significance for improving the accuracy and efficiency of reservoir development control and formulating long-term development strategies.

[0064] Existing flow rate data optimization methods mainly include reservoir engineering methods, numerical simulation methods, simplified physical model methods, machine learning surrogate modeling optimization methods, flow rate update criterion methods based on streamline simulation, etc. Among them, reservoir engineering methods are highly empirical and have low accuracy; numerical simulation methods require a large amount of data, have high computational costs, and are time-consuming and laborious; simplified physical models (such as CRM, INSIM, and FNM, etc.) have strong multi-solution characteristics, weak long-term prediction ability, and low reliability; for machine learning surrogate modeling optimization methods, surrogate construction is still time-consuming, and the optimization efficiency for high-dimensional problems is relatively low; the flow rate update criterion of streamline simulation still needs to be optimized, etc. That is, existing flow rate data optimization methods have a large amount of calculation, require a lot of computing resources (such as memory, CPU, etc.), and thus have problems such as low optimization efficiency and accuracy, high computational cost, time-consuming and laborious, and low reliability.

[0065] In view of the above problems existing in the existing methods and the specific reasons for these problems, this application considers introducing an optimization method and device for the flow rate data of target wells, which can clarify the optimization directions of different types of wells in the reservoir, effectively optimize the flow rate data of different types of wells, reduce the optimization cost, improve the optimization accuracy and efficiency, and thus improve the accuracy and efficiency of reservoir development control, providing valuable guidance for formulating long-term development strategies.

[0066] Based on the above ideas, this specification proposes an optimization method for the flow rate data of target wells. The method includes: inputting an initial weight scaling parameter into a parameter selection equation, and when the parameter selection termination condition of the parameter selection equation is satisfied, outputting a screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of wells in the reservoir. The flow field diagnosis index data is obtained from the injection-production data of wells in the reservoir. The parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to an injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model; determining whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition; when the optimization termination condition is reached, determining a target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target wells in the reservoir based on the target weight scaling parameter to obtain the flow rate data after one optimization; according to the type to which the target wells in the reservoir belong, determining whether it is necessary to perform secondary optimization processing on the flow rate data after one optimization. When secondary optimization processing is required, obtaining the target flow rate data of the target wells in the reservoir.

[0067] Figure 1 is a schematic flow chart of an optimization method for the flow rate data of target wells provided by an embodiment of this specification. Although this specification provides method operation steps or device structures as shown in the following embodiments or drawings, based on routine or non-creative labor, more or some combined and fewer operation steps or module units may be included in the method or device. In steps or structures where there is no necessary causal relationship logically, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments or drawings of this specification. When the method or module structure is applied to an actual device, server or terminal product, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or drawings (for example, in an environment with parallel processors or multi-threaded processing, and even including an implementation environment of distributed processing and server clusters). Specifically, when implemented, refer to Figure 1 as shown, the method may include the following content.

[0068] S101: Input the initial weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir. The flow field diagnosis index data is obtained based on the injection-production data of the wells in the reservoir. The parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to the injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model.

[0069] In some embodiments, the injection-production data of the wells in the reservoir can be obtained first, and then according to the injection-production data of the wells in the reservoir, the flow field diagnosis index data of the wells in the reservoir can be determined. Then, according to the flow field diagnosis index data of the wells in the reservoir, the initial weight scaling parameter can be determined.

[0070] In some embodiments, there can be multiple wells in the above reservoir wells. For example, there can be multiple injection wells (injection wells can include water injection wells and gas injection wells) or multiple production wells (production wells can also be called production wells, and production wells can include oil production wells and gas production wells), or there can be multiple injection wells and multiple production wells at the same time (one production well and one injection well form a well pair or are called an injection-production well pair).

[0071] In some embodiments, streamline simulation can be divided into two categories: one is to use a streamline simulator for operation, and the pressure and saturation solutions respectively adopt two systems of finite difference grids and streamline flight time grids; the other is a streamline generation method coupled with a finite difference simulator, which requires streamline tracing based on the pressure and velocity (flow rate) fields obtained by the finite difference simulator, so as to output the corresponding results of the streamline simulator, which is equivalent to the post-processing process of the results such as the pressure, flow rate, and saturation fields of the numerical simulator (streamline simulation is a reservoir numerical simulation method. Streamlines are instantaneous. At any time point, a definite streamline trajectory can be traced along the velocity direction. Streamline simulation reveals the movement laws and three-dimensional spatial distributions of fluids (such as oil and water) by simulating the saturation changes along the streamlines; its most significant difference from finite difference simulation lies in the way of calculating the saturation distribution. Finite difference simulation calculates the pressure and saturation at each time step in the real three-dimensional space. The pressure calculation of streamline simulation is similar to that of finite difference simulation, but the saturation is obtained by leading edge tracking along the streamline flight time coordinate. Therefore, the streamline flight time grid is a "time-varying" grid that changes with the time step. Through streamline simulation, the movement trajectories of fluids at any moment can be vividly displayed, the flow conditions of fluids between production wells and injection wells can be characterized, and the flow rate distribution relationship between production wells and injection wells can be shown).

[0072] After history matching the reservoir numerical model, the injection-production correspondence can be quantitatively identified with the help of streamline post-processing tools of streamline simulators or numerical simulators. The injection well can be numbered as i, the production well can be numbered as j, the injection volume of injection well i can be denoted as w i , the liquid production volume of production well j can be denoted as q j , injection well i and production well j form a well pair (i, j), the total number of effective injection wells can be denoted as N inj , the total number of effective production wells can be denoted as N pro . The injection fluid flow rate flowing from injection well i to production well j is q i,j (measured at the injection well), the liquid production volume of production well j from injection well i is q j,i (measured at the production well), the water cut of production well j is f w,j , the water cut of well pair (i, j) at production well j at the production end is f w,j,i . w i and q j can be set with the unit of m 3 / d (a volume unit name, read as cubic meters per day).

[0073] The above injection-production data can include: the injection volume of injection well i is w i , the liquid production volume of production well j is q j , the injection fluid flow rate q flowing from injection well i to production well j i,j , the liquid production volume q of production well j from injection well i j,i , the total number of effective injection wells N inj , the total number of effective production wells N pro , the water cut of production well j is f w,j , the water cut of well pair (i, j) at production well j at the production end is f w,j,i . Among them, the injection volume w of injection well i i , the liquid production volume q of production well j j , the injection fluid flow rate q flowing from injection well i to production well j i,j , the liquid production volume q of production well j from injection well i j,i can be defined as flow rate data or injection-production flow rate data.

[0074] Among them, the injection fluid flow rate q flowing from injection well i to production well j i,j , can be determined according to the flow rate distribution coefficient of injection well i to production well j (IWAF i,j ), the injection volume w of injection well i i , according to the following formula:

[0075]

[0076] Among them, qi,j represents the injection fluid flow rate flowing from injection well i to production well j; w i represents the injection volume of injection well i; IWAF i,j represents the flow rate distribution coefficient of injection well i to production well j; N pro represents the total number of effective production wells.

[0077] Among them, the liquid production volume q at production well j from injection well i j,i , can be determined according to the proportion (PWAF j,i ) of the contribution of injection well i in the liquid production volume of production well j, the liquid production volume q of production well j j , according to the following formula:

[0078]

[0079] Among them, q j,i represents the liquid production volume at production well j from injection well i; q j represents the liquid production volume of production well j; PWAF j,i represents the proportion of the contribution of injection well i in the liquid production volume of production well j; N inj represents the total number of effective injection wells.

[0080] In some embodiments, when there are multiple injection wells and multiple production wells injecting and producing simultaneously in the reservoir, streamline simulation can visually display the connection relationship between injection and production wells, and it is easy to judge the injection destination of injection wells and the liquid production source of production wells. Refer to Figure 2 , Figure 2 is the injection destination diagram of injection wells in a 5-injection 4-production model, Figure 2 Part A) of Figure 2 is the injection destination of all injection wells, Figure 3 Part B) of Figure 3 is the injection destination of injection well INJ3 (i.e., the third injection well), Figure 3 is the liquid production source diagram of production wells in a 5-injection 4-production model, Figure 2 Part A) of Figure 3 In Part A) of Figure 2 : INJ1 represents the first injection well, INJ2 represents the second injection well, INJ3 represents the third injection well, INJ4 represents the fourth injection well, INJ5 represents the fifth injection well, PRO1 represents the first production well, PRO2 represents the second production well, PRO3 represents the third production well, and PRO4 represents the fourth production well. Figure 2 The connecting lines in Figure 3The connection lines in it indicate the sources of produced fluid, such as: Figure 3 In part B) of Figure 3 , the sources of produced fluid from the first production well P1 are INJ1, INJ2, and INJ3. Through streamline simulation, the injection directions of injection wells and the sources of produced fluid from production wells can be clearly judged, so that the flow field diagnosis indexes defined in this specification can be counted.

[0081] The above flow field diagnosis index data may include at least one of the following: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency. Below, an explanation is given on how to determine the well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency of wells in the reservoir based on the injection-production data of the wells in the reservoir.

[0082] (1) The well pair injection efficiency is defined as the ratio of the oil production at the production end of the well pair (the well pair is, for example, (i,j)) to the injection volume at the injection end. The injection fluid flow rate q flowing from injection well i to production well j in the injection-production data i,j , the produced fluid volume q from injection well i at production well j j,i , and the water cut f at production well j at the production end of the well pair (i,j) w,j,i can be used to determine the well pair injection efficiency according to the following formula (3):

[0083]

[0084] where IE i,j represents the well pair injection efficiency; q j,i represents the produced fluid volume from injection well i at production well j; f w,j,i represents the water cut at production well j at the production end of the well pair (i,j); q i,j represents the injection fluid flow rate flowing from injection well i to production well j.

[0085] The well pair average injection efficiency at the oilfield level can also be determined according to the injection volume w of injection well i i , the produced fluid volume q of production well j j , the total number of effective injection wells N inj , the total number of effective production wells N pro , and the water cut f of production well j w,j in the injection-production data according to the following formula (4):

[0086]

[0087] where (IE i,j ) ave represents the well pair average injection efficiency; N inj represents the total number of effective injection wells; w i represents the injection volume of injection well i; N proRepresents the total number of effective production wells; q j Represents the liquid production of production well j; f w,j Represents the water cut of production well j.

[0088] After determining the injection efficiency of the well pair and the average injection efficiency of the well pair, the injection efficiency of the well pair and the average injection efficiency of the well pair can also be compared to obtain the first comparison result; then, according to the first comparison result, the injection effect of the injection-production well pair and the adjustment direction of the flow rate data (injection volume q i,j ) corresponding to the injection-production well pair can be determined. For example:

[0089] If IE i,j >(IE i,j ) ave It indicates that the injection effect is better than the overall level of the oilfield, and the adjustment direction of the injection volume q i,j is: the injection volume q i,j on this well pair should be increased; if IE i,j <(IE i,j ) ave It indicates that the injection effect is lower than the overall level of the oilfield, and the adjustment direction of the injection flow rate q i,j is: the injection volume q i,j on this well pair should be reduced.

[0090] (2) The production efficiency of the well pair is defined as the ratio of the oil production at the production end of the well pair (well pair such as: (i, j)) to the liquid production at the production end. The water cut f w,j,i at the production well j at the production end in the injection-production data of the well pair (i, j) can be used to determine the production efficiency of the well pair according to the following formula (5):

[0091]

[0092] where, PE j,i represents the production efficiency of the well pair; q j,i represents the liquid production from the injection well i at the production well j; f w,j,i represents the water cut at the production well j at the production end of the well pair (i, j).

[0093] The average production efficiency of the well pair at the oilfield level can also be determined according to the liquid production q j of the production well j, the total number of effective production wells N pro , and the water cut f w,j of the production well j in the injection-production data according to the following formula (6):

[0094]

[0095] where, (PE j,i ) ave represents the average production efficiency of the well pair; N proIndicates the total number of effective production wells; q j Indicates the liquid production of production well j; f w,j Indicates the water cut of production well j.

[0096] After determining the well pair production efficiency and the average well pair production efficiency, the well pair production efficiency and the average well pair production efficiency can be compared to obtain a second comparison result; then, based on the second comparison result, the oil production effect of the injection-production well pair and the adjustment direction of the flow rate data (liquid production q j,i ) corresponding to the injection-production well pair can be determined. For example:

[0097] If PE j,i >(PE j,i ) ave It indicates that the oil production effect is better than the overall level of the oilfield, and the adjustment direction of the liquid production q j,i is: the liquid production q j,i on this well pair should be increased; if PE j,i <(PE j,i ) ave It indicates that the oil production effect is lower than the overall level of the oilfield, and the adjustment direction of the liquid production q j,i is: the liquid production q j,i on this well pair should be reduced.

