Shale gas complex fracture network characteristic parameter intelligent inversion method and system

Abstract

The present application provides a shale gas complex fracture network characteristic parameter intelligent inversion method and system, the method adopts embedded discrete fracture model to establish the complex fracture network shale gas fracturing productivity model of shale reservoir; after determining the parameter value space to be inverted, the basic calculation examples are obtained based on random sampling, and parameter sensitivity analysis is carried out; the intelligent agent model is used to train the machine learning model with the basic calculation examples as the training samples, and the fracturing productivity prediction model is formed; the target example parameters meeting the error condition are optimized by using the set objective function based on the calculation results of the fracturing productivity prediction model; and then the target example parameters are brought into the complex fracture network shale gas fracturing productivity model to invert the target fracture network characteristic parameter results. The scheme can overcome the problems of time and cost consumption and limited applicability of the prior art, and can realize fast and reliable fitting by combining reservoir numerical simulation, intelligent machine learning model and optimization algorithm, and improve the shale gas complex fracture network characteristic parameter inversion efficiency.

Description

Technical Field

[0001] This invention relates to the field of shale gas exploration and development technology, and in particular to an intelligent inversion method and system for characteristic parameters of complex fracture networks in shale gas. Background Technology

[0002] Shale gas is a significant new area in natural gas exploration and development. Related companies are actively promoting the commercialization of the shale gas industry, and several shale gas producing areas have been established. Among them, the Fuling shale gas field, as the first commercially developed shale gas reservoir in the field, has achieved good development results. Due to the low porosity, low permeability, and high adsorption characteristics of shale reservoirs, hydraulic fracturing is the core technology for reservoir stimulation in shale gas extraction. The current expected technical goal is to form a "multi-layered, three-dimensional, large-well-cluster, factory-like" extraction approach. However, the high-investment, large-scale fracturing model still cannot meet the high-efficiency and economical requirements of shale gas development operations, necessitating research into intelligent and precise fracturing technology. Due to the development of shale reservoir bedding and natural fractures, the fracture network morphology after fracturing is complex, the effective supporting fractures are unevenly distributed, and they exhibit cross-scale characteristics, making fracture description and simulation quite challenging.

[0003] Traditional methods for inverting fracture network characteristic parameters based on historical data require technicians to manually adjust various reservoir characteristic parameters to match simulated production as closely as possible to historical real data. Then, the fracture network characteristic parameters are fitted and inverted based on the production data. This approach suffers from insufficient computational efficiency in practical applications. Existing methods for inverting characteristic parameters of complex shale gas fracture networks mainly include fracture monitoring technology, well test analysis methods, and dynamic production data analysis methods. Fracture monitoring technologies mainly include microseismic monitoring, wide-area electromagnetic methods, and well coring observation; however, fracture monitoring technologies are generally expensive and can only obtain post-compression fracture morphology. Well test analysis methods are mostly based on single-phase hydrocarbons and have strong ambiguity, making them unsuitable for gas-water two-phase flow in complex shale gas fracture networks. Dynamic production data analysis uses production data to fit and invert fracture network characteristic parameters and reservoir parameters. Historical fitting based on reservoir numerical simulation is the main analytical method, divided into three stages: the first stage is manual historical fitting based on reservoir numerical simulation; the second stage is automatic historical fitting based on optimization algorithms and reservoir numerical models; and the third stage is automatic historical fitting based on intelligent reservoir proxy models. The calculation process is cumbersome, has high requirements for data, and is not timely.

[0004] The information disclosed in the background section of this invention is intended only to enhance the understanding of the general background of this invention, and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an intelligent inversion method for characteristic parameters of complex fracture networks in shale gas. This method overcomes the time-consuming, costly, and limited applicability issues of existing technologies. By combining reservoir numerical simulation, intelligent machine learning models, and optimization algorithms, it achieves rapid and reliable fitting, improving the efficiency of inversion of characteristic parameters of complex fracture networks in shale gas. The method establishes a shale gas fracturing production capacity model for complex fracture networks in shale reservoirs using an embedded discrete fracture model. After determining the value space of the parameters to be inverted, basic calculation examples are obtained based on random sampling, and parameter sensitivity analysis is performed. Using the basic calculation examples as training samples, an intelligent surrogate model is used to train a machine learning model, forming a fracturing production capacity prediction model. Based on the calculation results of the fracturing production capacity prediction model, a target calculation example parameter that meets the error conditions is selected using a set objective function. Finally, the target calculation example parameter is substituted into the complex fracture network shale gas fracturing production capacity model to obtain the target fracture network characteristic parameter results. Preferably, in one embodiment, the method includes:

[0006] Step S100: Establish a shale gas fracturing production capacity model for complex fractures in shale reservoirs using an embedded discrete fracture model;

[0007] Step S200: Determine the value space of the parameters to be inverted, obtain basic calculation examples based on random sampling, and perform parameter sensitivity analysis;

[0008] Step S300: Using basic calculation examples as training samples, a machine learning model is trained using an intelligent agent model to form a fracturing capacity prediction model;

[0009] Step S400: Optimize the target case parameters that meet the error conditions based on the calculation results of the fracturing capacity prediction model using the set objective function;

[0010] Step S500: Substitute the selected target example parameters into the complex fracture web rock gas fracturing production capacity model inversion to obtain the target fracture network characteristic parameter results.

[0011] In an optional embodiment, in step S100, the analyzed reservoir is partitioned based on an embedded discrete fracture model. A gas-water two-phase flow production capacity model is established, taking into account various shale reservoir characteristics such as anti-condensation, micro- and nanoscale nonlinear seepage, reservoir stress sensitivity, and desorption. This model serves as a complex fracture network shale gas fracturing production capacity model, and the complex fracture network of the reservoir is explicitly characterized and calculated.

[0012] Furthermore, in one embodiment, the fluid flow exchange between physically connected but non-adjacent mesh cells in a complex fracture network is calculated, including fluid exchange between a fracture and its matrix cell, fluid exchange between two intersecting fractures, and fluid exchange between a single fracture and adjacent matrix cells.

[0013] In one embodiment, in step S200, the geological engineering data and production data of the set well are analyzed to determine the possible value range of the parameter to be fitted, thus forming the value space of the parameter to be inverted.

[0014] Preferably, in one embodiment, during the parameter sensitivity analysis, the number of effective connected cracks and the ratio of effective crack volume are introduced to quantitatively evaluate the complexity of the crack network.

[0015] In an optional embodiment, in step S300, the calculated basic examples are used as learning samples for the intelligent agent model and fed into the random forest intelligent agent model to train the machine learning model; based on the computational data analysis of the machine learning model, data index points are determined according to the production curve characteristics and computational results; when the production prediction accuracy of the obtained data point index numbers is greater than 95%, it is determined that the training requirements are met and a fracturing capacity prediction model is formed.

[0016] Furthermore, in one embodiment, in step S400, the following objective function is established:

[0017] minF(X)=[f1(Y1),f2(Y2),…,f n (Y n )]

[0018] Y n =[y1,y2,…,y m ] T

[0019] y = f(X)

[0020] X = [x1, x2, ..., x k ] T

[0021]

[0022] In the formula, F(X) represents the objective function; f1(Y1), f2(Y2), ..., f n (Y n Y represents the average absolute error between the calculated and actual values ​​of all production days under a certain fitted target. n Let represent the set of calculated results for all production times of the nth fitted target, where y1, y2, ..., y m These represent the calculation results from day 1 to day m; X represents the set of independent variables; g i (x) represents the boundary conditions; Z represents the parameter space.

[0023] In one embodiment, in step S400, an optimization algorithm is used to solve for the maximum and minimum values ​​of the objective function to obtain a combination of characteristic parameters of the complex crack network; the optimization algorithm uses Markov chain-Monte Carlo sampling, evolutionary algorithm or genetic algorithm.

[0024] Preferably, in one embodiment, in step S500, the values ​​of several sets of target calculation parameters with the smallest error are substituted into the shale gas fracturing production capacity model for recalculation, and then matched and fitted with historical production again. The values ​​of various fracture network characteristic parameters corresponding to the best fitting effect are the target inversion results.

[0025] Based on other aspects of the methods described in any one or more of the foregoing embodiments, the present invention also provides a storage medium storing program code that can implement the methods described in any one or more of the foregoing embodiments.

[0026] Based on the application aspects of the methods described in any one or more of the above embodiments, the present invention provides an intelligent inversion system for characteristic parameters of complex fracture networks in shale gas, which executes the methods described in any one or more of the above embodiments.

[0027] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0028] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0029] Figure 1 This is a flowchart illustrating the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas provided in an embodiment of the present invention.

[0030] Figure 2 This is a schematic diagram of the material SN curve of the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas provided in this embodiment of the invention;

[0031] Figure 3 This is a comparative diagram of gas production prediction data using the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas provided in this embodiment of the invention.

[0032] Figure 4 This is a schematic diagram of the intelligent inversion system for characteristic parameters of complex fracture networks in shale gas provided in another embodiment of the present invention. Detailed Implementation

[0033] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings and examples. Those skilled in the art will then fully understand how the present invention uses technical means to solve technical problems and achieve technical effects, and will be able to implement the present invention specifically based on the above-described implementation process. It should be noted that, as long as there is no conflict, the various embodiments and features of the present invention can be combined with each other, and the resulting technical solutions are all within the protection scope of the present invention.

