Fractured oil and gas well yield dynamic prediction method, equipment, medium and product

Abstract

The invention discloses a fractured oil and gas well yield dynamic prediction method, equipment, a medium and a product, and relates to the field of oil and gas field development. The method comprises the following steps: acquiring geological parameter data and yield dynamic data of different types of reservoirs of different oil and gas reservoirs; converting the geological parameter data into image data; classifying the image data by adopting an unsupervised clustering method; and constructing a data set based on the image data and the classification number to train a deep learning model, and using the deep learning model as a geological parameter classification model after training is completed to obtain a prediction classification number of the to-be-predicted image data and retrieve the target image data and the to-be-corrected image according to the prediction classification number. The to-be-corrected image is selected from a Glamer angle field image generated by yield dynamic data; calculating a difference image between the to-be-predicted image data and the target image data for correcting the to-be-corrected image; and the corrected image is inverted into predicted yield dynamic data, so that the calculation time consumption required by the yield dynamic prediction can be greatly reduced, and the prediction cost is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of oil and gas field development, and in particular to a method, equipment, medium and product for dynamically predicting the production of fractured oil and gas wells. Background Art

[0002] Tight oil and gas reservoirs have low permeability and low productivity of conventional wells. Generally, hydraulic fracturing technology is required to obtain economic productivity. Accurately predicting the production dynamics of fracturing oil and gas wells is the basis for formulating reliable development plans and accurately evaluating development benefits. At present, the production dynamics prediction of fracturing oil and gas wells is mostly based on reservoir numerical simulation, which must rely on expensive numerical simulation software, and the simulation calculation is time-consuming and labor-intensive. Summary of the invention

[0003] The purpose of this application is to provide a method, equipment, medium and product for dynamic prediction of production of fractured oil and gas wells, so as to significantly reduce the calculation time required for dynamic prediction of production and reduce the prediction cost.

[0004] To achieve the above objectives, this application provides the following solutions.

[0005] In a first aspect, the present application provides a method for dynamically predicting the production of a fractured oil and gas well, comprising: Acquire multiple sets of geological parameter data and corresponding production dynamic data of different types of reservoirs in different oil and gas reservoirs; the production dynamic data is production data at different times; Convert each set of geological parameter data into corresponding image data; The image data is classified using an unsupervised clustering method, and the classification results are optimized, and the optimized clusters are numbered as the final classifications; A data set is constructed based on the image data and the corresponding classification number, and a deep learning model is trained. After the training is completed, the model is used as a geological parameter classification model; the input of the geological parameter classification model is the image data, and the output is the corresponding classification number; The production dynamic data is processed using the Gram angle field to generate a Gram angle field image; The Gram angle field image and the image data converted from the corresponding geological parameter data are stored correspondingly to construct a database; The geological parameter data to be predicted is converted into image data to be predicted, and input into the geological parameter classification model for classification to obtain a prediction classification number; Based on the predicted classification number, the image data with the highest structural similarity under the same classification number is retrieved from the database as the target image data; Calculate the difference image between the image data to be predicted and the target image data using a structural similarity method; The Gram angle field image corresponding to the target image data in the database is used as the image to be corrected, and the image to be corrected is corrected based on the difference image to obtain a corrected image; The corrected image is reversely converted into predicted production dynamics data.

[0006] Optionally, the obtaining of multiple sets of geological parameter data of different types of reservoirs in different oil and gas reservoirs specifically includes: If the reservoir type is an oil layer, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, oil, gas and water saturation, oil, gas and water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability and relative permeability curve data; PVT data includes pressure, volume and temperature data; If the reservoir type is a tight gas reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability, and relative permeability curve data; If the reservoir type is a shale gas reservoir or a coalbed methane reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability, relative permeability curve, Langmuir volume and Langmuir pressure data.

[0007] Optionally, the converting each set of geological parameter data into corresponding image data specifically includes: According to the number of types of geological parameter data contained in each set of geological parameter data Determines the width of the image data to be converted and height ;That and , To fix the filling is an integer; Each geological parameter data contained in each set of geological parameter data is normalized to a value range between [0-1]; Map the normalized data to the range of [0-255], expressed as the grayscale value of the grayscale image; Arrange the mapped grayscale values ​​from top to bottom and from left to right and reconstruct them into OK A two-dimensional array of columns, each position in the two-dimensional array represents a pixel position, the grayscale value of each position represents the grayscale value of the pixel, and positions without grayscale values ​​are filled with zeros to form a grayscale image as the corresponding image data.