[0098] (3) The injection efficiency of the injection well is defined as the crude oil production that can be recovered per unit injection volume. The injection efficiency of the injection well can be determined according to the injection volume w i of injection well i, the liquid production q j of production well j, the contribution ratio PWAF j,i of injection well i in the liquid production of production well j, the total number of effective production wells N pro , and the water cut f w,j,i at production well j at the production end of the well pair (i, j) in the injection-production data, according to the following formula (7):

[0099]

[0100] Among them, IE i represents the injection efficiency of the injection well; w i represents the injection volume of injection well i; N pro represents the total number of effective production wells; q j represents the liquid production of production well j; PWAF j,i represents the contribution ratio of injection well i in the liquid production of production well j; f w,j,i represents the water cut at production well j at the production end of the well pair (i, j).

[0101] The injection volume w i of injection well i and the liquid production q j of production well j in the injection-production data can also be used., the total number of effective injection wells N inj , the total number of effective production wells N pro , the water cut f of production well j w,j , according to the following formula (8), determine the average injection efficiency of injection wells:

[0102]

[0103] Among them, (IE i ) ave represents the average injection efficiency of injection wells; N inj represents the total number of effective injection wells; w i represents the injection volume of injection well i; N pro represents the total number of effective production wells; q j represents the liquid production volume of production well j; f w,j represents the water cut of production well j.

[0104] After determining the injection efficiency of injection wells and the average injection efficiency of injection wells, the injection efficiency of injection wells and the average injection efficiency of injection wells can also be compared to obtain the third comparison result; then, according to the third comparison result, determine the injection volume of the injection well and the adjustment direction of the flow rate data corresponding to the injection well (injection volume w i ), for example:

[0105] If IE i >(IE i ) ave It indicates that the current injection volume is not sufficient to exert its injection advantage, and the adjustment direction of the injection volume w i is: the injection volume w i should be increased; if IE i <(IE i ) ave It indicates that the current injection volume is too large, and the adjustment direction of the injection volume w i is: the injection volume w i should be reduced.

[0106] (4) The production efficiency of the production well is defined as the proportion of the oil production in the unit liquid production, that is, the oil content rate. The production efficiency of the production well can be determined according to the water cut f of production well j in the injection-production data w,j , according to the following formula (9):

[0107]

[0108] Among them, PE j represents the production efficiency of the production well; q j represents the liquid production volume of production well j; f w,j represents the water cut of production well j.

[0109] It is also possible to determine the average production efficiency of production wells according to the total number of effective production wells N in the injection-production data pro , the liquid production q of production well j j , and the water cut f of production well j w,j , and determine the average production efficiency of production wells according to the following formula (10)

[0110]

[0111] where (PE j ) ave represents the average production efficiency of production wells; N pro represents the total number of effective production wells; q j represents the liquid production of production well j; f w,j represents the water cut of production well j.

[0112] After determining the production efficiency of production wells and the average production efficiency of production wells, it is also possible to compare the production efficiency of production wells with the average production efficiency of production wells to obtain the fourth comparison result; and then, according to the fourth comparison result, determine the adjustment direction of the liquid production of this production well and the flow data corresponding to this production well (liquid production q j ), for example

[0113] If PE j >(PE j ) ave it indicates that the current liquid production is not sufficient to exert its advantage of high production efficiency, and the adjustment direction of liquid production q j is: the liquid production q j should be increased; if PE j <(PE j ) ave it indicates that the current production efficiency is insufficient, and the adjustment direction of liquid production q j is: the liquid production q j should be decreased.

[0114] So far, two concepts of injection-production efficiency at the well pair level (well pair injection efficiency, well pair production efficiency) and two concepts of injection-production efficiency at the single well level (injection well injection efficiency, production well production efficiency) have been defined in this specification. Among them, the adjustment directions of the well pair injection efficiency and the well pair production efficiency are the same. Both determine the increase or decrease of the well pair flow rate according to whether the well pair injection-production efficiency is higher or lower than the average well pair injection-production efficiency. The amplitude of the increase or decrease is determined by the difference between the well pair injection-production efficiency and the average well pair injection-production efficiency (that is, the flow rate update weight in the subsequent flow rate update criterion determines the increase, decrease or adjustment amplitude of the well pair or single well flow rate data. The flow rate update criterion and the flow rate update weight will be described separately later, and will not be elaborated here in this specification). The adjustment directions of the injection well injection efficiency and the production well production efficiency are also similar. Both determine the increase or decrease of the injection-production well flow rate according to whether the corresponding injection-production efficiency is higher or lower than the average injection-production efficiency. The amplitude of the increase or decrease is determined by the difference between the single well injection-production efficiency and the single well average injection-production efficiency. It should be noted that different flow rate optimization or update methods can be selected according to different flow field diagnosis indicators to optimize or update the flow rate data. For example:

[0115] When the flow field diagnosis indicator is only the well pair injection efficiency, select the flow rate optimization or update method one: Optimize or update the flow rate q at the injection end of the well pair only according to the well pair injection efficiency i,j and the flow rate q at the production end of the well pair j,i ;

[0116] When the flow field diagnosis indicator is only the well pair production efficiency, select the flow rate optimization or update method two: Optimize or update the flow rate q at the injection end of the well pair only according to the well pair production efficiency i,j and the flow rate q at the production end of the well pair j,i ;

[0117] When the flow field diagnosis indicators are the well pair injection efficiency and the well pair production efficiency, select the flow rate optimization or update method three: Optimize or update the flow rate q at the injection end of the well pair according to the well pair injection efficiency i,j and optimize or update the flow rate q at the production end of the well pair according to the well pair production efficiency j,i .

[0118] When the flow field diagnosis indicator is only the injection well injection efficiency, select the flow rate optimization or update method four: Optimize or update the flow rate w at the injection end only according to the injection well injection efficiency i , and the flow rate q at the production end j depends on the reservoir constraints and system response.

[0119] When the flow field diagnosis indicator is only the production well production efficiency, select the flow rate optimization or update method five: Optimize or update the flow rate q at the production end only according to the production well production efficiency j , and the flow rate w at the injection endi It depends on reservoir constraints and system responses.

[0120] When the flow field diagnosis indicators are the injection efficiency of injection wells and the production efficiency of production wells, select flow rate optimization or update method six: optimize or update the flow rate w at the injection end according to the injection efficiency of injection wells i , and optimize or update the flow rate q at the production end according to the production efficiency of production wells j .

[0121] It should be noted that when the change in the pressure along the way between injection and production wells is not obvious, the calculation results of the above three methods, namely method one, method two, and method three, do not differ much; on the contrary, the calculation results may vary slightly due to different volume coefficients (the volume coefficient refers to the ratio of the volume of a certain amount of fluid under formation temperature and pressure conditions to its volume under standard surface conditions), and the convergence rate of the third method is faster.

[0122] In some embodiments, after defining four flow field diagnosis indicators: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency, a flow rate update criterion or a flow rate update equation can be constructed according to the well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency (which can include an unbounded weight criterion or equation, a bounded weight criterion or equation; if the amplitude limit of the flow rate change is not considered and the weight is determined only based on the difference between the flow field diagnosis indicator and the average value of the field diagnosis indicator, this flow rate update criterion is called the "unbounded weight criterion"; if a threshold is set for the flow rate change during the calculation of the flow rate update weight, it is called the "bounded weight criterion"; the "bounded weight criterion" can avoid excessive flow rate changes between adjacent time steps during the injection-production flow rate regulation process and make the flow rate update amplitude more gentle), and then determine the initial weight scaling parameter from the constructed flow rate update criterion or flow rate update equation (the initial weight scaling parameters corresponding to different criteria or equations are different. For the unbounded weight criterion, the corresponding initial weight scaling parameter can be α, and for the bounded weight criterion, the corresponding initial weight scaling parameter can be [α, R, W min , W max ) T .

[0123] Specifically, first find the normalized quantity (or normalized value) of the flow field diagnosis indicators: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency to obtain the normalized quantity of the flow field diagnosis indicators, which is uniformly represented by β (0 ≤ β ≤ 1), and then calculate the normalized quantity of the average values of the flow field diagnosis indicators: well pair average injection efficiency, well pair average production efficiency, injection well average injection efficiency, and production well average production efficiency, which is uniformly represented by β ave , and then construct a flow rate update criterion or a flow rate update equation according to β and β ave .

[0124] Among them, according to the normalized quantity (β) of the flow field diagnosis index and the normalized quantity (β ave ) of the average value of the flow field diagnosis index, the unbounded weight criterion or equation can be constructed according to the following formula (11):

[0125]

[0126] Among them, W represents the flow rate update weight; β represents the normalized quantity of the flow field diagnosis index, 0 ≤ β ≤ 1; β ave represents the normalized quantity of the average value of the flow field diagnosis index (i.e., the normalized quantity of the injection-production efficiency in the average sense of the oilfield); α represents the weight index (which can be used as the initial weight scaling parameter mentioned above) and is greater than 0.

[0127] Refer to Figure 4 as shown, Figure 4 is a schematic diagram of the relationship curve between W and β in the unbounded weight criterion, Figure 4 showing the variation of the weight W with β under different weight indices α. When α = 1, W varies linearly with β; when α > 1, as α increases, the variation rate of W with β increases, and W shifts downward in the range of β < β ave and rises in the range of β > β ave ; when α < 1, as α decreases, the variation rate of W with β decreases, and W shifts upward in the range of β < β ave and drops in the range of β > β ave . Therefore, the weight index α affects the flow rate update result. If the unbounded weight criterion is used to optimize the injection-production flow rate data, the optimal value of the weight index α needs to be selected.

[0128] Among them, according to the normalized quantity (β) of the flow field diagnosis index and the normalized quantity (β ave ) of the average value of the flow field diagnosis index, the bounded weight criterion or equation can be constructed according to the following formula (12):

[0129]

[0130] Among them, W represents the flow rate update weight; W min represents the linear scaling ratio of the difference degree when the normalized quantity of the flow field diagnosis index is less than the normalized quantity of the average value of the flow field diagnosis index (i.e., the linear scaling ratio of Δβ when β is less than β ave ), taking [-1, 0); W max represents the linear scaling ratio of the difference degree when the normalized quantity of the flow field diagnosis index is greater than the normalized quantity of the average value of the flow field diagnosis index (i.e., the linear scaling ratio of Δβ when β is greater than β ave ), taking (0, 1]; R represents the calculation range of controlling β [β min , βmax The coefficient of ], 0 < R ≤ 1; β min Represents the lower bound of the calculation range of β; β max Represents the upper bound of the calculation range of β; β represents the normalized quantity of the flow field diagnosis index, 0 ≤ β ≤ 1; β ave Represents the normalized quantity of the average value of the flow field diagnosis index.

[0131] It can be seen from formula (12) that the lower bound of the flow rate update weight W is 1 + W min , and the upper bound is 1 + W max , that is, wells or well pairs with poor injection-production efficiency can reduce the flow rate to 0 (when W min = -1), while wells or well pairs with high injection-production efficiency can increase the flow rate to at most twice the original (when W max = 1).

[0132] Among them, the derivation process of formula (12) is as follows:

[0133] During the injection-production flow rate regulation process, to avoid excessive flow rate changes between adjacent time steps, it is sometimes necessary to set limits on the weight to make the flow rate update more gentle, but at the same time follow the sorting result of the flow field diagnosis index. The flow rate update criterion adopted at this time is the "bounded weight criterion". Introduce the linear scaling ratio W lim To limit the change range of the weight, for the bounded weight rule, the flow rate update weight equation is shown in formula 13:

[0134] W = 1 + W lim ·(Δβ) α (13)

[0135] Among them, W lim Represents the linear scaling ratio of the difference degree (Δβ) of the flow field diagnosis index; Δβ represents the difference degree of the flow field diagnosis index; α represents the weight index.

[0136] When β < β ave , Δβ and W lim Are defined as:

[0137]

[0138] When β > β ave , Δβ and W lim Are defined as:

[0139]

[0140]

[0141] Among them, R represents the calculation range of controlling β [β min , βmax ], 0 <R≤1;β min Indicates the lower bound of the β calculation range; β max Indicates the upper limit of the β calculation range; W min Indicates that β is less than β ave The linear scaling ratio of Δβ is [-1,0); W max Indicates that β is greater than β ave The linear scaling ratio of Δβ is (0,1].

