Method and apparatus for determining fracture conductivity

Abstract

The application discloses a kind of fracture conductivity determination method and device, it is related to oil and gas field development technical field, main purpose is to improve the efficiency and accuracy of fracture conductivity determination method.The main technical scheme of the present application is: obtaining the fracture conductivity database of target work area;According to the fracture conductivity database of target work area, obtain the basic architecture of fracture conductivity neural network;Determine neural network weight and threshold training method based on preset training accuracy;Based on the basic architecture of fracture conductivity neural network, the preset training accuracy and the neural network weight and threshold training method, adjust neural network training method and carry out neural network training, obtain neural network model;According to the neural network model, determine the fracture conductivity of target work area.The present application is mainly used for efficiently and accurately determining fracture conductivity, and then provides reliable fracture conductivity parameters for hydraulic fracturing reconstruction, design.

Description

Technical Field

[0001] This invention relates to the field of oil and gas field development technology, and in particular to a method and apparatus for determining fracture conductivity. Background Technology

[0002] Hydraulic fracturing is a key technology for the stimulation of new wells in unconventional reservoirs and for enhancing production in older wells in conventional reservoirs. The efficient conductivity construction of fractures within complex fracture networks is crucial for increasing oil and gas well production. To determine the optimal production method for oil and gas wells, it is necessary to ascertain the conductivity of the fractures.

[0003] Currently, the method for determining the conductivity of fractures is usually as follows: the conductivity of fractures is experimentally measured before hydraulic fracturing, and the conductivity of the actually filled fractures is evaluated and analyzed after hydraulic fracturing.

[0004] However, the aforementioned experimental methods require downhole coring of oil and gas wells. Their limitations lie in the fact that not all oil and gas wells undergo downhole coring, and there are instances where the obtained core samples do not meet the requirements for fracture conductivity testing. Furthermore, experimental measurements require specialized equipment, and the complexity and timeliness of the experimental process cannot meet the needs of large-scale hydraulic fracturing design. On the other hand, the fracture conductivity obtained through well test analysis is an average value for the entire fracture system, presenting a bottleneck for assessing the conductivity of specific fractures. Therefore, a more accurate and efficient method for determining fracture conductivity is urgently needed. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method and apparatus for determining the conductivity of fractures, the main purpose of which is to improve the efficiency and accuracy of the method for determining the conductivity of fractures.

[0006] To achieve the above objectives, the present invention mainly provides the following technical solutions:

[0007] On one hand, embodiments of the present invention provide a method for determining the conductivity of a fracture, including:

[0008] Obtain a database of crack conductivity in the target work area;

[0009] The basic architecture of the crack flow guiding neural network is obtained based on the crack flow guiding capacity database of the target work area.

[0010] A training method that determines neural network weights and thresholds based on preset training precision;

[0011] Based on the basic architecture of the crack guiding neural network, the preset training accuracy, and the neural network weight and threshold training method, the neural network training method is adjusted and the neural network is trained to obtain a neural network model.

[0012] The crack flow conduction capacity of the target work area is determined based on the neural network model.

[0013] Specifically, the database for obtaining the crack conductivity of the target work area includes:

[0014] Obtain a basic database of crack flow conductivity;

[0015] The crack conductivity database for the target work area is updated based on the basic database of crack conductivity and the crack conductivity data of the target work area.

[0016] Specifically, updating the fracture conductivity database of the target work area based on the fracture conductivity base database and the fracture conductivity data of the target work area includes:

[0017] Based on the fracture conductivity database and the fracture conductivity data of the target work area, the fracture conductivity database of the target work area is updated using a first preset formula. The first preset formula is: in, For the updated crack conductivity database of the target work area, The crack conductivity data for the target work area before the update. To update the database The maximum value, To update the database The minimum value.