[0034] Although the flowchart describes the operations as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. The order of the operations can be rearranged. A process can terminate when its operation is complete, but it may also have additional steps not included in the diagram. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0035] Computer equipment includes user equipment and network equipment. User equipment or clients include, but are not limited to, computers, smartphones, and PDAs (Personal Digital Assistants); network equipment includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers. Computer equipment can operate independently to implement this invention, or it can connect to a network and implement this invention through interaction with other computer devices within the network. The network in which the computer equipment resides includes, but is not limited to, the Internet, wide area networks (WANs), metropolitan area networks (MANs), local area networks (LANs), and VPN networks.

[0036] The terms “first,” “second,” etc., may be used herein to describe various units, but these units should not be limited by these terms; they are used merely to distinguish one unit from another. The term “and / or” as used herein includes any and all combinations of one or more of the associated listed items. When a unit is referred to as “connected” or “coupled” to another unit, it may be directly connected or coupled to said other unit, or there may be intermediate units present.

[0037] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments. Unless the context clearly indicates otherwise, the singular forms “a” and “an” as used herein are also intended to include the plural. It should also be understood that the terms “comprising” and / or “including” as used herein specify the presence of the stated features, integers, steps, operations, units, and / or components, without excluding the presence or addition of one or more other features, integers, steps, operations, units, components, and / or combinations thereof.

[0038] Shale gas is one of the important new areas of natural gas exploration and development. Related companies are actively promoting the commercial development of the shale gas industry, and several shale gas production areas have been formed. Among them, the relevant blocks of the Fuling Shale Gas Field, as the first commercially developed shale gas reservoir in the field, have achieved good development results.

[0039] Due to the low porosity, low permeability, and high adsorption characteristics of shale reservoirs, hydraulic fracturing is the core technology for reservoir stimulation in shale gas extraction. The current expected technical goal is to develop a "multi-layered, three-dimensional, large-well-cluster, factory-like" extraction approach. However, the high-investment, large-scale fracturing model still cannot meet the high-efficiency and economical requirements of shale gas development operations, necessitating research into intelligent and precise fracturing technologies. Because of the well-developed bedding and natural fractures in shale reservoirs, the fracture network morphology after fracturing is complex, the effective supporting fractures are unevenly distributed, and they exhibit cross-scale characteristics, making fracture description and simulation quite challenging.

[0040] Traditional methods for inverting fracture network characteristic parameters based on historical data require technicians to manually adjust various reservoir characteristic parameters to match simulated production as closely as possible to historical real data. Then, the fracture network characteristic parameters are fitted and inverted based on the production data. This approach suffers from insufficient computational efficiency in practical applications. Existing methods for inverting characteristic parameters of complex shale gas fracture networks mainly include fracture monitoring technology, well test analysis methods, and dynamic production data analysis methods. Fracture monitoring technologies mainly include microseismic monitoring, wide-area electromagnetic methods, and well coring observation; however, fracture monitoring technologies are generally expensive and can only obtain post-compression fracture morphology. Well test analysis methods are mostly based on single-phase hydrocarbons and have strong ambiguity, making them unsuitable for gas-water two-phase flow in complex shale gas fracture networks. Dynamic production data analysis uses production data to fit and invert fracture network characteristic parameters and reservoir parameters. Historical fitting based on reservoir numerical simulation is the main analytical method, divided into three stages: the first stage is manual historical fitting based on reservoir numerical simulation; the second stage is automatic historical fitting based on optimization algorithms and reservoir numerical models; and the third stage is automatic historical fitting based on intelligent reservoir proxy models. The calculation process is cumbersome, has high requirements for data, and is not timely.

[0041] Therefore, in order to more accurately characterize fracture parameters and better guide the design of fracturing schemes, it is necessary to carry out intelligent inversion research on the characteristic parameters of complex fracture networks in shale gas.

[0042] The researchers of this invention considered that using a combination of numerical simulation, artificial intelligence models, and optimization algorithms can significantly accelerate the fitting process and improve the efficiency of inverting complex fracture network characteristic parameters. Therefore, using intelligent algorithms to invert complex fracture network parameters is expected to provide support for achieving integrated shale oil geological engineering, cost reduction and efficiency improvement in development, and promoting the construction of smart oil and gas fields. This patent combines reservoir numerical simulation, artificial intelligence models, and optimization algorithms to accelerate the historical fitting process and improve the efficiency of inverting complex fracture network characteristic parameters in shale gas.