[0142] Substituting formula (16) into formula (15) and formula (14) and rearranging them, we can obtain:

[0143]

[0144] By segmenting the β range according to formula (17), we can obtain:

[0145]

[0146] Substituting formula (18) into formula (13) yields formula (12), which is the final form of the bounded weight criterion or equation.

[0147] If R = 0.85, W min =-1, W max =1,β ave =0.47, then formula (12) can be transformed into:

[0148]

[0149] See also Figure 5 As shown, Figure 5 It is a diagram of the relationship curve between W and β in the bounded weight criterion. Figure 5 The W-β relationship curve corresponding to different α values ​​corresponding to formula (19) is shown as follows: Figure 5 It can be seen that the normalized diagnostic index β corresponding to a single well or an injection-production well pair (streamline bundle) deviates from β ave When the weight exceeds 1, the flow rate increases, and vice versa, the flow rate decreases. If α = 1, then the increase or decrease of the flow rate is proportional to the deviation of β from the average value β ave The degree of is linearly proportional; when α>1, in β ave The influence of nearby wells or well pairs is relatively weakened because their weights are close to 1, while the weight change rate of single wells or well pairs (streamline bundles) whose attribute values ​​deviate from the average value will increase. When 0<α<1, the situation is the opposite. ave The weight changes near β will be strengthened, while those away from β ave The weight change will be weakened when min ,βmax The coefficient R of [] is less than 1, which may cause poor performance of shut-in wells ( Figure 5 in the case where β < 0.0705 and the weight is 0), and at the same time limits the flow rate of the best-performing wells ( Figure 5 when β > 0.8695 in [], the weight no longer increases).

[0150] In some embodiments, after constructing a flow rate update criterion or a flow rate update equation according to the flow field diagnosis index, different weight scaling parameters can be determined according to the actually selected flow rate update criterion. For example: if the actually selected flow rate update criterion is the "unbounded weight criterion", the weight scaling parameter can be determined as α; when the actually selected flow rate update criterion is the "bounded weight criterion", the weight scaling parameter can be determined as [α, R, W min , W max T .

[0151] In some embodiments, different flow rate update methods can be selected according to the type of the flow field diagnosis index, and combined with a randomly selected flow rate update criterion to optimize or update the flow rate data of well pairs or single wells (wherein, the flow rate update criterion can also be called a flow rate update weight equation, which can determine the flow rate update weight W, and the flow rate update weight W can in turn determine the amplitude of the optimization of the flow rate data of well pairs or single wells).

[0152] Referring to Table 1, if the flow field diagnosis index is only the well pair injection efficiency at the well pair level, the selected flow rate update method needs to be Method 1, and the flow rate update criterion can be any one of the "unbounded weight criterion" (abbreviated as unbounded) and the "bounded weight criterion" (abbreviated as bounded), and the flow rates q i,j at the injection end of the well pair and q j,i at the production end of the well pair can be optimized or updated based on the well pair injection efficiency.

[0153] Table 1

[0154] Statistical level Flow field diagnosis index Flow rate update method Flow rate update criterion Well pair Injection efficiency of well pair Method 1 Bounded / unbounded Well pair Production efficiency of well pair Method 2 Bounded / unbounded Well pair Injection efficiency of well pair, production efficiency of well pair Method 3 Bounded / unbounded Single well Injection efficiency of injection well Method 4 Bounded / unbounded Single well Production efficiency of production well Method 5 Bounded / unbounded Single well Injection efficiency of injection well, production efficiency of production well Method 6 Bounded / unbounded

[0155] Among them, the flow rate data of well pairs or single wells can be optimized or updated according to the following formula (Formula (20) can represent the flow rate update equation):

[0156] q new = q old ·W (20)

[0157] Among them, q new represents the updated flow rate, m 3 / d; q old represents the flow rate before update (which can be w i , q j or q​i,j , q j,i ), m 3 / d; W represents the flow rate update weight.

[0158] That is, through the flow rate update criterion, the flow rate update weight W can be determined (defining the ratio of the updated flow rate to the flow rate before update as the flow rate update weight W), and then according to the flow rate update weight W, the flow rate data q of the well pair or single well old is optimized or updated to obtain the updated flow rate data q of the well pair or single well new , that is, the optimal flow rate update scheme of the well pair or single well (i.e., the target flow rate update scheme of the target well) is obtained, so that the optimal reservoir injection-production scheme of the well pair or single well (i.e., the target reservoir injection-production scheme of the target well) can be determined, and thus based on the optimal reservoir injection-production scheme, the accuracy and efficiency of reservoir development regulation can be improved.

[0159] In some embodiments, the flow field diagnosis results obtained by generating streamlines according to streamline simulation or finite difference simulation can clearly judge the injection direction of the injection well and the liquid production source of the production well, so that the flow field diagnosis indexes in this specification can be statistically analyzed. Combining with the flow rate update criterion, the injection-production flow rate can be optimized. This injection-production optimization method based on streamline simulation flow field diagnosis involves a "sequential optimization" process, and only one forward simulation is required to optimize the injection-production flow rate settings of all optimization time steps, and it is not sensitive to the problem dimension (the number of injection-production wells), so the calculation cost can be significantly saved.

[0160] For example: for the injection-production optimization problem with the number of injection wells being N inj and the number of production wells being N pro , and the optimization step being N t , the dimension of the decision variable vector is (N inj +N pro )×N t , that is, each optimization step needs to optimize (N inj +N pro ) flow rate settings. In the first optimization step, first, the streamline simulation results at the beginning of optimization (i.e., the end of historical production) are statistically analyzed, and the selected flow field diagnosis indexes are statistically analyzed according to this simulation result. Then, according to the flow field diagnosis indexes and the flow rate update criterion, the injection-production flow rate of the first optimization step is optimized. After that, the optimized injection-production flow rate is set as the injection-production target and the first optimization step is simulated; the second optimization step is to statistically analyze the flow field diagnosis indexes according to the simulation results of the first optimization step, and then optimize the injection-production flow rate setting of this time step; the entire optimization process is carried out sequentially. The N t th optimization step is to statistically analyze the flow field diagnosis indexes according to the simulation results of the N t -1th optimization step, and then optimize the injection-production flow rate setting of the last optimization step; thus, the entire streamline simulation optimization process only requires one time including Nt Numerical simulation for one time step.

[0161] The streamline simulation injection-production optimization method fully implements flow rate allocation optimization based on the physical meaning of the flow field. When the flow rate update criterion is set appropriately, its optimization result is generally better than the historical benchmark plan or the artificially set injection-production plan. However, it may not be the optimal plan because the flow rate update criterion on which the streamline simulation production optimization depends can still be further optimized (i.e., the initial weight scaling parameters in the flow rate update criterion, such as: α, R, W min , W max can still be further optimized. Different values of α, R, W min , W max will affect the calculation result of the flow rate update weight W, thereby affecting the flow rate data optimized or updated by formula (20), and further affecting the formulation of the flow rate update plan, making the flow rate update plan unable to reach the best, resulting in a reduction in reservoir development efficiency.

[0162] In some embodiments, in order to obtain a set of optimal initial weight scaling parameters such as: α, R, W min , W max (the optimal initial weight scaling parameters are the above-mentioned target weight scaling parameters), the existing method is as follows: Based on the initial weight scaling parameters, a sub-function based on the reservoir numerical simulator is pre-constructed, and the objective function is calculated according to the simulation results of the reservoir numerical simulator. The objective function can be the net present value function of the reservoir, the cumulative oil production function, etc. The role of this sub-function based on the reservoir numerical simulator is to receive a set of specific values of the initial weight update parameters input by the user and calculate the objective function value. When the objective function value reaches the maximum (such as: the maximum NPV, the maximum cumulative oil production, etc.), the corresponding initial weight scaling parameter is the optimal value, that is, the target weight scaling parameter can be obtained when ensuring that the objective function value reaches the maximum. However, taking the initial weight scaling parameters as [α, R, W min , W max T as an example, since α, R, W min , W max all have different value range limitations (i.e., preset constraints, which can include equality constraints and inequality constraints), [α, R, W min , W max T can have hundreds or thousands of different value combinations. Inputting hundreds or thousands of different value combinations into the reservoir numerical simulator for calculation or directly using the global optimization algorithm to call the simulator to obtain the maximum value of the objective function will consume a large amount of computing power and a large amount of time, and may take several months. Obviously, this method cannot be applied to specific reservoir exploitation scenarios. ​​

[0163] In order to address the problem that existing methods cannot obtain the maximum value of the objective function at a low cost, accurately and quickly, and thus cannot determine the optimal initial weight scaling parameter (i.e., the target weight scaling parameter) at a low cost, accurately and quickly. This specification pre-constructs a parameter selection equation. The initial weight scaling parameter can be input into the pre-constructed parameter selection equation. When the parameter selection termination condition of the parameter selection equation is met, the screened weight scaling parameter is output. Then, the screened weight scaling parameter is used to optimize the pre-constructed injection-production optimization model, thereby obtaining the target injection-production optimization model (the target injection-production optimization model is the optimal or best injection-production optimization model). Finally, the target weight scaling parameter can be determined based on the target injection-production optimization model. Among them, the injection-production optimization model can also be called a surrogate model. The surrogate model is a function of the same type as the objective function. If the objective function is a net present value function, the surrogate model is also a net present value function. However, the form of the surrogate model is simpler and the calculation speed is faster than that of the objective function. The surrogate model optimization algorithm is to select some weight update parameter values with good representativeness and a small number, input them into the objective function to calculate their corresponding objective function values, and then combine the weight scaling parameter with its corresponding objective function value to form multiple sample points. The surrogate model is used to fit the sample points to make the surrogate model as close as possible to the objective function. Finally, a target surrogate model with high accuracy, very close to the objective function, but a simpler function form (i.e., the above-mentioned target injection-production optimization model) is obtained. When the output of the target surrogate model reaches the maximum value, the corresponding input data is a set of globally optimal weight scaling parameters (i.e., the target weight scaling parameters). Based on the surrogate model optimization algorithm, the number of times of calling the reservoir numerical simulator for calculation can be reduced, the calculation time can be reduced, and at the same time, a set of optimal weight scaling parameters (target weight scaling parameters) can be obtained. After obtaining the target weight scaling parameter, the flow rate data of the target well can be optimized or updated by combining the globally optimal weight scaling parameter with formula (20) to obtain the optimal flow rate update plan. The well pairs or single wells' flow rate data are regulated according to the optimal flow rate update plan, and the net present value or cumulative oil production obtained will be approximately equal to the maximum value of the net present value function or cumulative oil production in the objective function. This process is injection-production optimization.

[0164] In some embodiments, the above-mentioned parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. In specific implementation, it may include:

[0165] S1: Construct a comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter;

[0166] S2: According to the comprehensive weight evaluation equation and the preset constraints of the initial weight scaling parameter, construct the parameter selection equation according to the following formula:

[0167]

[0168] Among them, L B is the parameter selection equation; x is the initial weight scaling parameter; μ is the multiplier vector corresponding to the equality constraint in the preset constraint; λ is the multiplier vector corresponding to the inequality constraint condition in the preset constraint; σ is the penalty factor; f merit is the comprehensive weight evaluation equation; l is the number of equality constraint conditions in the preset constraint; m is the number of inequality constraint conditions in the preset constraint; h i (x) is the equality constraint in the preset constraint; g j (x) is the inequality constraint in the preset constraint.