[0018] Specifically, the basic architecture of the fracture conductivity neural network obtained based on the fracture conductivity database of the target work area includes:

[0019] Based on the number of parameters and the number of samples in the crack diversion capacity database of the target work area, the number of hidden layers of the neural network, the number of neurons in the hidden layers of the neural network, and the type of activation function are obtained by the second preset formula, the third preset formula, and the fourth preset formula, respectively.

[0020] The parameters in the fracture conductivity database include one or more of the following: rock mechanical properties, fracture closure pressure, proppant concentration, proppant particle size, proppant type, test temperature, and test time.

[0021] Specifically, the second preset formula is: Where H is the number of hidden layers in the neural network, m is the number of parameters in the fracture conductivity database of the target work area, n is the number of samples in the fracture conductivity database of the target work area, and mod is an integer; and / or,

[0022] The third preset formula is: Where C is the number of neurons in the hidden layer of the neural network, m is the number of parameters in the crack conductivity database of the target work area, n is the number of samples in the crack conductivity database of the target work area, and mod is an integer; and / or,

[0023] The fourth preset formula is:

[0024]

[0025] Where Type is the activation function type, H is the number of hidden layers in the neural network, C is the number of neurons in the hidden layers of the neural network, mod is to round to the nearest integer, and Linear, Log-sigmoid, Tangent sigmoid, and ReLU are the activation function type names of the neural network model, respectively.

[0026] Specifically, the method for determining neural network weights and thresholds based on a preset training precision includes:

[0027] Based on the preset training precision, a neural network weight and threshold training method is obtained through a fifth preset formula, wherein the fifth preset formula is: Where F represents the neural network weight and threshold training method, e represents the preset training precision, and batch gradient descent, stochastic gradient descent, Momentum, Nesterov Momentum, RMSProp, and Adam are the names of the neural network weight and threshold training methods, respectively.

[0028] Specifically, determining the crack conduction capacity of the target work area based on the neural network model includes:

[0029] Obtain parameter values ​​for the target work area, which include any one or more of the following parameters: rock mechanical properties, fracture closure pressure, proppant particle size, proppant type, and proppant sand concentration.

[0030] Based on the parameter values ​​of the target work area, the crack conductivity of the target work area is calculated using the neural network model.

[0031] When the parameter values ​​of the target work area include multiple sets of parameter values, the corresponding crack conduction capacity is calculated by the neural network model according to each set of parameter values.

[0032] On the other hand, embodiments of the present invention also provide a device for determining the conductivity of a fracture, comprising:

[0033] The acquisition unit is used to acquire a database of crack conductivity in the target work area.

[0034] A construction unit is used to obtain the basic architecture of a crack flow-guiding neural network based on the crack flow-guiding capacity database of the target work area obtained by the acquisition unit.

[0035] The first determining unit is used to determine the neural network weights and threshold training method based on a preset training precision.

[0036] The training unit is used to adjust the neural network training method and perform neural network training based on the basic architecture of the crack guiding neural network obtained by the construction unit, the preset training accuracy, and the neural network weights and threshold training method determined by the first determining unit, so as to obtain a neural network model.

[0037] The second determining unit is used to determine the crack flow conduction capacity of the target work area based on the neural network model obtained by the training unit.

[0038] On the other hand, embodiments of the present invention also provide a system for determining the conductivity of a fracture, comprising:

[0039] The system includes a memory and one or more processors, the processors being configured to execute program instructions stored in the memory, which, when executed, perform the aforementioned method for determining the flow capacity of cracks.

[0040] On the other hand, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program, when executed by one or more processors, implements the above-described method for determining the flow capacity of cracks.