[0043] To address the aforementioned issues, this invention provides a data-driven, physically constrained intelligent inversion method for complex fracture network parameters in shale gas fracturing horizontal wells. Targeting different reservoir and fluid characteristics, it combines artificial intelligence technology to achieve intelligent inversion of complex fracture network characteristic parameters, improving inversion rate while ensuring reliability. A gas-water two-phase flow model suitable for shale gas reservoirs is established. Based on an embedded fracture network model, a complex fracture shale gas fracturing production capacity model is built. Through analysis of geological, engineering, and production data, the parameter value space for inversion is determined. Basic calculation examples are generated using random sampling, and the results are calculated and saved in the numerical model to analyze parameter sensitivity. The calculated basic examples are used as learning samples for the intelligent proxy model, with reservoir physical parameters and fracture network characteristic parameters as input variables, and gas production, water production, and bottom hole flowing pressure as output variables. An objective function is established, and an optimization algorithm is used to solve for the objective function value. The parameter values ​​of the n sets of examples with the smallest errors are re-introduced into the production capacity model for calculation, and the parameter values ​​of the example with the best fitting effect are the optimal inversion results. This invention combines an optimization algorithm, resulting in a fast inversion speed and enabling intelligent inversion of characteristic parameters of different pressure fracture networks.

[0044] The following describes the detailed flow of the method according to an embodiment of the present invention with reference to the accompanying drawings, the steps of which can be executed in a computer system containing, for example, a set of computer-executable instructions. Although the logical order of the steps is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0045] Example 1

[0046] Figure 1 This diagram illustrates the flow chart of the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas provided in Embodiment 1 of the present invention. Figure 1 As can be seen, the method includes the following steps.

[0047] Step S100: Establish a shale gas fracturing production capacity model for complex fractures in shale reservoirs using an embedded discrete fracture model;

[0048] Step S200: Determine the value space of the parameters to be inverted, obtain basic calculation examples based on random sampling, and perform parameter sensitivity analysis;

[0049] Step S300: Using basic calculation examples as training samples, a machine learning model is trained using an intelligent agent model to form a fracturing capacity prediction model;

[0050] Step S400: Optimize the target case parameters that meet the error conditions based on the calculation results of the fracturing capacity prediction model using the set objective function;

[0051] Step S500: Substitute the selected target example parameters into the complex fracture web rock gas fracturing production capacity model inversion to obtain the target fracture network characteristic parameter results.

[0052] The embodiments of this invention combine reservoir numerical simulation, artificial intelligence models and optimization algorithms to accelerate the historical fitting process and use intelligent algorithms to invert complex fracture network parameters. This is expected to provide support for achieving integrated shale oil geology and engineering, cost reduction and efficiency improvement in development, and promoting the construction of smart oil and gas fields.

[0053] In a preferred embodiment, in step S100, the reservoir under analysis is partitioned based on an embedded discrete fracture model to explicitly characterize and calculate the complex fracture network of the reservoir. A complex fracture network gas-water two-phase flow productivity model is established, taking into account various shale reservoir characteristics such as anti-condensation, micro-nano scale nonlinear seepage, reservoir stress sensitivity, and desorption.

[0054] Specifically, in step S100, when the analyzed reservoir is characterized by an embedded discrete fracture model, the matrix domain of the reservoir is first discretized into a structured grid, i.e., divided into multiple matrix elements. Then, fractures are embedded into the matrix domain, and the fractures are divided into multiple fracture elements by the boundaries of the matrix elements. At this point, the entire computational domain consists of matrix elements, fracture elements, and empty elements (grids without fractures or matrix). The fluid flow exchange between physically connected but non-adjacent grid elements is computed, including fluid exchange between fractures and their matrix elements, fluid exchange between two intersecting fractures, and fluid exchange between adjacent matrix elements for a single fracture.

[0055] A complex fractured network gas-water two-phase flow productivity model was established, considering various shale reservoir characteristics such as anti-condensation, micro- and nano-scale nonlinear seepage, reservoir stress sensitivity, and desorption. Specifically, shale reservoir characteristics such as anti-condensation, micro- and nano-scale nonlinear seepage, reservoir stress sensitivity, and desorption were implemented by setting parameters according to corresponding formulas.

[0056] In the process of partitioning the analyzed reservoir based on the embedded discrete fracture model, the complex fracture network of the reservoir is explicitly characterized and calculated using the embedded discrete fracture model. In an optional embodiment, this is achieved by the following operation: firstly, the matrix domain of the analyzed reservoir is discretized into a structured mesh, that is, divided into multiple matrix units.