[0169] For the convenience of understanding, the preset constraint of the initial weight scaling parameter, the response value of the initial weight scaling parameter, how to construct the comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter, and finally how to construct the parameter selection equation will be described separately as follows:

[0170] In some embodiments, the preset constraint of the above-mentioned initial weight scaling parameter may include the equality constraint h i (x) and the inequality constraint g j (x) of the initial weight scaling parameter. If the initial weight scaling parameters are α, R, W min , W max (i.e., the "bounded weight criterion" mentioned above is selected during traffic update), then the equality constraint h i (x) may include but is not limited to:

[0171] R = 1; W min = -1; W max = 1

[0172] The inequality constraint g j (x) may include but is not limited to:

[0173] α > 0; 0 < R < 1; -1 < W min < 0; 0 < W max < 1

[0174] In some embodiments, the injection-production optimization model can be constructed first according to the initial weight scaling parameter, and then the response value of the initial weight scaling parameter can be determined by using the constructed injection-production optimization model. Among them, the process of constructing the injection-production optimization model according to the initial weight scaling parameter can be as follows:

[0175] First, the initial weight scaling parameter is sampled by using a preset sampling method (such as: Latin Hypercube Sampling (LHS), Symmetric Latin Hypercube Sampling (SLHS)) to obtain the sampled weight scaling parameter;

[0176] Then, the sampling weight scaling parameter can be input into the objective function or objective equation (assuming the objective function is as shown in formula (22), which is the net present value function), and the objective function value corresponding to the sampling weight scaling parameter (i.e., the net present value corresponding to the sampling weight scaling parameter) can be obtained;

[0177] Among them, the objective function or objective equation can be constructed according to the following formula:

[0178]

[0179] Among them, NPV represents the net present value (objective function, i.e., the net present value function); u represents the number of the time step; N t represents the total number of time steps; N pro is the total number of effective production wells; r o (t u ) represents the oil price at time t u ; t u represents the termination time corresponding to the time step u; q o,j (t u ) represents the oil production of production well j at time t u ; r w (t u ) represents the water treatment cost at time t u ; q w,j (t u ) represents the water production of production well j at time t u ; r wi (t u ) represents the water injection cost at time t u ; N inj is the total number of effective injection wells; w i (t u ) represents the injection volume of injection well i at time t u ; t u-1 represents the termination time corresponding to u - 1; b represents the annual discount rate; t0 represents the initial optimization time;

[0180] Finally, the sampling weight scaling parameter and the objective function value corresponding to the sampling weight scaling parameter can be fitted (a sample point can be constructed according to the sampling weight scaling parameter and the objective function value corresponding to the sampling weight scaling parameter, and then this sample point can be fitted), and an injection-production optimization model can be obtained (this injection-production optimization model can be used as the initial injection-production optimization model, and subsequent further optimization of this initial injection-production optimization model is required to obtain the target injection-production optimization model that best approximates the objective function).

[0181] In some embodiments, the specific form of the above injection-production optimization model can be as shown in the following formula (23):

[0182]

[0183] p(x) = [x T , 1]·[c1, c2,..., c n , c0] T = p·c (24)

[0184] Where, S ur (x) represents the response value of x, that is, the output value of the injection-production optimization model; x represents the decision variable, [x1, x2, x3,..., x n T , that is, the input value of the injection-production optimization model. In this application, α or is used as the decision variable; h represents the number of sample points for which the true objective function has been evaluated; λ a represents the RBF interpolation model coefficient vector of the a-th sample point for which the true objective function has been evaluated; RBF(·) represents the RBF interpolation model; ||x - x a || represents the Euclidean norm / distance (or the modulus of the vector) between x and x a ; x a represents the decision variable of the a-th sample point for which the true objective function has been evaluated; p(x) represents the linear tail of the RBF interpolation model; p represents the row vector constructed by adding the element 1 to the decision variable, equal to [x1, x2, x3,..., x n , 1]; n represents the dimension of the decision variable (problem dimension); c represents the coefficient vector of the linear tail of the RBF interpolation model, equal to [c1, c2, c3,..., c n , c0] T .

[0185] To solve for the RBF interpolation model coefficient vector λ and the coefficient vector c of the linear tail of the RBF interpolation model, the following linear equations can be considered for construction:

[0186]

[0187]

[0188] R ad = RBF(||x a - x d ||), (1 ≤ a, d ≤ h) (27)

[0189]

[0190]

[0191] Fun = [f(x1), f(x2), f(x3),..., f(x h )] T (30)​

[0192] Among them, Phi represents the RBF interpolation model; R ad represents the function value of the RBF interpolation model; P represents the constructed sample matrix; P T represents the transpose matrix of P; x a,d represents the d-th element of the a-th decision variable for which the true objective function has been evaluated; λ represents the RBF interpolation model coefficient; Fun represents the objective function vector corresponding to the sample points for which the true objective function has been evaluated; f(x) represents the objective function corresponding to the decision variable x; x1, x2, x3, x h represent different decision variables for which the true objective function has been evaluated.

[0193] When the rank of the matrix P is n + 1 (h ≥ n + 1), the coefficient matrix in the linear equation system expressed by formula (25) is invertible, and it is denoted as A rbf , then:

[0194]

[0195] C = A rbf \B (32)

[0196]

[0197] According to formula (32), the matrix C to be solved for the RBF interpolation model can be obtained, and then the RBF interpolation model coefficient λ and the coefficient vector c of the linear tail of the RBF interpolation model can be obtained.

[0198] In some embodiments, the RBF interpolation model specifically has the following 6 forms:

[0199] (1) Cubic spline function:

[0200] RBF(r) = r 3 (34)

[0201] (2) Thin plate spline function:

[0202] RBF(r) = r 2 ln r (35)

[0203] (3) Gaussian function:

[0204]

[0205] (4) Multiquadric function:

[0206] RBF(r) = (r 2 + f 2 ) g ,(f ≠ 0, g > 0, Na represents natural numbers)(37)

[0207] (5)Inverse multiquadric function:

[0208] RBF(r) = (r 2 + f 2 ) -g ,(f ≠ 0, g > 0) (38)

[0209] (6)Surface spline function:

[0210]

[0211] where r represents the input value of the RBF interpolation model, g and f are both parameters in the RBF interpolation model, uneven represents odd numbers in natural numbers, and even represents even numbers in natural numbers.

[0212] For the surface spline function, when g = 1, it becomes a linear function; when g = 3 or 2, it is converted into a cubic spline function or a thin plate spline function.

[0213] The RBF interpolation model in the above injection-production optimization model in this specification can be a cubic spline function. Of course, according to different usage scenarios, other forms of RBF interpolation models can also be selected as the above injection-production optimization model. For example, an exact radial basis network (RBE), a generalized regression neural network (GRNN), a radial basis network (RB), or a radial basis function neural network (RBFN) can also be used as the above injection-production optimization model, and this specification does not make specific limitations on this.

[0214] In some embodiments, after constructing the injection-production optimization model (formula (23)), the initial weight scaling parameter can be input or substituted into formula (23), and the response value of the initial weight scaling parameter can be obtained using formula (23). The response value can also be referred to as the surrogate response value, which is the output value of the injection-production optimization module or the surrogate model, and the response value changes with the change of the input initial weight scaling parameter.

[0215] In some embodiments, after obtaining the response value of the initial weight scaling parameter, a comprehensive weight evaluation equation can be constructed based on the response value of the initial weight scaling parameter, that is, the above step S1 (construct a comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter). In specific implementation, it may include:

[0216] S11: Obtain the distance from the initial weight scaling parameter to the weight scaling parameter of the already obtained objective function value. According to the distance and the response value of the initial weight scaling parameter, construct a comprehensive weight evaluation equation according to the following formula:

[0217]

[0218] where f merit (x) represents the comprehensive weight evaluation equation; w represents the weight of the surrogate minimization criterion, where 0 < w < 1; represents the normalized value of the response value of the initial weight scaling parameter (i.e., the normalized value of the surrogate response at the sample point x); represents the normalized value of the target distance from the initial weight scaling parameter to the weight scaling parameter of the evaluated objective function value (i.e., the normalized value of the minimum distance from the sample point x to the evaluated sample point).

[0219] The comprehensive weight value evaluation equation combines the surrogate minimization criterion and the search maximization criterion (the surrogate minimization criterion requires selecting the sample point with the minimum surrogate model response value (the weight scaling parameter corresponding to the minimum response value) from the feasible candidate range, and this criterion focuses on the search near the current minimum value of the surrogate model. The search maximization criterion requires selecting the sample point farthest from the evaluated sample point from the feasible candidate range, and this criterion focuses on expanding the search range and avoiding falling into local optimal solutions), and can comprehensively consider the balance between accelerating the optimization and expanding the search range; the comprehensive weight value function should ensure the comparability of the physical quantities represented by the two criteria (i.e., the surrogate response and the distance), so it needs to be normalized. The surrogate response data and distance data can be normalized using the mean and standard deviation, or the maximum and minimum values or other methods can be used for normalization.

[0220] Therefore, the normalized value of the response value in the above formula (40) can also be determined by the following formula (41) or formula (42) and the normalized value of the target distance

[0221]

[0222]

[0223] where min(·) represents the minimum value operator; max(·) represents the maximum value operator; ave(·) represents the arithmetic mean operator; sd(·) represents the standard deviation operator; S ur (x) represents the response value of the initial weight scaling parameter (i.e., the response value of the sample point x); represents the normalized value of the response value of the initial weight scaling parameter (i.e., the normalized value corresponding to S ur (x)); S ur represents the set of response values of the weight scaling parameters within the feasible candidate range (i.e., the set of response values of the sample points within the feasible candidate range); D is(x) represents the target distance from the initial weight scaling parameter to the weight scaling parameter of the obtained objective function value (i.e., the minimum distance from the sample point x to the sample point with the evaluated value); represents the normalized value of the target distance from the initial weight scaling parameter to the weight scaling parameter of the obtained objective function value (i.e., the normalized value corresponding to D is (x)); D is represents the set of target distances from the sample points within the feasible candidate range to the weight scaling parameter of the obtained objective function value (i.e., the set of minimum distances between the weight scaling parameters within the feasible candidate range and the sample points with the obtained objective function value).

[0224] It should be noted that normalization can be performed using (41) or formula (42), or by using the maximum and minimum values or other methods. This specification does not make specific limitations in this regard.

[0225] In some embodiments, the feasible candidate range in the above surrogate minimization criterion and search maximization criterion can be determined in the following manner:

[0226] First, it is necessary to determine the initial candidate range of the next screening weight scaling parameter (i.e., determine the next function evaluation point, that is, the next evaluation point (weight scaling parameter) for which the objective function value needs to be obtained). Since the computational cost of the injection-production optimization model (or surrogate model) is very low, this specification sets 4 sources for the initial candidate range: 1) randomly perturbing near the current best feasible point; 2) performing Latin hypercube sampling within the bounded constraint range; 3) performing symmetric Latin hypercube sampling within the bounded constraint range; 4) performing random sampling within the bounded constraint range. Different numbers of sampling points can be set for each source, generally set to 100·n (n is the problem dimension).

[0227] Secondly, in the sampling problem, if in addition to the bounded constraints, there are also complex constraints such as linear equalities, linear inequalities, nonlinear equalities, and nonlinear inequalities in the optimization problem, taking an n-dimensional bounded nonlinear constraint optimization problem f2(x) as an example:

[0228]

[0229] Among them, f2(x) represents the optimization problem for which the minimum value needs to be solved; s.t. represents the constraint conditions of f2(x); x represents the decision variable; x ub represents the upper bound of the decision variable, x lb represents the lower bound of the decision variable; A represents the coefficient matrix of the linear inequality constraint; b1 represents the linear inequality constraint constant, which is a column vector; A eq represents the coefficient matrix of the linear equality constraint; b eq represents the linear equality constraint constant, which is a column vector; non ineqRepresents the row vector of non - linear inequality constraints; non eq Represents the row vector of non - linear equality constraints.

[0230] For each decision variable x, define its constraint violation degree function as:

[0231]

[0232] If f pen ≤ 0 or less than the pre - set constraint tolerance error, it indicates that the sample point x is feasible. Using the above - mentioned constraint violation degree function (Formula (44)), the feasible points that meet f pen ≤ 0 (less than the pre - set constraint tolerance error) can be screened out from the above - mentioned initial candidate range; in addition, the candidate points that are too close to the evaluated sample points in the initial candidate range need to be excluded (or set the value function of these points to infinity), so as to obtain the feasible candidate range.

[0233] In some embodiments, after constructing the comprehensive weight evaluation equation Formula (40), in combination with the preset constraints of the initial weight scaling parameter, the parameter - selection equation shown in Formula (21) can be constructed. The specific construction idea of the parameter - selection equation is as follows:

[0234] For the general constrained optimization problem as follows:

[0235]

[0236] Among them, x is an n - dimensional vector; h i (x) is the equality constraint; g j (x) is the inequality constraint; l is the number of equality constraints; m is the number of inequality constraints.

[0237] First, consider the equality constraint in Formula (45), and construct its augmented Lagrange function as:

[0238]

[0239]

[0240] Among them, L A Represents the augmented Lagrange function; L represents the Lagrange function; μ represents the Lagrange multiplier vector of the equality constraint, [μ1, μ2, …, μ l T ; σ represents the penalty coefficient.