[0041] This invention proposes a method and apparatus for determining fracture conductivity. By acquiring a fracture conductivity database of a target work area, a basic architecture of a fracture conductivity neural network is obtained. The network weights and thresholds are determined based on a preset training precision, and a neural network model is obtained through adaptive adjustment and training of the neural network architecture. This model is then used to determine the fracture conductivity of the target work area, providing reliable fracture conductivity parameters for hydraulic fracturing and design. This method integrates expertise from rock mechanics, fluid mechanics, databases, and neural network technology, enabling comprehensive, efficient, and accurate determination of fracture conductivity under multi-factor coupling conditions. Attached Figure Description

[0042] Figure 1 A flowchart illustrating a method for determining the conductivity of a crack, as provided in an embodiment of the present invention;

[0043] Figure 2 A flowchart of another method for determining the flow conductivity of a crack provided in an embodiment of the present invention;

[0044] Figure 3 This is a block diagram of a crack flow-conducting capacity determination device provided in an embodiment of the present invention. Detailed Implementation

[0045] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0046] This invention provides a method for determining fracture conductivity. This method achieves fracture conductivity determination under multi-factor coupled control conditions based on neural networks, aiming to improve the efficiency and accuracy of fracture conductivity determination methods. The specific steps of this method are as follows: Figure 1 As shown, it includes:

[0047] 101. Obtain the crack conductivity database for the target work area.

[0048] The target work area refers to the work area where fracture conductivity is to be determined; in this embodiment, it refers to oil and gas reservoirs. The fracture conductivity database can include parameter data related to fracture conductivity, such as test temperature, test time, rock type, closure pressure, proppant particle size, and proppant concentration. These parameters can be dynamically updated based on parameter data obtained from fracture conductivity tests on the target work area and a pre-set basic database, using methods such as normalization. Multiple parameter data can be obtained from multiple tests on the same reservoir, thus forming multiple sample data sets.

[0049] The basic database can be obtained through experiments and well test interpretation, and can be the target reservoir or historical test results similar to the target reservoir. By updating the basic database with parameter data from the target reservoir, the resulting fracture conductivity database is more suitable for the target reservoir and provides a richer and more accurate data source.

[0050] 102. The basic architecture of the crack flow guiding neural network is obtained based on the crack flow guiding capacity database of the target work area.

[0051] After obtaining the crack conductivity database of the target work area, the basic architecture of the crack conductivity neural network can be constructed based on the size of the crack conductivity database and computational efficiency.

[0052] In this embodiment of the invention, the basic architecture of the crack flow-guiding neural network includes the number of hidden layers, the number of neurons in the hidden layers, and the activation function. The number of hidden layers, the number of neurons in the hidden layers, and the activation function can be determined based on the number of parameters and samples in the crack flow-guiding capacity database, taking into account both computational accuracy and processing time.

[0053] Specifically, the following implementation methods can be adopted:

[0054] Through formula The number of hidden layers H in the neural network is calculated.

[0055] Through formula The number of neurons C in the hidden layer of the neural network was calculated.

[0056] Through formula Obtain the activation function type Type;

[0057] Where m is the number of parameters in the fracture conductivity database of the target work area, n is the number of samples in the fracture conductivity database of the target work area, mod is an integer, and Linear, Log-sigmoid, Tangentsigmoid, and ReLU are the activation function type names of the neural network model. Linear represents a linear function, Log-sigmoid represents a Log-type sigmoid function, Tangentsigmoid represents a hyperbolic tangent sigmoid function, and ReLU represents a linear rectified function.

[0058] 103. A training method for determining neural network weights and thresholds based on preset training accuracy.

[0059] The preset accuracy can be determined in advance based on the actual test scenario of the target work area. Based on the size of the crack conductivity database, the neural network weights and threshold training methods are determined using computational accuracy as the standard.

[0060] In this embodiment of the invention, the following implementation methods may be adopted:

[0061] Through formula A method for training neural network weights and thresholds was derived.

[0062] Where F represents the neural network weight and threshold training method, e represents the preset training precision, and batch gradient descent, stochastic gradient descent, Momentum, Nesterov Momentum, RMSProp, and Adam are the names of the weight and threshold training methods for the neural network, respectively. Specifically, batch gradient descent is the batch gradient descent algorithm, stochastic gradient descent is the stochastic gradient descent algorithm, Momentum is Newton's momentum method, Nesterov Momentum is the Nesterov momentum method, RMSProp (root mean square propagation) is the forward root mean square gradient descent algorithm, and Adam is the Adam algorithm.