[0057] The cracks are then embedded into the matrix domain, and the cracks are divided into multiple crack elements by the boundaries of the matrix elements. At this point, the entire computational domain consists of matrix elements, crack elements, and empty elements (mesh without cracks or matrix).

[0058] To calculate the fluid flow exchange between physically connected but non-adjacent grid cells in the computational domain, non-neighboring connections (NNCs) are introduced based on the type of grid cell connectivity. The original connection method on the reservoir simulator is replaced by the newly defined NNCs, thereby enabling explicit characterization and calculation of the complex fracture network of the reservoir.

[0059] In practical applications, different reservoir objects may need to be analyzed. When the characteristics of the reservoirs being analyzed are different, it is necessary to establish corresponding reservoir numerical simulation models based on different reservoir characteristics, such as reservoir thickness, burial depth, fluid properties, etc. However, the implementation process is the same, including matrix domain discretization and fracture element partitioning, and establishing connections between elements.

[0060] Further execute step S200 to determine the numerical space used for inversion, and obtain the basic calculation example based on random sampling.

[0061] In an optional embodiment, in step S200, the value space of the inversion parameters is determined, and random sampling is used to calculate the production capacity model, generate basic calculation examples, and perform parameter sensitivity analysis.

[0062] Organize target well data, including production data and geological engineering data, such as pore saturation, Langmuir volume and pressure, production curves, etc.

[0063] By analyzing the acquired and organized data, the possible values ​​and ranges of the parameters to be fitted are determined, and a value space for the parameters to be inverted is generated. The parameters to be fitted refer to the parameters that need to be input into the model but are not yet certain, in addition to existing geological data, logging data, and drilling and completion data. The possible values ​​and ranges of the parameters to be fitted include, for example, the hydraulic / natural fracture half-length and the hydraulic / natural fracture conductivity. In an optional embodiment, the value range is determined and set based on data from adjacent wells, block data, etc.

[0064] When different parameters are to be fitted, the value space should correspond to the parameters to be fitted. The value space of the parameters to be inverted consists of the value ranges of different parameters, such as the half-length of hydraulic fractures ranging from 100m to 200m, and the number of natural fractures ranging from 1000 to 2000. By random sampling, n sets of parameter combinations are randomly determined from the parameter space, such as a half-length of hydraulic fractures of 150m and a number of natural fractures of 1000.

[0065] Considering that the n parameter combinations selected in the current parameter space may not necessarily contain the final inversion results; this embodiment of the invention uses the production results obtained by calculating the production capacity of the n randomly determined parameter combinations through the production capacity model as basic examples, which are used together to train the intelligent agent model; n parameter combinations are randomly determined from the parameter space through random sampling and substituted into the gas-water two-phase flow production capacity numerical model for calculation; the n calculated examples are used as basic examples; in practical applications, such as Figure 2 As shown, the productivity calculation of a complex fractured network gas-water two-phase flow model incorporating shale reservoir characteristics is presented. The n calculated examples serve as the base examples. Examples of the calculation results can be found in [link to example]. Figure 2 Chinese data.

[0066] A parameter sensitivity analysis was performed; two quantitative evaluation parameters for the complexity of the fracture network were introduced: the number of effective connected fractures and the ratio of effective fracture volume. The parameter sensitivity analysis was conducted to provide sensitive parameters for step 300.

[0067] During the parameter sensitivity analysis, the objective function of the quantitative evaluation parameter of the seam mesh complexity is implemented in stages for computational analysis.

[0068] Specifically, the effective connected fracture number (ECFN) is calculated as follows: The effective fracture network of a coal seam consists of hydraulic fractures and natural fractures (surface cleavage and end cleavage in the coal seam) that intersect with the hydraulic fractures. Mapping the three-dimensional fracture network model onto a two-dimensional plane, and representing fractures with line segments, determines whether fractures are connected, which is equivalent to determining whether line segments intersect. The Quick Rejection Test and Straddle Test are computer graphics-based methods for determining line segment intersection. Combining these two methods allows for the rapid determination of whether two line segments intersect, i.e., whether two fractures are connected, thereby calculating the effective connected fracture number.

[0069] Effective fracture volume ratio (EFVR): The ratio of the effective fracture network volume to the coal seam volume. A higher EFVR indicates better fracture connectivity in the coal seam. The formula for calculating EFVR is:

[0070]

[0071] In the formula, V EFVR The effective crack volume ratio; V FFV For the volume of the connected fracture network, m 3 V RES It is the coal seam volume within the computational domain, in m. 3 .

[0072] In practical applications, the sensitivity requirements are determined by analyzing the importance of the surrogate model features and combining this with the set numerical conditions.