[0241] Denote the gradient of the equality constraint vector h(x) as:

[0242] ​

[0243]

[0244] Substituting formula (47) into formula (46), we get:

[0245]

[0246] Taking the partial derivatives of formula (50) with respect to each component of x, we get:

[0247]

[0248] Written in the form of a gradient, it is:

[0249]

[0250] Taking the gradient of formula (47) with respect to x, we get:

[0251]

[0252] Observing the above two formulas, it can be found that the update method of the multiplier sequence {μ k} is:

[0253] μ k+1 = μ k - σ·h(x k ) (54)

[0254] Now consider the inequality constraint in formula (45). Introducing the auxiliary variable s j (>0), the inequality constraint is transformed into:

[0255] g j (x) - s j = 0 (55)

[0256] Similarly, the augmented Lagrange function at this time is:

[0257]

[0258] where λ represents the Lagrange multiplier vector of the inequality constraint, [λ1, λ2,..., λ l T ;

[0259] Taking the partial derivative of formula (56) with respect to s, we get

[0260]

[0261] Since each element of the slack vector s is non-negative, then

[0262] ​

[0263] Therefore

[0264]

[0265] Substituting Equation (59) into Equation (56) gives:

[0266]

[0267] That is

[0268]

[0269] Equation (61) is the augmented Lagrange function of the inequality constraint; by following Equation (54), the iterative formula for the Lagrange multiplier vector of the inequality constraint can be written as:

[0270]

[0271] Or

[0272] λ k+1 = max[λ k - σ·g(x k ), 0] (63)

[0273] In summary, for the constrained optimization problem of Equation (45), the augmented Lagrange function including both equality constraints and inequality constraints can be written as:

[0274]

[0275] where σ is the penalty factor (penalty coefficient); μ and λ are the multiplier vectors corresponding to the equality constraint and the inequality constraint respectively, and their iterative formulas are:

[0276]

[0277] The Lagrange multiplier method adds equality and inequality constraint conditions to the objective function, converts the general constrained optimization into an unconstrained optimization problem, and continuously solves for x by continuously updating the Lagrange multipliers (μ, λ), thereby seeking the extreme point that satisfies all constraints. Its termination criterion can be set as:

[0278]

[0279] where: ε represents the maximum error of the constraint condition (i.e., the preset error threshold), such as 1×10 -4 ;

[0280] The Lagrange multiplier method is used to handle the constraint conditions. This kind of method is very effective for non-time-consuming constraint optimization problems. However, when the evaluation cost of the objective function is relatively large, it still needs to be combined with the above-mentioned surrogate optimization algorithm. The comprehensive evaluation equation (40) can be substituted into formula (64) to obtain the parameter selection equation (21) in this specification. The parameter selection equation is used to obtain the screening weight scaling parameter (i.e., the weight scaling parameter for which the objective function needs to be evaluated using formula (22) next) to realize the optimization and update of the injection-production optimization model.

[0281] In some embodiments, after the parameter selection equation is constructed, the initial weight scaling parameter can be input into the pre-constructed parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, the screening weight scaling parameter is output. Among them, the parameter selection termination condition may include: the sum of the errors of the equality constraint in the preset constraint and the inequality constraint in the preset constraint is less than the preset error threshold and the comprehensive weight evaluation value is less than the preset comprehensive weight evaluation threshold. The comprehensive weight evaluation value is determined according to the comprehensive weight evaluation equation. Among them, the sum of the errors of the equality constraint in the preset constraint and the inequality constraint in the preset constraint being less than the preset error threshold may include (formula (66)):

[0282]

[0283] where h i (x) is the equality constraint in the preset constraint; g j (x) is the inequality constraint in the preset constraint; l is the number of equality constraint conditions in the preset constraint; m is the number of inequality constraint conditions in the preset constraint; ε represents the preset error threshold; σ is the penalty factor; λ represents the multiplier vector of the inequality constraint; λ j represents the multiplier of the jth one (the multiplier vector can also be called the Lagrange multiplier vector).

[0284] Among them, the comprehensive weight evaluation value being less than the preset comprehensive weight evaluation threshold means that the comprehensive weight evaluation value obtained by using formula (40) is the minimum comprehensive weight evaluation value. When the preset constraint error and the comprehensive function evaluation value in formula (21) reach the minimum, the output screening weight scaling parameter is the optimal weight scaling parameter. This optimal weight scaling parameter can be used to optimize the above-mentioned injection-production optimization model (i.e., the initial injection-production optimization model).

[0285] In some embodiments, a global optimization algorithm (such as genetic algorithm GA, pattern search PS, particle swarm optimization PSO, simulated annealing SA) can be combined with the above-constructed parameter selection equation (Lagrange multiplier method) to solve the constrained optimization problem of the weight scaling parameter, so as to quickly find the point with the most function evaluation value (candidate point, that is, the optimal screened weight scaling parameter). The optimal point can be added to the above-mentioned feasible candidate range. For example, if multiple unconstrained optimization algorithms are combined with the above parameter selection equation, then the candidate points obtained by each optimization algorithm can be all added to the feasible candidate range. Finally, a sample point (i.e., weight scaling parameter) with the smallest comprehensive weight evaluation value and preset constraint error can be selected from the feasible candidate range as the next function evaluation point (i.e., screened weight scaling parameter), and its objective function is evaluated (i.e., the objective function is evaluated using formula (22)) and added to the evaluated sample points (i.e., the screened weight scaling parameter with the evaluated objective function value), so as to update the injection-production optimization model or surrogate model and continue the next candidate point sampling. This process is cycled until the termination condition is met. It should be noted that the weight w of the surrogate minimization criterion in the comprehensive weight value equation should not be kept fixed, but should be updated and changed with the optimization process. The present invention sets two update methods:

[0286] 1) Take values cyclically within a fixed sequence (such as the point set composed of 10 equal division points of [10 -6 ,1]);

[0287] 2) Take values cyclically within a fixed sequence and supplement with the optimization effect response. For example, when the true objective function values obtained by two consecutive samplings are both lower than the minimum value of the previous true objective function, w is increased to accelerate the optimization; on the contrary, if the true objective function values of two consecutive samplings are both higher than the minimum value of the previous true objective function, w is decreased to expand the search area.

[0288] In some embodiments, the above method may further include:

[0289] When the parameter selection termination condition of the parameter selection equation is not satisfied, re-obtain the initial weight scaling parameter and input it into the parameter selection equation, and continuously update the screened weight scaling parameter until the parameter selection termination condition is satisfied, and the updated screened weight scaling parameter is obtained;

[0290] Wherein, the parameter selection termination condition includes: the sum of the errors of the equality constraint in the preset constraint and the inequality constraint in the preset constraint is less than the preset error threshold and the comprehensive weight evaluation value is less than the preset comprehensive weight evaluation threshold, and the comprehensive weight evaluation value is determined according to the comprehensive weight evaluation equation.

[0291] In some embodiments, if the selection parameter termination condition of the selection parameter equation is not met, a new round of iteration needs to be started, that is, the initial weight scaling parameter needs to be obtained again and input into the selection parameter equation, and the screened weight scaling parameter is continuously updated until the selection parameter termination condition is met, and the updated screened weight scaling parameter is output. By inputting the initial weight scaling parameter into the selection parameter equation, the screened weight scaling parameter can be accurately and quickly obtained and updated, thereby improving the efficiency and accuracy of optimizing the injection-production optimization model based on the screened weight scaling parameter in the subsequent stage.

[0292] S102: Determine whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition.

[0293] In some embodiments, after obtaining the screened weight scaling parameter, it can be first determined whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition. The optimization termination condition can be that the number of optimization times reaches a preset number threshold (that is, the number of times of evaluating the objective function with the screened weight scaling parameter reaches the preset evaluation number threshold). If the optimization termination condition is reached, the injection-production optimization model can be directly optimized based on the screened weight scaling parameter to obtain the target injection-production optimization model (the optimal or best injection-production optimization model).

[0294] In some embodiments, the above method may further include:

[0295] When the optimization termination condition is not reached, the screened weight scaling parameter is input into the selection parameter equation. When the selection parameter termination condition of the selection parameter equation is met, the second screened weight scaling parameter is output, and the second screened weight scaling parameter is different from the screened weight scaling parameter;

[0296] Calculate the second objective function value of the second screened weight scaling parameter, and update the data set where the screened weight scaling parameter is located based on the second screened weight scaling parameter and the second objective function value to obtain a second data set. The data set where the screened weight scaling parameter is located includes the screened weight scaling parameter and the first objective function value of the screened weight scaling parameter;

[0297] Fit the screened weight scaling parameters in the second data set and the objective function values corresponding to the screened weight scaling parameters to obtain a second injection-production optimization model;

[0298] Determine again whether the optimization termination condition is reached. When the optimization termination condition is reached, the second injection-production optimization model is used as the optimized target injection-production optimization model.

[0299] In some embodiments, if the number of times the screening weight scaling parameter optimizes the injection-production optimization model does not reach the optimization termination condition, the screening weight scaling parameter obtained last time can be continuously input into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, the second screening weight scaling parameter is output (the second screening weight scaling parameter can represent the screening weight scaling parameter output for the second time by the parameter selection equation, abbreviated as the second screening weight scaling parameter, which is different from the screening weight scaling parameter last time). Then, the second objective function value of the second screening weight scaling parameter is obtained (for example: the net present value of the second screening weight scaling parameter), and then the data set where the screening weight scaling parameter last time is located is updated based on the second screening weight scaling parameter and the second objective function value of the second screening weight scaling parameter. The data set where the screening weight scaling parameter last time is located can include the screening weight scaling parameter and the first objective function value of the screening weight scaling parameter. The updated data set can include the screening weight scaling parameter, the first objective function value of the screening weight scaling parameter, the second screening weight scaling parameter, and the second objective function value of the second screening weight scaling parameter. Then, each screening weight scaling parameter in the second data set and the objective function value corresponding to each screening weight scaling parameter are fitted, and a new injection-production optimization model can be obtained, that is, the second injection-production optimization model mentioned above. Then, it is determined whether the current optimization times reach the optimization termination condition. If the optimization termination condition is reached, the second injection-production optimization model can be used as the target injection-production optimization model. If the optimization termination condition is still not reached, the above process can be repeated, that is: the second screening weight scaling parameter is input into the parameter selection equation, the third screening weight scaling parameter is output, the objective function value of the third screening weight scaling parameter is obtained, the second data set is updated according to the third screening weight scaling parameter and the objective function value of the third screening weight scaling parameter to obtain the third data set, and then the third data set is fitted to obtain the third injection-production optimization model, and it is determined whether the optimization termination condition is reached. If the optimization termination condition is reached, the third injection-production optimization model is used as the target injection-production optimization model mentioned above.

[0300] In some embodiments, when specifically implementing the obtaining of the second objective function value of the second screening weight scaling parameter, it may include:

[0301] Construct an objective function according to the following formula:

[0302]

[0303] where NPV represents the net present value (objective function); u represents the number of the time step; N t represents the total number of time steps; N pro is the total number of effective production wells; r o (t u ) represents the oil price at time t u ; t uDenote the termination time corresponding to time step u; q o,j (t u ) represents the oil production rate of production well j at time t u ; r w (t u ) represents the water treatment cost at time t u ; q w,j (t u ) represents the water production rate of production well j at time t u ; r wi (t u ) represents the water injection cost at time t u ; N inj is the total number of effective injection wells; w i (t u ) represents the injection volume of injection well i at time t u ; t u-1 Denote the termination time corresponding to u - 1; b represents the annual discount rate; t0 represents the initial time;

[0304] Input the second screening weight scaling parameter into the objective function, and output the second objective function value.

[0305] In some embodiments, the second screening weight scaling parameter can be input into the objective function, and then combined with formula (20) for reservoir simulation calculation. The q corresponding to different time steps can be calculated in sequence o,j 、q w,j 、w i , and then the objective function value NPV can be obtained by combining formula (22). It is equivalent to calling a numerical simulator to execute a subroutine for pure streamline simulation production optimization. The input of this program is the weight scaling parameter, and the output is the defined objective function (e.g., NPV).

[0306] By judging whether the number of times of optimizing the injection-production optimization model with the screening weight scaling parameter reaches the optimization termination condition, when the optimization termination condition is not reached, the next screening weight scaling parameter can be obtained in time to effectively optimize the injection-production model, and then the target injection-production optimization model can be obtained accurately and quickly.

[0307] S103: When the optimization termination condition is reached, determine the target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target wells in the reservoir based on the target weight scaling parameter to obtain the flow rate data after the first optimization.