[0063] 104. Based on the basic architecture of the crack guiding neural network, the preset training accuracy, and the neural network weight and threshold training method, adjust the neural network training method and perform neural network training to obtain a neural network model.

[0064] After determining the neural network weights and threshold training method, the neural network training method can be adjusted to train the neural network model to the preset training accuracy. The training process usually includes model training and model validation. Once both the training accuracy and validation accuracy reach the preset training accuracy standard, the model training is complete, and a trained neural network model is obtained.

[0065] 105. Determine the crack conduction capacity of the target work area based on the neural network model.

[0066] The process involves first obtaining the parameter values ​​for the target work area, then preprocessing these values ​​to conform to the input data format of the neural network model. This preprocessing includes tasks such as padding missing data and normalization. Finally, the preprocessed parameter values ​​are input into the neural network model to obtain the flow guidance capacity result for the target work area.

[0067] A specific implementation method may be as follows: the parameter values ​​of the target work area may include rock mechanical properties, fracture closure pressure, proppant particle size, proppant type and proppant sand concentration, etc., and the fracture conductivity of the target work area is calculated by a neural network model based on the parameter values ​​of the target work area.

[0068] When the target work area has multiple sets of parameter values, the corresponding crack conduction capacity is calculated using a neural network model based on each set of parameter values.

[0069] The fracture conductivity determination method provided in this invention obtains a fracture conductivity database for a target work area, then derives the basic architecture of a fracture conductivity neural network. Based on a preset training precision, it determines the neural network weights and thresholds for training. Through adaptive adjustment and training of the fracture conductivity neural network architecture, a neural network model is obtained, thereby determining the fracture conductivity of the target work area. This provides reliable fracture conductivity parameters for hydraulic fracturing and design. This method integrates expertise from rock mechanics, fluid mechanics, databases, and neural network technology, enabling comprehensive, efficient, and accurate determination of fracture conductivity under multi-factor coupling conditions.

[0070] Based on the above explanation, such as Figure 2 As shown in the figure, this embodiment of the invention also provides a method for determining the conductivity of a fracture, the method comprising:

[0071] 201. Obtain the basic database of fracture conductivity.

[0072] The basic database can be obtained through experiments and well tests, and can be the target reservoir or historical test results similar to the target reservoir.

[0073] 202. The crack conductivity database of the target work area is updated based on the crack conductivity database and the crack conductivity data of the target work area.

[0074] Specifically, fracture conductivity data for the target work area can be obtained through experimental testing. The parameter types of this experimental test data are then matched with the parameter types in the basic database. If a match is found, the experimental test data corresponding to the matched parameter type is updated in the corresponding parameter data of the basic database. If a match is not found, the mismatched parameter type is added to the basic database, and the current parameter data corresponding to the mismatched parameter type is updated in the corresponding parameter data of the basic database. The parameter data corresponding to each parameter type in the sample database to be updated are then normalized to obtain the fracture conductivity database for the target work area. A specific implementation method can be as follows:

[0075] According to the formula A database of crack conductivity in the target work area was obtained, among which... For the updated crack conductivity database of the target work area, The crack conductivity data for the target work area before the update. To update the database The maximum value, To update the database The minimum value.

[0076] In the method for determining the flow capacity of cracks in this embodiment of the invention, the crack flow capacity database of the target work area is updated based on the basic database and the crack flow capacity data of the target work area, which can improve the accuracy and comprehensiveness of the determination of crack flow capacity.

[0077] 203. Based on the number of parameters and the number of samples in the crack diversion capacity database of the target work area, the number of hidden layers of the neural network, the number of neurons in the hidden layers of the neural network, and the activation function type are obtained by using the second preset formula, the third preset formula, and the fourth preset formula, respectively.