[0073] Furthermore, in step S300, a machine learning model is trained using a smart agent model with basic computational examples as training samples.

[0074] In step S300, the calculated basic examples are used as learning samples for the intelligent agent model and fed into the intelligent agent model to train the machine learning model.

[0075] The intelligent agent model can be constructed using a random forest model; when the output prediction accuracy of the data point index numbers obtained from the training operation of the random forest agent model is greater than 95%, it is determined that the training requirements are met.

[0076] During the training process, reservoir physical properties and fracture network characteristics that have a significant impact on production results are used as input variables, while gas production, water production, and bottom hole flowing pressure are used as output variables. This way, the well-trained machine learning model can greatly improve the efficiency of production simulation.

[0077] In the basic example, n sets of parameter combinations are obtained by random sampling. Each set of parameter combinations corresponds to a set of input parameter values ​​and a set of output parameter values.

[0078] Production results refer to information such as gas production, water production, and bottom hole flowing pressure contained in the basic calculation example. They are the results obtained by calculating the production capacity model using n sets of parameters, and together with the n sets of parameters, they form the basic calculation example.

[0079] After the random forest surrogate model is trained, data index points are determined based on the characteristics of the production curve and the training results to reduce the complexity of the surrogate model and ensure the accuracy and efficiency of subsequent optimization algorithms.

[0080] In an optional embodiment, the output data uses production data from a specific date rather than production data from every day within the production cycle as the data point index number. The selection principle for the data point index number is: to select as few data point index numbers as possible while accurately reflecting the trend of output changes; and at the same time, the output prediction accuracy at that point in time is high. Figure 3 The accuracy of the gas production model was demonstrated. A well-trained machine learning model, used as a capacity prediction model, can greatly improve the efficiency of production simulation.

[0081] Further, step S400 is executed to establish an objective function and, based on an optimization algorithm, to select the parameter values ​​of group a of the calculation examples with the smallest error; in an optional embodiment, the objective function adopts the average absolute error between actual production data and simulation results.

[0082] Step S400: An optimization algorithm is used to solve for the maximum and minimum values ​​of the objective function, obtaining the combination of characteristic parameters for the complex fracture network. Markov chain-Monte Carlo sampling, evolutionary algorithms, and genetic algorithms are used to solve for the maximum and minimum values ​​of the objective function, thereby calculating possible combinations of characteristic parameters for the complex fracture network, and selecting the optimal parameter values ​​from the set of examples (a) with the smallest error. The complete mathematical model for the inversion of the characteristic parameters of the complex fracture network can be expressed as:

[0083]

[0084] In the formula, F(X) represents the objective function; f1(Y1), f2(Y2), ..., f n (Y n Y represents the average absolute error between the calculated and actual values ​​of all production days under the nth fitted target. The fitted target can be daily gas (water) production, cumulative gas (water) production, bottom hole flowing pressure, etc. n Let represent the set of calculated results for all production times of the nth fitted target, where y1, y2, ..., y m These represent the calculation results from day 1 to day m; X represents the set of independent variables; g i (x) represents the boundary conditions; Z represents the parameter space; k represents the number of input parameters; T represents the transpose sign.

[0085] Furthermore, by executing step S500 and inputting the parameter values ​​into the production capacity model, the parameter values ​​corresponding to the best fit are used to obtain the best inversion result.

[0086] Step S500: Sort the target case parameters in groups a with the smallest average absolute error from smallest to largest, and input the parameter values ​​of the group with the smallest error into the shale gas fracturing production capacity model for recalculation. Match the model with the historical production again, and the parameter values ​​corresponding to the best fitting effect are the best inversion results.

[0087] The optimal inversion result is determined by finding the parameter values ​​corresponding to the best fitting effect, with the goal of minimizing the gas production fitting error.

[0088] The technical solution provided by the embodiments of the present invention makes full use of existing geological-engineering-production data and adopts a combination of numerical simulation, artificial intelligence model and optimization algorithm to realize intelligent and rapid inversion of complex fracture network parameters of shale gas.

[0089] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0090] It should be noted that, in other embodiments of the present invention, the method can also combine one or more of the above embodiments to obtain a new intelligent inversion method for the characteristic parameters of complex fracture networks in shale gas, so as to achieve a quantitative and reliable analysis of the characteristics of complex fracture networks in shale.

[0091] Example 2

[0092] It should be noted that, based on the methods in any one or more embodiments of the present invention described above, the present invention also provides a storage medium storing program code that can implement the methods described in any one or more embodiments. When the program code is executed by the operating system, it can implement the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas as described above.