[0308] In some embodiments, when the optimization termination condition is reached, a target injection-production optimization model (i.e., the optimal injection-production optimization model) can be obtained. This target injection-production optimization model is closest to the defined objective function (e.g., the net present value function). Thus, an optimal function value can be output based on the target injection-production optimization model, which is the optimal net present value (i.e., the maximum net present value). Correspondingly, the input data corresponding to the output optimal function value can be used as the target weight scaling parameter. After obtaining the target weight scaling parameter, the target weight scaling parameter can be substituted into the flow rate update criterion or equation (e.g., it can be substituted into the above formula (11) or formula (12)) to obtain the target flow rate update weight. The flow rate data of the target well can be optimized by using the above formula (20) in combination with the determined target flow rate update weight (e.g., when the target well is an injection-production well pair, optimizing the flow rate data q i,j 、q j,i ; when the target well is an injection well or a production well, optimizing the flow rate data w i 、q j ). The flow rate data obtained by optimizing the flow rate data of the target well with the target flow rate update weight can be used as the flow rate data of the first optimization. It is also possible to directly find the injection-production optimization plan corresponding to the weight scaling parameter in the optimization simulation results.

[0309] S104: According to the type of the target well in the reservoir, determine whether secondary optimization processing is required for the flow rate data of the first optimization. When secondary optimization processing is required, obtain the target flow rate data of the target well in the reservoir.

[0310] In some embodiments, when secondary optimization processing is required as described above to obtain the target flow rate data of the target well in the reservoir, in specific implementation, it may include:

[0311] Perform secondary optimization processing on the flow rate data of the first optimization according to the following formula:

[0312]

[0313]

[0314]

[0315] Wherein, is the injection volume of injection well i in the target flow rate data; is the updated value of the lost injection volume of injection well i; is the flow rate at injection well i in well pair (i, j) in the flow rate data of the first optimization; is the liquid production volume of production well j in the target flow rate data; is the updated value of the injection volume of the aquifer to production well j; is the flow rate at production well j in well pair (i, j) in the once-optimized flow rate data; AQ i is the injection volume lost by injection well i due to injection into the aquifer; r AQ is the reduction ratio of the lost injection volume.

[0316] In some embodiments, the type of the target well in the above reservoir may include well pairs or single wells. For well pairs, after the flow rate data is optimized once, it is also necessary to perform a secondary optimization process on the once-optimized flow rate data (that is, after the flow rate data on the well pair is optimized or updated, it needs to be reflected to the single well level by superposition, and the secondary optimization process is to superpose the once-optimized flow rate data to the single well level). For single wells, after the flow rate data is optimized once, there is no need to perform a secondary optimization process. By judging whether it is necessary to perform a secondary optimization process on the once-optimized flow rate data according to the type of the target well in the reservoir, and when a secondary optimization process is required, the target flow rate data of the target well in the reservoir can be obtained, which can effectively optimize the flow rate data at the well pair level.

[0317] In some embodiments, the above method may further include:

[0318] judging whether there is a flow rate upper bound constraint on the target flow rate data;

[0319] When there is a flow rate upper bound constraint, perform a scaling process on the target flow rate data to obtain the second target flow rate data, and the scaling process is carried out according to the following formula:

[0320]

[0321]

[0322] where is the injection volume of injection well i in the second target flow rate data; N inj is the total number of effective injection wells; is the injection volume of injection well i in the target flow rate data; w sum is the target or upper bound of the total injection volume of all injection wells; is the liquid production volume of production well j in the second target flow rate data; N pro is the total number of effective production wells; is the liquid production volume of production well j in the target flow rate data; q sum is the target or upper bound of the total liquid production volume of all production wells.

[0323] In some embodiments, after superimposing the flow rate data of the wells onto the single well level to obtain the target flow rate data, it is also possible to determine whether there is a flow rate upper bound constraint for the target flow rate data (the flow rate upper bound constraint such as: the upper bound of the total injection volume and production fluid volume or the target constraint). If there is a flow rate upper bound constraint, it is necessary to perform a scaling process on the target flow rate data to obtain the second target flow rate data.

[0324] In some embodiments, if the type of the target well is a single well, there is no need to perform a secondary optimization process on the flow rate data optimized once. At this time, it is also possible to determine whether there is a flow rate upper bound constraint for the target well. If there is a flow rate upper bound constraint, the flow rate of the single well can also be scaled according to the above formulas (70) and (71). Additionally, if there are other constraints set for the single well, such as economic index constraints like the maximum injection pressure, the minimum bottom hole pressure of the production well, the minimum injection volume, and the maximum water cut of the production well, it is also necessary to optimize or update the flow rate under the condition of satisfying these constraints.

[0325] It should be noted that the above-mentioned primary optimization process and secondary optimization process can be executed in the flow field diagnosis and optimization sub-function, and all optimization steps (the 1st optimization step to the Nth t optimization step) will execute the above-mentioned primary optimization process and secondary optimization process.

[0326] By determining whether there is a flow rate constraint, and performing a scaling process on the flow rate data of the target well when there is a constraint, the optimization accuracy of the flow rate data can be improved, so that a more accurate flow rate update plan can be obtained, and thus reservoir exploration and development can be carried out accurately and efficiently.

[0327] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and the key points of each embodiment are the differences from other embodiments. Specifically, reference can be made to the descriptions of the relevant processing-related embodiments above, and details will not be repeated here.

[0328] The above is an explanation of the method. However, it should be noted that this specific embodiment is only for better explaining the present application, and a specific embodiment of the specification is described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be executed in a different order from that in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In certain embodiments, multi-tasking and parallel processing are also possible or may be beneficial.

[0329] In a specific scenario example, refer to Figure 6 as shown Figure 6It is a flow diagram of the proxy model optimization algorithm. Figure 6 In it, the abscissa x represents the weight scaling parameter, and the ordinate f(x) represents the output value of the injection-production optimization model or the proxy model. As Figure 6 shown, the main steps to sort out and optimize the injection-production optimization model or the proxy model to obtain the target injection-production optimization model or the target proxy model are as follows:

[0330] Step 1) Initial experimental design: Use the preset sampling method to sample the initial weight scaling parameter to obtain the sampled weight scaling parameter (which can be used as a sampling point), and evaluate the objective function to obtain the objective function value corresponding to the sampled weight scaling parameter. The sampled weight scaling parameter and the objective function value corresponding to the sampled weight scaling parameter can be used as the sample point set (the sample point set is the above-mentioned data set), that is, Figure 6 The multiple points in Step 1 represent the sample point set;

[0331] Step 2) Construct a proxy model: Fit the sample point set to obtain the initial injection-production optimization model or the initial proxy model (that is, Figure 6 the curve in Step 2 represents the constructed initial injection-production optimization model);

[0332] Step 3) Select a new function evaluation point: Input the initial weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter, calculate the objective function value of the screened weight scaling parameter, and add the screened weight scaling parameter for which the objective function value is calculated to the sample point set to obtain the first sample point set (that is, Figure 6 the multiple points in Step 3 represent the first sample point set, where the newly appeared points represent the screened weight scaling parameters for which the objective function values have been calculated) (the first sample point set is the above-mentioned first data set);

[0333] Step 4) Update the proxy model: Fit the first sample point set to obtain the optimized injection-production optimization model or the optimized proxy model (that is, Figure 6 the curve in Step 4 represents the optimized injection-production optimization model or the optimized proxy model);

[0334] Step 5) Select a new function evaluation point: When the optimization termination condition is not reached (or the parameter selection termination condition is not satisfied) (when the optimization termination condition is not reached, it means that the number of times of evaluating the objective function for the screened weight scaling parameter has not reached the preset evaluation times threshold), continue to obtain the second screened weight scaling parameter, calculate the second objective function value of the second screened weight scaling parameter, and add the second screened weight scaling parameter for which the second objective function value is calculated to the first sample point set to obtain the second sample point set (that is, Figure 6 the multiple points in Step 5 represent the second sample point set, where the newly appeared points represent the second screened weight scaling parameters for which the objective function values have been calculated) (the second sample point set is the above-mentioned second data set);

[0335] Step 6) Update the surrogate model: Fit the second set of sample points. When the optimization termination condition is reached (the optimization termination condition means that the number of times of screening the weight scaling parameter for objective function evaluation reaches the preset evaluation times threshold), obtain the target injection-production optimization model or the target surrogate model (that is, Figure 6 the curve in Step 6 represents the target injection-production optimization model or the target surrogate model). If the optimization termination condition is still not reached, repeat Step 5 and Step 6 until the optimization termination condition is met.

[0336] In a specific scenario example, refer to Figure 7 、 Figure 8 as shown, Figure 7 FIG. is a schematic diagram of the calculation process of the GA_AL algorithm for the non-linear constraint optimization example, Figure 8 and FIG. is a schematic diagram of the calculation process of the PSO_AL algorithm for the non-linear constraint optimization example. When using the genetic algorithm combined with the Lagrange multiplier method (GA_AL) to solve the optimization problem in Equation (70), the optimal solution is [78.1019, 35.4090, 31.5147, 43.7894, 33.6046] T , and the best objective function value is -30364.8388. The optimization process is as Figure 7 shown, Figure 7 where the horizontal axis Generation in FIG. represents the number of iterations, the vertical axis Penalty value represents the penalty value, Bestpenalty value represents the best penalty value, and Mean penalty value represents the average penalty value. It can be seen from Figure 7 that the best penalty value Best penalty value is -30364.8, and the average penalty value Mean penalty value is -30355.3. The combination of the Lagrange multiplier method (parameter selection equation (21)) and the particle swarm algorithm (GA) can effectively solve complex constraint optimization problems.

[0337] When using the particle swarm algorithm combined with the Lagrange multiplier method (PSO_AL) to solve the optimization problem as shown in Equation (72), the optimal solution is [78, 33, 29.9952, 45, 36.7758] T , and the best objective function value is -30665.5585. The optimization process is as Figure 8 shown, Figure 8 where the horizontal axis Iteration in FIG. represents the number of iterations, the vertical axis Function value represents the function value, and BestFunction value represents the best function value. It can be seen from Figure 8It can be seen that the Best Function value is -30665.6. The combination of the Lagrange multiplier method and the classical gradient-free intelligent algorithm can effectively solve complex constrained optimization problems. For this example, the Particle Swarm Optimization (PSO) has obtained better optimization results than the Genetic Algorithm (GA).

[0338]

[0339] In a specific scenario example, Figure 9 It represents a schematic flow diagram of an optimization method based on target well flow rate data. First, through streamline model simulation or finite difference simulation, streamline tracing, and flow field diagnosis, flow field diagnosis indicators are constructed: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency. Then, according to the preset flow rate update criteria (unbounded weight criterion, bounded weight criterion), injection and production flow rates are allocated, that is, a flow rate update equation (Formula (20)) is constructed. When operating the flow rate update equation, a surrogate model optimization algorithm is introduced to determine the target flow rate update weight. For the surrogate model optimization algorithm, first, an initial experimental design is carried out, that is, according to the preset sampling method (such as: Latin Hypercube Sampling (LHS), Symmetric Latin Hypercube Sampling (SLHS)), the initial weight scaling parameter is sampled to obtain the sampled weight scaling parameter, and the number of samplings should be greater than the problem dimension. Then, an initial injection and production optimization model or surrogate model is constructed, which can use a Radial Basis Function Interpolation Model (RBF) or a Radial Basis Function Neural Network (RBFN). When selecting the weight update parameter, a sampling method (strategy) also needs to be set, and the cyclic weight value evaluation sampling method (Formula (40) in Formula (21)) is adopted. The cyclic weight value evaluation sampling method reflects both the surrogate minimization criterion and the search maximization criterion (comprehensively using the two criteria of surrogate minimization and search maximization, and according to the comprehensive weight value evaluation equation, selecting the point with the highest potential for evaluating the objective function from the feasible candidate range jointly composed of the candidate points screened based on the constraint violation function and the candidate points given by the Lagrange multiplier method (the process of constructing the parameter selection equation is the Lagrange multiplier method)) as the next batch of function evaluation points. This method is called the "modified cyclic weight value evaluation sampling method").

[0340] In a specific scenario example, Figure 10 It represents a schematic flow diagram of an optimization method based on target well flow rate data. It can be carried out according to the following steps:

[0341] 1. Define the optimization problem (determine the decision variables, constraints, and objective function): For example, in the bounded weight rule, α, R, W min ,W maxAs decision variables, the net present value function (NPV) obtained based on the flow field diagnosis index and the bounded weight rule is used as the objective function, and the decision variable constraint conditions and the optimization termination conditions are specified.