[0078] The parameters in the fracture conductivity database include one or more of the following: rock mechanical properties, fracture closure pressure, proppant concentration, proppant particle size, proppant type, test temperature, and test time.

[0079] The second preset formula is Where H is the number of hidden layers in the neural network, m is the number of parameters in the crack conductivity database of the target work area, n is the number of samples in the crack conductivity database of the target work area, and mod is an integer;

[0080] The third preset formula is: Where C is the number of neurons in the hidden layer of the neural network, m is the number of parameters in the crack diversion capacity database of the target work area, n is the number of samples in the crack diversion capacity database of the target work area, and mod is an integer.

[0081] The fourth preset formula is:

[0082]

[0083] Where Type is the activation function type, H is the number of hidden layers in the neural network, C is the number of neurons in the hidden layers of the neural network, mod is to round to the nearest integer, and Linear, Log-sigmoid, Tangent sigmoid, and ReLU are the activation function type names of the neural network model, respectively.

[0084] The relevant descriptions of the number of hidden layers, the number of neurons in the hidden layers of the neural network, and the type of activation function obtained in step 203 have been described in step 102 of the aforementioned embodiments, and will not be repeated here. For details, please refer to the description in step 102.

[0085] 204. Based on the preset training precision, obtain the neural network weights and threshold training method through the fifth preset formula.

[0086] The fifth preset formula is: Where F represents the neural network weight and threshold training method, e represents the preset training precision, and batch gradient descent, stochastic gradient descent, Momentum, Nesterov Momentum, RMSProp, and Adam are the names of the weight and threshold training methods for the neural network, respectively.

[0087] The relevant description of the neural network weight and threshold training method obtained in step 204 has been described in step 103 of the aforementioned embodiments, and will not be repeated here. For details, please refer to the description in step 103.

[0088] In the fracture conductivity determination method of this invention, the number of hidden layers, the number of neurons in the hidden layers of the neural network, and the activation function type are obtained based on the number of parameters and the number of samples in the fracture conductivity database. The basic architecture of the fracture conductivity neural network is constructed, which is more in line with the actual geological characteristics and data acquisition characteristics of the current target work area, thereby improving the accuracy of model construction.

[0089] 205. Based on the basic architecture of the crack guiding neural network, the preset training accuracy, and the neural network weight and threshold training method, adjust the neural network training method and perform neural network training to obtain a neural network model.

[0090] In step 205, adjusting the neural network training method and training the neural network to obtain the relevant description of the neural network model has been described in step 104 of the aforementioned embodiments, and will not be repeated here. For details, please refer to the description in step 104.

[0091] In the fracture conductivity determination method of this invention, the algorithm for adjusting the weights and thresholds of the neural network is optimized. Based on the preferred model architecture parameters and the optimization algorithm, the prediction model is constructed, which makes the prediction results of the constructed model more consistent with the actual geological characteristics and data acquisition characteristics of the current target work area, thereby improving the accuracy of the model prediction.

[0092] 206. Obtain the parameter values ​​of the target work area, which include any one or more of the following parameters: rock mechanical properties, fracture closure pressure, proppant particle size, proppant type, and proppant sand concentration.

[0093] 207. Based on the parameter values ​​of the target work area, the crack conduction capacity of the target work area is calculated using the neural network model.

[0094] When the parameter values ​​of the target work area include multiple sets of parameter values, the corresponding crack conduction capacity is calculated by the neural network model according to each set of parameter values.

[0095] The relevant descriptions of steps 206 and 207 have been described in step 105 of the aforementioned embodiments, and will not be repeated here. For details, please refer to the description in step 105.