[0008] Optionally, the unsupervised clustering method is used to classify the image data, and the classification results are optimized, and the optimized clusters are numbered as the final classifications, specifically including: Use the Gaussian kernel function to calculate the similarity between every two image data and construct a similarity matrix; Use the k-nearest neighbor graph to construct a similarity graph corresponding to the similarity matrix, and construct a Laplacian matrix based on the similarity graph; Perform eigendecomposition on the Laplace matrix and calculate the pre- feature vector, this The feature vectors are composed as columns The matrix U, each row in the matrix U is a dimensional row vector, total row vectors; To this The row vectors are clustered using the K-means clustering method to obtain the distribution of clusters, where each cluster represents a classification; The clusters contained in the classification results are optimized based on the DB index, and the optimized clusters are numbered as the final classifications.

[0009] Optionally, the production dynamic data is processed using the Gram angle field to generate a Gram angle field image, specifically including: The dynamic data of production is expressed in the form of a tuple as ; where each pair of values It refers to time The production data at ; , is the number of sampling points; The production data Zoom to Within the range, get the scaled production data ; Representing scaled production data in polar coordinates ,get Polar angle in polar coordinates and polar diameter ; Based on polar angle and polar diameter Generate a Gram angular field image.

[0010] Optionally, the step of calculating a difference image between the image data to be predicted and the target image data using a structural similarity method specifically includes: Using a structural similarity method to calculate structural similarity data between the image data to be predicted and the target image data; The structural similarity data is converted into corresponding image data as a difference image.

[0011] Optionally, the correcting the image to be corrected based on the difference image to obtain the corrected image specifically includes: Divide the image to be corrected into pixel blocks; based on The grayscale value of the corresponding pixel in the difference image of size is The RGB channel color values ​​of the pixel blocks at corresponding positions in the pixel blocks are adjusted to obtain a corrected image.

[0012] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for dynamically predicting the production of fractured oil and gas wells.

[0013] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for dynamically predicting the production of fractured oil and gas wells.

[0014] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the method for dynamically predicting the production of fractured oil and gas wells.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects.

[0016] The present application provides a method, device, medium and product for dynamic prediction of the production of fractured oil and gas wells. A data set consisting of image data and corresponding classification numbers is pre-established, and a deep learning method is used to train the deep learning model. The trained geological parameter classification model can be used to quickly classify the geological parameter data to be predicted; then, based on the predicted classification number, the target image data can be quickly retrieved from the database; the difference image between the image data to be predicted and the target image data is used to correct the image to be corrected, and the corrected image is reversed to obtain the predicted production dynamic data. The present application method does not need to rely on expensive numerical simulation software for reservoir numerical simulation, thus greatly reducing the computational time required for dynamic production prediction and reducing the prediction cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of a flow chart of a method for dynamically predicting the production of a fractured oil and gas well in this application; Figure 2 is a schematic diagram of structural similarity data between image data to be predicted and target image data in a certain exemplary embodiment; Figure 3 It is a schematic diagram of dividing the image to be corrected into pixel blocks in an exemplary embodiment. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0020] The present application proposes a method, device, medium and product for dynamic prediction of production of fractured oil and gas wells, aiming to significantly reduce the computational time required for dynamic prediction of production and reduce the prediction cost.

[0021] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0022] In an exemplary embodiment, Figure 1 As shown, a method for dynamic prediction of the production of a fractured oil and gas well is provided, comprising the following steps 1 to 11.

[0023] Step 1: Obtain multiple sets of geological parameter data and corresponding production dynamic data for different types of reservoirs in different oil and gas reservoirs.

[0024] During the data collection phase, multiple sets of geological parameter data and corresponding production dynamic data are collected for different types of reservoirs in different oil and gas reservoirs. The reservoir types include oil layers, tight gas reservoirs, shale gas reservoirs, and coalbed methane reservoirs.

[0025] Specifically, if the reservoir type is an oil layer, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, oil, gas and water saturation, oil, gas and water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability and relative permeability curve, etc. PVT data includes pressure, volume and temperature data; where P represents the pressure of the oil and gas reservoir, V represents the volume, and T represents the temperature.