[0342] 2. Write the flow field diagnosis optimization sub-function: The process of calling the reservoir numerical simulator for flow field diagnosis optimization is used as the flow field diagnosis optimization sub-function. Its input is the weight update parameter in the flow rate update rule. The injection and production flow rates at each optimization time step are optimized for flow field diagnosis (sequential optimization) by combining the flow field diagnosis index with the flow rate update rule. Through the writing of the numerical simulation data file and the reading of the simulation result file, the objective function value is output; the pressure and flow rate constraints of single wells, well pairs, or oilfields involved in the optimization are set in the numerical simulation file.

[0343] 3. Set the surrogate model optimization algorithm: Set the sampling method, the function type of the initial injection-production optimization model or surrogate model, the adjustment times threshold, the minimum distance of the sample point spacing, the number of initial weight update parameters, the minimum number of sampling points for updating the injection-production optimization model or surrogate model, etc.

[0344] 4. Determine the initial sampling range: Determine the candidate range of the weight scaling parameter; and select the weight scaling parameter from the candidate range; at the same time, use the sampling method to select the sampling weight scaling parameter, and call the flow field diagnosis optimization sub-function to evaluate the objective function, generating sample points. This process can use parallel computing.

[0345] 5. Construct the initial injection-production optimization model or initial surrogate model; Based on the function of the initial injection-production optimization model or initial surrogate model, use the set of sample points for curve fitting to construct the initial surrogate model.

[0346] 6. Determine the potential candidate points: That is, input the initial weight scaling parameter into the parameter selection equation, output the screened weight scaling parameter, and use the modified cyclic weight value evaluation sampling method to select the next batch (one) of valuable function evaluation points.

[0347] 7. Evaluate the potential candidate points: Call the flow field diagnosis optimization sub-function to evaluate the objective function of the screened weight scaling parameter, obtain the first sample point (the screened weight scaling parameter for which the objective function value has been obtained), and this process can use asynchronous parallel computing.

[0348] 8. Update the surrogate model: Add the first sample point to the set of sample points to obtain the second set of sample points, and use the second set of sample points to update the initial surrogate model; increment the adjustment times by 1; determine whether the termination condition (termination judgment) is reached, that is, whether the number of objective function evaluations of the screened weight scaling parameter reaches the preset evaluation times threshold. If the termination condition is reached, go to step 9; if the termination condition is not reached, repeat steps 6 to 8;

[0349] 9. Optimal injection-production control scheme: When the termination condition is reached, the current surrogate model is used as the target surrogate model. Determine the target weight update parameter according to the optimal surrogate model; solve the flow rate update weight equation using the target weight update parameter to obtain the target flow rate update weight; update the flow rate data using the target flow rate update weight, and output the optimal injection-production control scheme (or injection-production control scheme). If the numerical simulation results are saved throughout the entire iterative process of running the flow field diagnosis and optimization sub-function under the surrogate optimization framework, then at the end of the optimization, the injection-production optimization scheme corresponding to the optimal weight scaling parameter can be directly found from the saved simulation results.

[0350] Although this specification provides method operation steps or device structures as described in the following embodiments or attached Figure 11 figures, more or fewer operation steps or module units may be included in the method or device based on routine or non-creative labor. In steps or structures where there is no necessary causal relationship logically, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments of this specification or the attached figures. When the method or module structure is applied to an actual device, server, or terminal product, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or the attached figures (for example, in an environment with parallel processors or multi-threaded processing, or even including an environment of distributed processing and server clusters).

[0351] Based on the above optimization method for target well flow rate data, this specification also presents an embodiment of an optimization method device for target well flow rate data. As Figure 11 shown, the device may specifically include the following modules:

[0352] The screening weight scaling parameter output module 1101 can be used to input the initial weight scaling parameter into the parameter selection equation, and when the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir. The flow field diagnosis index data is obtained from the injection-production data of the wells in the reservoir. The parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to the injection-production optimization model. The screened weight scaling parameter is used to optimize the injection-production optimization model;

[0353] The judgment module 1102 can be used to judge whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition;

[0354] The primary optimization module 1103 can be used to determine a target weight scaling parameter based on the optimized target injection-production optimization model when the optimization termination condition is reached, so as to optimize the flow rate data of target wells in the reservoir based on the target weight scaling parameter and obtain the primarily optimized flow rate data.

[0355] The secondary optimization module 1104 can be used to determine whether it is necessary to perform secondary optimization processing on the primarily optimized flow rate data according to the type to which the target wells in the reservoir belong. When secondary optimization processing is required, the target flow rate data of the target wells in the reservoir is obtained.

[0356] In some embodiments, the flow field diagnosis index data in the above-mentioned screening weight scaling parameter output module 1101 may include at least one of the following: well pair injection efficiency, well pair production efficiency, injection well injection efficiency, and production well production efficiency.

[0357] In some embodiments, the above-mentioned screening weight scaling parameter output module 1101 can specifically be used to construct a comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter.

[0358] According to the comprehensive weight evaluation equation and the preset constraints of the initial weight scaling parameter, a parameter selection equation is constructed according to the following formula:

[0359]

[0360] where L B is the parameter selection equation; x is the initial weight scaling parameter; μ is the multiplier vector corresponding to the equality constraint in the preset constraint; λ is the multiplier vector corresponding to the inequality constraint condition in the preset constraint; σ is the penalty factor; f merit is the comprehensive weight evaluation equation; l is the number of equality constraint conditions in the preset constraint; m is the number of inequality constraint conditions in the preset constraint; h i (x) is the equality constraint in the preset constraint; g j (x) is the inequality constraint in the preset constraint.

[0361] In some embodiments, the above-mentioned screening weight scaling parameter output module 1101 can specifically also be used to, when the parameter selection termination condition of the parameter selection equation is not satisfied, re-obtain the initial weight scaling parameter and input it into the parameter selection equation, continuously update the screening weight scaling parameter until the parameter selection termination condition is satisfied, and obtain the updated screening weight scaling parameter; wherein, the parameter selection termination condition includes: the total error of the equality constraint in the preset constraint and the inequality constraint in the preset constraint is less than the preset error threshold and the comprehensive weight evaluation value is less than the preset comprehensive weight evaluation threshold, and the comprehensive weight evaluation value is determined according to the comprehensive weight evaluation equation.

[0362] In some embodiments, the above-mentioned determination module 1102 may specifically be configured to input the screening weight scaling parameter into the parameter selection equation when the optimization termination condition is not met, and output a second screening weight scaling parameter different from the screening weight scaling parameter when the parameter selection termination condition of the parameter selection equation is satisfied; calculate the second objective function value of the second screening weight scaling parameter, and update the data set where the screening weight scaling parameter is located based on the second screening weight scaling parameter and the second objective function value to obtain a second data set, where the data set where the screening weight scaling parameter is located includes the screening weight scaling parameter and the first objective function value of the screening weight scaling parameter; fit each screening weight scaling parameter and the corresponding objective function value in the second data set to obtain a second injection-production optimization model; determine again whether the optimization termination condition is met, and when the optimization termination condition is met, use the second injection-production optimization model as the target injection-production optimization model obtained through optimization.

[0363] In some embodiments, the above-mentioned determination module 1102 may specifically further be configured to construct an objective function according to the following formula:

[0364]

[0365] where NPV represents the net present value (objective function); u represents the number of the time step; N t represents the total number of time steps; N pro is the total number of effective production wells; r o (t u ) represents the oil price at time t u ; t u represents the termination time corresponding to the time step u; q o,j (t u ) represents the oil production of production well j at time t u ; r w (t u ) represents the water treatment cost at time t u ; q w,j (t u ) represents the water production of production well j at time t u ; r wi (t u ) represents the water injection cost at time t u ; N inj is the total number of effective injection wells; w i (t u ) represents the injection volume of injection well i at time t u ; t u-1 represents the termination time corresponding to u - 1; b represents the annual discount rate; t0 represents the optimization initial time;

[0366] Input the second screening weight scaling parameter into the objective function to output the second objective function value.

[0367] In some embodiments, the above-mentioned quadratic optimization module 1104 may specifically be configured to perform quadratic optimization processing on the once-optimized flow data according to the following formula:

[0368]

[0369]

[0370]

[0371] Where, is the injection volume of injection well i in the target flow data; is the updated value of the lost injection volume of injection well i; is the flow rate at injection well i in well pair (i, j) in the once-optimized flow data; is the liquid production volume of production well j in the target flow data; is the updated value of the injection volume of the aquifer into production well j; is the flow rate at production well j in well pair (i, j) in the once-optimized flow data; AQ i is the injection volume lost by injection well i due to injection into the aquifer; r AQ is the proportion of the reduction in the lost injection volume.

[0372] In some embodiments, after the above-mentioned quadratic optimization module 1104, it may specifically further be configured to determine whether there is a flow upper bound constraint on the target flow data;

[0373] When there is a flow upper bound constraint, perform scaling processing on the target flow data to obtain the second target flow data, and the scaling processing is performed according to the following formula:

[0374]

[0375]

[0376] Where, is the injection volume of injection well i in the second target flow data; N inj is the total number of effective injection wells; is the injection volume of injection well i in the target flow data; w sum is the target or upper bound of the total injection volume of all injection wells; is the liquid production volume of production well j in the second target flow data; N pro is the total number of effective production wells; is the liquid production volume of production well j in the target flow data; q sumis the total liquid production target or upper bound for all production wells.

[0377] It should be noted that the units, devices, modules, etc. illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. For the convenience of description, when describing the above devices, they are divided into various modules according to functions for separate description. Of course, when implementing this specification, the functions of each module can be implemented in the same or multiple software and / or hardware, or the modules implementing the same function can be realized by the combination of multiple sub-modules or sub-units, etc. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be in electrical, mechanical or other forms.

[0378] As can be seen from the above, an optimization device for target well flow rate data provided based on the embodiments of this specification can solve the drawbacks of existing optimization methods such as time-consuming, laborious, high cost, insufficient accuracy, low efficiency, and low intelligence level, and can contribute to the establishment of an automated, standardized, and process-based paradigm for intelligent closed-loop production optimization, and can significantly improve the accuracy and efficiency of reservoir development regulation. Specifically, it has the following advantages:

[0379] Redefine the injection-production efficiency at the single-well and well-pair levels from the injection end and production end, and the optimization process converges faster; according to the flow-line injection-production efficiency for the flow rate update criterion of injection-production flow rate optimization, only one flow-line simulation is required to achieve injection-production flow rate optimization, significantly saving the calculation cost;

[0380] Combine the surrogate optimization algorithm and the Lagrange multiplier method to construct a parameter selection equation, which can efficiently solve the time-consuming constrained optimization problem with various linear equality, linear inequality, non-linear equality, and non-linear inequality constraint conditions;

[0381] Combine the optimization strategy based on flow field diagnosis with the surrogate optimization algorithm, and use the proposed surrogate optimization algorithm combined with a parallel program to optimize the parameters in the flow rate update criterion, greatly reducing the number of numerical simulations required for simply using the surrogate optimization, and improving the accuracy, efficiency, and automation level of flow-line simulation optimization;

[0382] While reducing the computational cost of the high-confidence optimization model, it improves the optimization accuracy. Its time cost is comparable to that of the real-time optimization method. It comprehensively utilizes the physical driving characteristics of streamline simulation and the data-driven characteristics of surrogate optimization. Therefore, it can serve as a bridge between short-term real-time optimization and medium- and long-term optimization strategies, developing a comprehensive production optimization concept that combines geological reservoir understanding, short-term real-time optimization, and medium- and long-term optimization strategies under the intelligent closed-loop reservoir development and management framework, with physical-data dual driving. In terms of optimization efficiency, it can meet the needs of real-time adjustment, and in terms of optimization accuracy and reliability, it can also provide valuable guidance for the formulation of long-term development strategies.

[0383] The embodiment of this specification also provides an electronic device for an optimization method based on the above target well flow rate data, including a processor and a memory for storing processor-executable instructions. When specifically implemented, the processor can execute the following steps according to the instructions: Input the initial weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir. The flow field diagnosis index data is obtained from the injection-production data of the wells in the reservoir. The parameter selection equation is constructed based on the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to the injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model; Judge whether the number of times the screened weight scaling parameter optimizes the injection-production optimization model reaches the optimization termination condition; When the optimization termination condition is reached, determine the target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target well in the reservoir based on the target weight scaling parameter to obtain the flow rate data of one-time optimization; According to the type of the target well in the reservoir, judge whether it is necessary to perform secondary optimization processing on the flow rate data of one-time optimization. When secondary optimization processing is required, obtain the target flow rate data of the target well in the reservoir.