[0096] In one scenario example, fracture conductivity tests have been conducted on reservoir B in an oilfield. A total of 3 sets of experiments were conducted, resulting in 6 parameters and 3 samples. The parameters are test temperature, test time, rock type, closure pressure, proppant particle size, and proppant concentration. The test data can be added to the basic database and the database can be updated. The updated database is more suitable for testing the fracture conductivity of reservoir B in this oilfield.

[0097] For example, the updated target work area's crack diversion capacity database has 6 parameters and 100 samples. The above-described embodiment can be used to obtain a neural network with 1 hidden layer, 7 hidden layer neurons, and a linear activation function, thus constructing the basic architecture of the crack diversion neural network.

[0098] Assuming that the preset training accuracy is determined to be 10 based on the actual testing environment of reservoir B in oilfield. -5 Based on the above embodiment, the neural network weight and threshold training method is determined to be Nesterov Momentum. Then, through adaptive adjustment of the crack flow guiding capacity neural network architecture and neural network training, the neural network model is trained to a preset training accuracy of 10. -5 This ensures that both training and validation accuracy reach the required level, completing model training and yielding the neural network model.

[0099] Then, measured data on parameters such as rock mechanical properties, fracture closure pressure, proppant particle size, proppant type, and proppant concentration in the target work area can be obtained. These data are then input into the neural network model to obtain the fracture conductivity of reservoir B in oilfield B. If fracture conductivity with different closure pressures, proppant particle sizes, and proppant types is required, the measured data can be adjusted accordingly, and the data can be input into the neural network model again to obtain the fracture conductivity of reservoir B in oilfield B.

[0100] Assume the measured parameters of reservoir B are as follows: Young's modulus of rock is 58 GPa, Poisson's ratio is 0.25, fracture closure pressure is 70 MPa, proppant particle size is 40 / 70 mesh ceramsite, and proppant concentration is 4 kg / m³. 2 The fracture conductivity obtained based on the aforementioned method is 1.03D.cm.

[0101] The fracture conductivity determination method provided in this invention dynamically updates the fracture conductivity database to obtain a database of fracture conductivity for the target work area. This leads to the basic architecture of a fracture conductivity neural network. The method then determines the neural network weights and thresholds based on a preset training precision. Through adaptive adjustment and training of the neural network architecture, a neural network model is obtained. By considering factors such as reservoir rock mechanics properties, fracture closure pressure, proppant particle size, proppant type, and proppant concentration, the neural network predicts fracture conductivity, providing reliable fracture conductivity parameters for hydraulic fracturing and design. This method integrates expertise from rock mechanics, fluid mechanics, databases, and neural network technology, enabling comprehensive, efficient, and accurate determination of fracture conductivity under multi-factor coupling conditions.

[0102] Furthermore, as an implementation of the above-mentioned method for determining fracture conductivity, this embodiment of the invention provides a device for determining fracture conductivity, which is mainly used to efficiently and accurately determine fracture conductivity. It should be noted that this device embodiment will not repeat the details of the aforementioned method embodiments one by one, but it should be clear that the device in this embodiment can correspondingly implement all the contents of the aforementioned method embodiments. The device is as follows... Figure 3 As shown, it specifically includes:

[0103] Acquisition unit 31 is used to acquire the crack conductivity database of the target work area;

[0104] Construction unit 32 is used to obtain the basic architecture of the crack conduction neural network based on the crack conduction capacity database of the target work area obtained by acquisition unit 31;

[0105] The first determining unit 33 is used to determine the neural network weights and threshold training method based on a preset training precision.

[0106] Training unit 34 is used to adjust the neural network training method and perform neural network training based on the basic architecture of the crack guiding neural network obtained by the construction unit 32, the preset training accuracy, and the neural network weights and threshold training method determined by the first determining unit 33, so as to obtain a neural network model.

[0107] The second determining unit 35 is used to determine the crack flow conduction capacity of the target work area based on the neural network model obtained by the training unit 34.