[0093] Example 3

[0094] The methods described in the above-disclosed embodiments of the present invention are detailed. These methods can be implemented using various devices or systems. Therefore, based on other aspects of the methods described in any one or more of the above embodiments, the present invention also provides an intelligent inversion system for characteristic parameters of complex fracture networks in shale gas. This system is used to execute the intelligent inversion method for characteristic parameters of complex fracture networks in shale gas described in any one or more of the above embodiments. Specific embodiments are given below for detailed description.

[0095] Specifically, Figure 4 The diagram shows a schematic representation of the intelligent inversion system for complex fracture network characteristic parameters of shale gas provided in an embodiment of the present invention. Figure 4 As shown, the system includes:

[0096] The numerical production capacity model establishment module is configured to establish a complex fracture shale gas fracturing production capacity model for shale reservoirs using an embedded discrete fracture model.

[0097] The basic case determination module is configured to determine the value space of the parameters to be inverted, obtain basic cases based on random sampling, and perform parameter sensitivity analysis.

[0098] The intelligent production capacity model training module is configured to use a basic calculation example as training sample and an intelligent agent model to train a machine learning model to form a fracturing production capacity prediction model.

[0099] The target case selection module is configured to select target case parameters that meet the error conditions based on the calculation results of the fracturing capacity prediction model using a set objective function.

[0100] The feature parameter inversion module is configured to input the selected target case parameters into the complex fracture web rock gas fracturing production capacity model to obtain the target fracture network feature parameter results.

[0101] In one optional embodiment, the numerical production capacity model building module is based on an embedded discrete fracture model to partition the analyzed reservoir, and considers shale reservoir characteristics such as anti-condensation, micro- and nano-scale nonlinear seepage, reservoir stress sensitivity, and desorption to establish a gas-water two-phase flow production capacity model as a complex fractured shale gas fracturing production capacity model, and explicitly characterizes and calculates the complex fracture network of the reservoir.

[0102] Furthermore, in one embodiment, the numerical productivity model building module is further configured to: calculate the fluid flow exchange between physically connected but non-adjacent grid cells in a complex fracture network, including fluid exchange between a fracture and its matrix cell, fluid exchange between two intersecting fractures, and fluid exchange between a single fracture and adjacent matrix cells.

[0103] In one embodiment, the basic calculation case determination module is configured to: analyze the geological engineering data and production data of the set well to determine the possible value range of the parameter to be fitted, and form the value space of the parameter to be inverted.

[0104] Preferably, in one embodiment, during the parameter sensitivity analysis process of the basic calculation case determination module, the number of effective connected cracks and the ratio of effective crack volume are introduced to quantitatively evaluate the complexity of the crack network.

[0105] In an optional embodiment, the intelligent production capacity model training module is configured as follows: the calculated basic examples are used as learning samples for the intelligent agent model and fed into the random forest intelligent agent model to train the machine learning model; based on the computational data analysis of the machine learning model, data index points are determined according to the production curve characteristics and computational results; when the production prediction accuracy of the obtained data point index numbers is greater than 95%, the training requirements are met, and a fracturing production capacity prediction model is formed.

[0106] Furthermore, in one embodiment, in step S400, the following objective function is established:

[0107] minF(X)=[f1(Y1),f2(Y2),…,f n (Y n )]

[0108] Y n =[y1,y2,…,y m ]T

[0109] y = f(X)

[0110] X = [x1, x2, ..., x k ] T

[0111]

[0112] In the formula, F(X) represents the objective function; f1(Y1), f2(Y2), ..., f n (Y n Y represents the average absolute error between the calculated and actual values ​​of all production days under a certain fitted target. n Let represent the set of calculated results for all production times of the nth fitted target, where y1, y2, ..., y m These represent the calculation results from day 1 to day m; X represents the set of independent variables; g i (x) represents the boundary conditions; Z represents the parameter space.

[0113] In one embodiment, the target case selection module uses an optimization algorithm to solve for the maximum and minimum values ​​of the objective function to obtain a combination of characteristic parameters of the complex fracture network; the optimization algorithm uses Markov chain-Monte Carlo sampling, evolutionary algorithm or genetic algorithm.

[0114] Preferably, in one embodiment, the feature parameter inversion module is configured to: input the values ​​of several target calculation example parameters with the smallest error into the shale gas fracturing production capacity model for recalculation, and then match and fit them with historical production again. The feature parameter values ​​of each fracture network corresponding to the best fitting effect are the target inversion results.

[0115] In the intelligent inversion system for characteristic parameters of complex fracture networks in shale gas provided in this embodiment of the invention, each module or unit structure can operate independently or in combination according to actual parameter setting requirements and model calculation requirements to achieve the corresponding technical effects.