[0026] If the reservoir type is a tight gas reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability and relative permeability curve.

[0027] If the reservoir type is a shale gas reservoir or a coalbed methane reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability, relative permeability curve, Langmuir volume and Langmuir pressure data.

[0028] The production dynamic data refers to the production data at different moments (times), that is, the time series of production data. Each set of geological parameter data corresponds to a set of production dynamic data.

[0029] Step 2: Convert each set of geological parameter data into corresponding image data.

[0030] Each set of geological parameter data in the multiple sets of geological parameter data is processed as a whole, and each set of geological parameter data is converted into a corresponding image data. The number of types of geological parameter data contained in each set of geological parameter data is recorded as , that is, each set of geological parameter data includes geological parameter data. Determines the width of the image data to be converted and height , to Geological parameter data are used to reshape the shape. and Satisfies the following formula: (1); in To fix the filling is an integer.

[0031] For the deep learning model to be used in step 4, in order to ensure efficiency and performance during training and inference, the convolution kernel and stride must be consistent, and the side lengths of the reshaped image data must be equal, that is, and The following formula must also be satisfied: (2).

[0032] Determine Width and height After that, each geological parameter data contained in each set of geological parameter data is normalized, and its value range is normalized to between 〔0-1〕. Then the normalized data is mapped to the range of 〔0-255〕, which is expressed as the grayscale value of the grayscale image.

[0033] After mapping The gray values ​​are arranged from top to bottom and from left to right and reconstructed into OK A two-dimensional array of columns, each position in the two-dimensional array represents a pixel position, and the grayscale value of each position represents the grayscale value of the pixel, thereby forming a grayscale image as the corresponding image data. OK In the two-dimensional array of columns, the positions without grayscale values ​​are filled with zeros. Positions need to be filled with zeros.

[0034] Step 3: Use unsupervised clustering methods to classify image data, optimize the classification results, and number the optimized clusters as the final classifications.

[0035] For multiple sets of geological parameter data converted into multiple image data, spectral clustering + K-means clustering method is used for classification. Spectral clustering transforms the clustering problem into a graph partitioning problem by utilizing the graph structure of the data, while K-means clustering performs data grouping by iteratively optimizing the position of the cluster center. Obtaining low-dimensional feature representation through spectral clustering and then performing K-means clustering can effectively process geological parameter data of complex shapes and improve the clustering effect. The step 3 specifically includes the following steps 3.1 to 3.5.

[0036] Step 3.1: Use the Gaussian kernel function to calculate the similarity between every two image data and construct a similarity matrix.

[0037] Each image data in multiple images is taken as a sample, and the Gaussian kernel function is used to calculate the similarity between samples. It can reduce the similarity of samples with a long distance to near zero, thereby highlighting the local structure. The calculation formula of Gaussian kernel similarity is: (3); in, and Represents two different samples; It is a sample and The Euclidean distance between It is a parameter that controls the decay speed of similarity. is the similarity matrix No. Line Column elements.

[0038] The Euclidean distance calculation formula is: (4); in and Respectively represent samples and Middle The gray value of the pixel position.

[0039] Step 3.2: Use the k-nearest neighbor graph to construct a similarity graph corresponding to the similarity matrix, and construct a Laplacian matrix based on the similarity graph.

[0040] According to the calculated similarity matrix Construct a similarity graph using the k-nearest neighbor graph. Specifically, the similarity matrix Each element in is taken as a node, and the similarity graph constructed is to model the local neighborhood relationship between nodes. The line connecting two nodes is taken as an edge, and the result is a directed graph. Because the neighbor relationship is asymmetric, after all nodes are connected, for each node in the directed graph, the k neighbor nodes with the highest structural similarity are found, and then only the similarities of these neighbor nodes are retained, and the other similarities are set to 0 to construct a similarity graph. After the similarity graph is constructed, weights are assigned to the edges according to the similarity of the nodes, and the degree matrix D and adjacency matrix W of the nodes are calculated. The Laplace matrix formula is L=DW. Where L is the Laplace matrix.

[0041] Step 3.3: Perform eigendecomposition on the Laplace matrix L and calculate the pre-calculation of the Laplace matrix feature vector, this The feature vectors are composed as columns The matrix U, each row in the matrix U is a dimensional row vector, total A row vector.