[0108] Furthermore, this embodiment of the invention also provides a fracture conductivity determination system, including: a memory and one or more processors, wherein the processors are configured to execute program instructions stored in the memory, and the program instructions execute the above-described fracture conductivity determination method when they are executed.

[0109] The crack flow capacity determination system includes a memory and a processor. That is, the acquisition unit, construction unit, first determination unit, training unit and second determination unit in the above embodiments are all stored in the memory as program units, and the processor executes the above program units stored in the memory to realize the corresponding functions.

[0110] Furthermore, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program, when executed by one or more processors, implements the above-described method for determining the flow capacity of cracks.

[0111] In summary, the fracture conductivity determination method and apparatus proposed in this invention obtains a fracture conductivity database of the target work area, thereby deriving the basic architecture of a fracture conductivity neural network. Based on a preset training precision, the neural network weights and thresholds are determined for training. Through adaptive adjustment of the fracture conductivity neural network architecture and neural network training, a neural network model is obtained, thus determining the fracture conductivity of the target work area. This provides reliable fracture conductivity parameters for hydraulic fracturing and design. This fracture conductivity determination method integrates expertise from rock mechanics, fluid mechanics, databases, and neural network technology, enabling comprehensive, efficient, and accurate determination of fracture conductivity under multi-factor coupling conditions.

[0112] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0113] It is understood that the relevant features in the above methods and apparatus can be referenced interchangeably. Furthermore, the terms "first," "second," etc., in the above embodiments are used to distinguish between embodiments and do not represent the superiority or inferiority of any particular embodiment.

[0114] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0115] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this invention is not directed to any particular programming language. It should be understood that the contents of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0116] In addition, the memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0117] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0122] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0123] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0124] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0125] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for determining the conductivity of a fracture, characterized in that, include: Obtain a database of crack conductivity in the target work area; The basic architecture of the crack flow guiding neural network is obtained based on the crack flow guiding capacity database of the target work area. A training method that determines neural network weights and thresholds based on preset training precision; Based on the basic architecture of the crack guiding neural network, the preset training accuracy, and the neural network weight and threshold training method, the neural network training method is adjusted and the neural network is trained to obtain a neural network model. The crack flow conduction capacity of the target work area is determined based on the neural network model. The database for obtaining the fracture conductivity of the target work area includes: Obtain a basic database of crack flow conductivity; The crack conductivity database for the target work area is updated based on the crack conductivity database and the crack conductivity data for the target work area. The basic architecture of the fracture conductivity neural network obtained based on the fracture conductivity database of the target work area includes: Based on the number of parameters and the number of samples in the crack diversion capacity database of the target work area, the number of hidden layers of the neural network, the number of neurons in the hidden layers of the neural network, and the type of activation function are obtained by the second preset formula, the third preset formula, and the fourth preset formula, respectively. The parameters in the fracture conductivity database include one or more of the following: rock mechanical properties, fracture closure pressure, sand concentration, proppant particle size, proppant type, test temperature, and test time. The second preset formula is ,in, The number of hidden layers in the neural network. The number of parameters in the crack conductivity database for the target work area. The number of samples in the crack conductivity database for the target work area. To take an integer; and / or, The third preset formula is: ,in, This represents the number of neurons in the hidden layer of the neural network. The number of parameters in the crack conductivity database for the target work area. The number of samples in the crack conductivity database for the target work area. To take an integer; and / or, The fourth preset formula is: ， ， in, It is an activation function type. The number of hidden layers in the neural network. This represents the number of neurons in the hidden layer of the neural network. To take integer values, Linear, Log-sigmoid, Tangent sigmoid, and ReLU are the activation function type names of the neural network model, respectively. The method for determining neural network weights and thresholds based on preset training precision includes: Based on the preset training precision, a neural network weight and threshold training method is obtained through a fifth preset formula, wherein the fifth preset formula is: ,in, This refers to a method for training neural network weights and thresholds. To preset the training accuracy, batch gradient descent, stochastic gradient descent, Momentum, Nesterov Momentum, RMSProp, and Adam are the names of the training methods for the weights and thresholds of the neural network, respectively.