[0116] It should be understood that the embodiments disclosed herein are not limited to the specific structures, processing steps, or materials disclosed herein, but should be extended to equivalent substitutions of these features as understood by those skilled in the art. It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

[0117] The phrase "an embodiment" in the specification means that a specific feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Therefore, the phrase "an embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment.

[0118] While the embodiments disclosed in this invention are as described above, the content is merely for the purpose of facilitating understanding of the invention and is not intended to limit the invention. Any person skilled in the art to which this invention pertains may make any modifications and variations in form and detail of the implementation without departing from the spirit and scope disclosed herein; however, the scope of patent protection for this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A method for intelligent inversion of characteristic parameters of complex fracture networks in shale gas, characterized in that, The method includes: Step S100: Establish a shale gas fracturing production capacity model for complex fractures in shale reservoirs using an embedded discrete fracture model; Step S200: Determine the value space of the parameters to be inverted, obtain basic calculation examples based on random sampling, and perform parameter sensitivity analysis; Step S300: Using basic calculation examples as training samples, a machine learning model is trained using an intelligent agent model to form a fracturing capacity prediction model; Step S400: Optimize the target case parameters that meet the error conditions based on the calculation results of the fracturing capacity prediction model using the set objective function; Step S500: Substitute the selected target example parameters into the complex fracture web rock gas fracturing production capacity model inversion to obtain the target fracture network characteristic parameter results.

2. The method according to claim 1, characterized in that, In step S100, the analyzed reservoir is partitioned based on an embedded discrete fracture model. A gas-water two-phase flow production capacity model is established, taking into account various shale reservoir characteristics such as anti-condensation, micro-nano scale nonlinear seepage, reservoir stress sensitivity, and desorption. This model serves as a complex fracture network shale gas fracturing production capacity model, and the complex fracture network of the reservoir is explicitly characterized and calculated.

3. The method according to claim 1, characterized in that, The fluid flow exchange between physically connected but non-adjacent mesh cells in a complex fracture network is calculated, including fluid exchange between a fracture and its matrix cell, fluid exchange between two intersecting fractures, and fluid exchange between a single fracture and its adjacent matrix cells.

4. The method according to claim 1, characterized in that, In step S200, the geological and engineering data and production data of the set well are analyzed to determine the possible range of values ​​for the parameters to be fitted, thus forming the value space of the parameters to be inverted.

5. The method according to claim 1, characterized in that, In the process of parameter sensitivity analysis, the number of effective connected cracks and the ratio of effective crack volume are introduced to quantitatively evaluate the complexity of the crack network.

6. The method according to claim 1, characterized in that, In step S300, the calculated basic examples are used as learning samples for the intelligent agent model and fed into the random forest intelligent agent model to train the machine learning model. Based on the computational data analysis of the machine learning model, data index points are determined according to the characteristics of the production curve and the computational results. When the production prediction accuracy of the obtained data point index numbers is greater than 95%, it is determined that the training requirements are met and a fracturing capacity prediction model is formed.

7. The method according to claim 1, characterized in that, In step S400, the following objective function is established: minF(X)=[f1(Y1),f2(Y2),…,f n (Y n )] Y n =[y1,y2,…,y m ] T y = f(X) X=[x1,x2,…,x k ] T In the formula, F(X) represents the objective function; f1(Y1), f2(Y2), ..., f n (Y n Y represents the average absolute error between the calculated and actual values ​​of all production days under a certain fitted target. n Let represent the set of calculated results for all production times of the nth fitted target, where y1, y2, ..., y m These represent the calculation results from day 1 to day m; X represents the set of independent variables; g i (x) represents the boundary conditions; Z represents the parameter space.

8. The method according to claim 1, characterized in that, In step S400, an optimization algorithm is used to solve for the maximum and minimum values ​​of the objective function to obtain the combination of characteristic parameters of the complex crack network; the optimization algorithm adopts Markov chain-Monte Carlo sampling, evolutionary algorithm or genetic algorithm.

9. The method according to claim 1, characterized in that, In step S500, the parameter values ​​of several target calculation cases with the smallest error are substituted into the shale gas fracturing production capacity model for recalculation, and then matched and fitted with historical production again. The values ​​of various fracture network characteristic parameters corresponding to the best fitting effect are the target inversion results.

10. A storage medium, characterized in that, The storage medium stores program code capable of implementing the method as described in any one of claims 1 to 9.

11. An intelligent inversion system for characteristic parameters of complex fracture networks in shale gas, characterized in that, The system performs the method as described in any one of claims 1 to 9.

Images