[0384] To be able to complete the above instructions more accurately, refer to Figure 12 As shown, the embodiment of this specification also provides another specific electronic device. Among them, the electronic device includes a network communication port 1201, a processor 1202, and a memory 1203. The above structures are connected by internal cables so that each structure can perform specific data interactions.

[0385] Among them, the network communication port 1201 can specifically be used to input the initial weight scaling parameter into the parameter selection equation;

[0386] The processor 1202 can be specifically configured to output a screening weight scaling parameter when a parameter selection termination condition of a parameter selection equation is satisfied. The initial weight scaling parameter is determined according to the flow field diagnosis index data of wells in the reservoir. The flow field diagnosis index data is obtained based on the injection-production data of wells in the reservoir. The parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to an injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screening weight scaling parameter is used to optimize the injection-production optimization model; determine whether the number of times of optimizing the injection-production optimization model with the screening weight scaling parameter reaches an optimization termination condition; when the optimization termination condition is reached, determine a target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of target wells in the reservoir based on the target weight scaling parameter to obtain the flow rate data after one optimization; according to the type to which the target wells in the reservoir belong, determine whether secondary optimization processing needs to be performed on the flow rate data after one optimization, and when secondary optimization processing is required, obtain the target flow rate data of the target wells in the reservoir.

[0387] The memory 1203 can be specifically configured to store corresponding instruction programs.

[0388] In this embodiment, the network communication port 1201 can be bound to different communication protocols, so as to send or receive different data. For example, the network communication port can be a virtual port responsible for web data communication, or a port responsible for FTP data communication, or a port responsible for mail data communication. In addition, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM, CDMA, etc.; it can also be a Wifi chip; it can also be a Bluetooth chip.

[0389] In this embodiment, the processor 1202 can be implemented in any suitable manner. For example, the processor can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc. This specification does not make any limitations.

[0390] In this embodiment, the memory 1203 may include multiple levels. In a digital system, anything that can store binary data can be a memory; in an integrated circuit, a circuit with a storage function without a physical form is also called a memory, such as RAM, FIFO, etc.; in a system, a storage device with a physical form is also called a memory, such as a memory module, a TF card, etc.

[0391] The embodiment of this specification also provides a computer storage medium for an optimization method based on the above target well flow rate data. The computer storage medium stores computer program instructions, which when executed implement: inputting an initial weight scaling parameter into a parameter selection equation, and when the parameter selection termination condition of the parameter selection equation is satisfied, outputting a screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir, the flow field diagnosis index data is obtained from the injection-production data of the wells in the reservoir, the parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, the response value of the initial weight scaling parameter is determined according to an injection-production optimization model, the injection-production optimization model is determined according to the initial weight scaling parameter, and the screened weight scaling parameter is used to optimize the injection-production optimization model; determining whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition; when the optimization termination condition is reached, determining a target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target well in the reservoir based on the target weight scaling parameter to obtain the flow rate data after one optimization; according to the type to which the target well in the reservoir belongs, determining whether secondary optimization processing needs to be performed on the flow rate data after one optimization, and when secondary optimization processing needs to be performed, obtaining the target flow rate data of the target well in the reservoir.

[0392] In this embodiment, the above storage medium includes but is not limited to a random access memory (RAM), a read-only memory (ROM), a cache, a hard disk drive (HDD), or a memory card. The memory can be used to store computer program instructions. The network communication unit can be set according to the standards specified by the communication protocol and is used for an interface for network connection communication.

[0393] In this embodiment, the functions and effects specifically implemented by the program instructions stored in the computer storage medium can be explained by comparison with other embodiments and will not be elaborated here.

[0394] The embodiments of this specification also provide a computer program product for an effectiveness testing method based on the above site data, including a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the following steps:

[0395] Although this specification provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative means. The step sequences listed in the embodiments are only one way among numerous step execution sequences and do not represent the only execution sequence. When an actual device or client product executes, it can be executed in the method sequence shown in the embodiments or the drawings or in parallel (for example, in an environment with parallel processors or multi-threaded processing, or even in a distributed data processing environment). The term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, product or device including a series of elements not only includes those elements but also includes other elements not explicitly listed, or also includes elements inherent to such a process, method, product or device. Without further limitation, the presence of additional identical or equivalent elements in a process, method, product or device including the said elements is not excluded. Words such as first, second, etc. are used to denote names and do not denote any particular order.

[0396] Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, the method steps can be logically programmed to enable the controller to implement the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Therefore, such a controller can be regarded as a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0397] This specification can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types. This specification can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0398] From the descriptions of the above embodiments, those skilled in the art can clearly understand that this specification can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of this specification can essentially be embodied in the form of a software product, and this computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to enable a computer device (which can be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this specification.

[0399] The various embodiments in this specification are described in a progressive manner. For the same or similar parts among the various embodiments, reference can be made to each other, and the key points of each embodiment are the differences from other embodiments. This specification can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on.

[0400] Although this specification is depicted through embodiments, those of ordinary skill in the art know that this specification has many variations without departing from the spirit of this specification, and it is hoped that the appended claims will include these variations without departing from the spirit of this specification.

Claims

1. An optimization method for target well flow rate data, characterized in that Including: Input the initial weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the screened weight scaling parameter. The initial weight scaling parameter is determined according to the flow field diagnosis index data of wells in the reservoir. The flow field diagnosis index data is obtained based on the injection-production data of wells in the reservoir. The parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter. The response value of the initial weight scaling parameter is determined according to the injection-production optimization model. The injection-production optimization model is determined according to the initial weight scaling parameter. The screened weight scaling parameter is used to optimize the injection-production optimization model; Judge whether the number of times of optimizing the injection-production optimization model with the screened weight scaling parameter reaches the optimization termination condition; When the optimization termination condition is reached, determine the target weight scaling parameter based on the optimized target injection-production optimization model, so as to optimize the flow rate data of the target well in the reservoir based on the target weight scaling parameter to obtain the flow rate data after one optimization; According to the type of the target well in the reservoir, judge whether it is necessary to perform secondary optimization processing on the flow rate data after one optimization. When secondary optimization processing is required, obtain the target flow rate data of the target well in the reservoir.

2. The method according to claim 1, wherein The flow field diagnosis index data includes at least one of the following: injection efficiency of well pairs, production efficiency of well pairs, injection efficiency of injection wells, and production efficiency of production wells.

3. The method according to claim 1, wherein The parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, including: Construct a comprehensive weight evaluation equation according to the response value of the initial weight scaling parameter; Construct a parameter selection equation according to the comprehensive weight evaluation equation and the preset constraints of the initial weight scaling parameter according to the following formula: Among them, L B is a selection parameter equation; x is the initial weight scaling parameter; μ is the multiplier vector corresponding to the equality constraint in the preset constraint; λ is the multiplier vector corresponding to the inequality constraint condition in the preset constraint; σ is the penalty factor; f merit is the comprehensive weight evaluation equation; l is the number of equality constraint conditions in the preset constraint; m is the number of inequality constraint conditions in the preset constraint; h i (x) is the equality constraint in the preset constraint; g j (x) is the inequality constraint in the preset constraint.

4. The method according to claim 1, wherein The method further includes: When the parameter selection termination condition of the parameter selection equation is not satisfied, re-obtain the initial weight scaling parameter and input it into the parameter selection equation, and continuously update the screened weight scaling parameter until the parameter selection termination condition is satisfied to obtain the updated screened weight scaling parameter; Among them, the parameter selection termination condition includes: the sum of the errors of the equality constraint in the preset constraint and the inequality constraint in the preset constraint is less than the preset error threshold and the comprehensive weight evaluation value is less than the preset comprehensive weight evaluation threshold, and the comprehensive weight evaluation value is determined according to the comprehensive weight evaluation equation.

5. The method according to claim 1, wherein The method further includes: When the optimization termination condition is not reached, input the screened weight scaling parameter into the parameter selection equation. When the parameter selection termination condition of the parameter selection equation is satisfied, output the second screened weight scaling parameter, and the second screened weight scaling parameter is different from the screened weight scaling parameter; Obtain the second objective function value of the second screened weight scaling parameter, and update the data set where the screened weight scaling parameter is located based on the second screened weight scaling parameter and the second objective function value to obtain the second data set. The data set where the screened weight scaling parameter is located includes the screened weight scaling parameter and the first objective function value of the screened weight scaling parameter; Fit the screened weight scaling parameters and the corresponding objective function values in the second data set to obtain the second injection-production optimization model; Judge again whether the optimization termination condition is reached. When the optimization termination condition is reached, use the second injection-production optimization model as the target injection-production optimization model obtained through optimization.

6. The method according to claim 5, wherein The obtaining of the second objective function value of the second screening weight scaling parameter includes: Construct an objective function according to the following formula: Among them, NPV represents the net present value (objective function); u represents the number of the time step; N t represents the total number of time steps; N pro is the total number of effective production wells; r o (t u ) represents the oil price at time t u ; t u represents the termination time corresponding to the time step u; q o,j (t u ) represents the oil production of production well j at time t u ; r w (t u ) represents the water treatment cost at time t u ; q w,j (t u ) represents the water production of production well j at time t u ; r wi (t u ) represents the injection cost at time t u ; N inj is the total number of effective injection wells; w i (t u ) represents the injection volume of injection well i at time t u ; t u-1 represents the termination time corresponding to the time step u - 1; b represents the annual discount rate; t0 represents the initial optimization time; Input the second screening weight scaling parameter into the objective function and output the second objective function value.

7. The method according to claim 1, characterized in that When secondary optimization processing is required, the obtaining of the target flow rate data of the target well in the reservoir includes: Perform secondary optimization processing on the flow rate data of the primary optimization according to the following formula: wherein, is the injection volume of injection well i in the target flow rate data; is the updated value of the lost injection volume of injection well i; is the flow rate at injection well i in well pair (i, j) in the first optimized flow rate data; is the liquid production volume of production well j in the target flow rate data; is the updated value of the injection volume of the aquifer into production well j; is the flow rate at production well j in well pair (i, j) in the first optimized flow rate data; AQ i is the injection volume lost by injection well i due to injection into the aquifer; r AQ is the proportion of the reduction in the lost injection volume.

8. The method according to claim 1, wherein The method further includes: Judge whether there is a flow rate upper bound constraint on the target flow rate data; When there is a flow rate upper bound constraint, perform scaling processing on the target flow rate data to obtain the second target flow rate data, and the scaling processing is performed according to the following formula: Among them, is the injection volume of injection well i in the second target flow rate data; N inj is the total number of effective injection wells; is the injection volume of injection well i in the target flow rate data; w sum is the target or upper bound of the total injection volume of all injection wells; is the liquid production volume of production well j in the second target flow rate data; N pro is the total number of effective production wells; is the liquid production volume of production well j in the target flow rate data; q sum is the target or upper bound of the total liquid production volume of all production wells.

9. An optimization device for target well flow rate data, characterized in that, including: A screening weight scaling parameter output module, configured to input an initial weight scaling parameter into a parameter selection equation, and output a screening weight scaling parameter when the parameter selection termination condition of the parameter selection equation is satisfied. The initial weight scaling parameter is determined according to the flow field diagnosis index data of the wells in the reservoir, the flow field diagnosis index data is obtained according to the injection-production data of the wells in the reservoir, the parameter selection equation is constructed according to the preset constraints of the initial weight scaling parameter and the response value of the initial weight scaling parameter, the response value of the initial weight scaling parameter is determined according to the injection-production optimization model, and the screening weight scaling parameter is used to optimize the injection-production optimization model; A judgment module, configured to judge whether the number of times of optimizing the injection-production optimization model by the screening weight scaling parameter reaches the optimization termination condition; A primary optimization module, configured to determine a target weight scaling parameter based on the optimized target injection-production optimization model when the optimization termination condition is reached, so as to optimize the flow rate data of the target well in the reservoir based on the target weight scaling parameter to obtain the flow rate data of the primary optimization; A secondary optimization module, configured to judge whether it is necessary to perform secondary optimization processing on the flow rate data of the primary optimization according to the type of the target well in the reservoir, and obtain the target flow rate data of the target well in the reservoir when secondary optimization processing is required.

10. A computer storage medium, characterized in that, The computer storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images