2. The method according to claim 1, characterized in that, The process of updating the fracture conductivity database of the target work area based on the fracture conductivity base database and the fracture conductivity data of the target work area includes: Based on the fracture conductivity database and the fracture conductivity data of the target work area, the fracture conductivity database of the target work area is updated using a first preset formula. The first preset formula is: ,in, For the updated crack conductivity database of the target work area, The crack conductivity data for the target work area before the update. To update the database The maximum value, To update the database The minimum value.

3. The method according to claim 1, characterized in that, The step of determining the crack conduction capacity of the target work area based on the neural network model includes: Obtain parameter values ​​for the target work area, which include any one or more of the following parameters: rock mechanical properties, fracture closure pressure, proppant particle size, proppant type, and proppant sand concentration. Based on the parameter values ​​of the target work area, the crack conductivity of the target work area is calculated using the neural network model. When the parameter values ​​of the target work area include multiple sets of parameter values, the corresponding crack conduction capacity is calculated by the neural network model according to each set of parameter values.

4. A device for determining the flow conductivity of a fracture, characterized in that, include: The acquisition unit is used to acquire a database of crack conductivity in the target work area. A construction unit is used to obtain the basic architecture of a crack flow-guiding neural network based on the crack flow-guiding capacity database of the target work area obtained by the acquisition unit. The first determining unit is used to determine the neural network weights and threshold training method based on a preset training precision. The training unit is used to adjust the neural network training method and perform neural network training based on the basic architecture of the crack guiding neural network obtained by the construction unit, the preset training accuracy, and the neural network weights and threshold training method determined by the first determining unit, so as to obtain a neural network model. The second determining unit is used to determine the crack flow conduction capacity of the target work area based on the neural network model obtained by the training unit. The acquisition unit is specifically used for: Obtain a basic database of crack flow conductivity; The crack conductivity database for the target work area is updated based on the crack conductivity database and the crack conductivity data for the target work area. The construction unit is specifically used to: obtain the number of hidden layers of the neural network, the number of neurons in the hidden layers of the neural network, and the activation function type according to the number of parameters and the number of samples in the crack diversion capacity database of the target work area, respectively, through the second preset formula, the third preset formula, and the fourth preset formula; The parameters in the fracture conductivity database include one or more of the following: rock mechanical properties, fracture closure pressure, sand concentration, proppant particle size, proppant type, test temperature, and test time. The second preset formula is ,in, The number of hidden layers in the neural network. The number of parameters in the crack conductivity database for the target work area. The number of samples in the crack conductivity database for the target work area. To take an integer; and / or, The third preset formula is: ,in, This represents the number of neurons in the hidden layer of the neural network. The number of parameters in the crack conductivity database for the target work area. The number of samples in the crack conductivity database for the target work area. To take an integer; and / or, The fourth preset formula is: ， ， in, It is an activation function type. The number of hidden layers in the neural network. This represents the number of neurons in the hidden layer of the neural network. To take integer values, Linear, Log-sigmoid, Tangent sigmoid, and ReLU are the activation function type names of the neural network model, respectively. The first determining unit is specifically used for: obtaining a neural network weight and threshold training method based on a preset training precision using a fifth preset formula, wherein the fifth preset formula is... ,in, This refers to a method for training neural network weights and thresholds. To preset the training accuracy, batch gradient descent, stochastic gradient descent, Momentum, Nesterov Momentum, RMSProp, and Adam are the names of the training methods for the weights and thresholds of the neural network, respectively.

5. A system for determining the conductivity of a fracture, characterized in that, include: A memory and one or more processors, the processors being configured to execute program instructions stored in the memory, which, when executed, perform the method for determining the flow capacity of a crack as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by one or more processors, implements the method for determining the flow capacity of a crack as described in any one of claims 1-3.

Images