[0042] Step 3.4: The row vectors are clustered using the K-means clustering method to classify the image data converted from all geological parameter data to obtain the cluster distribution, and each cluster represents a new classification.

[0043] Step 3.5: Optimize the clusters contained in the classification results based on the DB index, and number the optimized clusters as the final classifications.

[0044] The number threshold of samples in each classification (cluster) is set to 5. When the number of samples in a cluster is less than or equal to 5, the DB index is calculated using the following formula: (5); in, is the number of clusters, that is, the number of clusters in the classification results; and They represent two different clusters respectively; Cluster The average radius is defined as the average distance from all points in a cluster to the centroid of the cluster; Cluster The average radius is defined as the average distance from all points in a cluster to the centroid of the cluster; and Represents clusters and Cluster The centroid of for and The distance between For each cluster , find the cluster with the largest difference compared to it . The calculated DB index is: the closer the value is to 0, the higher the compactness and the better the separation of the clusters; otherwise, the lower the compactness and the worse the separation of the clusters.

[0045] In this application, the compactness and separation of clusters are evaluated using the DB index. When the value is less than or equal to 0.3, the compactness is considered to be high and the separation is good. When the value is greater than 0.3, the compactness is considered to be low and the separation is poor. When it is less than or equal to 0.3, it indicates that the cluster is too small, the compactness is low and the separation is poor. It is considered to be data noise and needs to be deleted. After deleting all clusters that do not meet the requirements, the optimized clusters are numbered as the final classification. For example, after deleting all clusters that do not meet the requirements, there are still ten optimized clusters left, and these ten clusters are numbered C1~C10 respectively.

[0046] Step 4: Build a data set based on the image data and the corresponding classification number, train the deep learning model, and use it as a geological parameter classification model after training.

[0047] The image data converted from the geological parameter data is used as the input variable, and the classification number corresponding to the geological parameter data is used as the output variable to construct a data set, and the data set is stored in the database.

[0048] In an exemplary embodiment, a MobileNet V2 convolutional neural network model is selected as a deep learning model for training. The MobileNet V2 convolutional neural network model uses an inverted residual structure and a linear layer as the last layer, which has the advantages of higher accuracy and smaller model compared to other deep learning models.

[0049] The entire data set is randomly divided into a training set and a validation set in a ratio of 7:3, and the deep learning model is trained and validated. After the training is completed, it is used as a geological parameter classification model. The input of the geological parameter classification model is image data, and the output is the corresponding classification number.

[0050] Step 5: Process the production dynamic data using the Gram angle field to generate a Gram angle field image.

[0051] The production dynamic data are processed using the Gram angle field, and then converted into a Gram angle field image or a Gram angle difference field image.

[0052] The step 5 specifically includes the following steps 5.1 to 5.4.

[0053] Step 5.1: Represent the production dynamic data in tuple form as ; where each pair of values It refers to time The production data at ; , is the number of sampling points. Then for each pair of values Create a time dependency matrix by transforming time With production data One to one correspondence, and then expressed in the form of a matrix.

[0054] Step 5.2: Output data Zoom to Within the range, get the scaled production data , the calculation formula is as follows: (6); in, Represents all production data in the time correlation matrix A collection of; and Respectively represent sets The maximum and minimum values ​​in . Indicates time The corresponding production data, Indicates that Zoom to Results in range.

[0055] Step 5.3: Express the scaled production data in polar coordinates ,get Polar angle in polar coordinates and polar diameter , the formula is as follows: (7); in, For all A collection of; is a constant factor for the span of the regularized polar coordinate system.

[0056] Step 5.4: Based on polar angle and polar diameter Generate Gram angle field image and Gram angle difference field image.

[0057] Based on polar angle and polar diameter The Gram angle field image and the Gram angle difference field image can be drawn. The corresponding values ​​of the Gram angle field and the Gram angle difference field can be calculated according to the following formula: (8); in and represents the adjacent polar angle; Represents traversal of a collection For each value in ; and They represent the calculated values ​​of the Gram angle field and the Gram angle difference field respectively.

[0058] in The calculation method of is as follows: (9).

[0059] The generated Gram angle field image and Gram angle difference field image are both RGB color images, and the Gram angle field image is used as an example for detailed description in this application. Of course, in practical applications, the production dynamic data can also be converted into a Gram angle difference field image for subsequent processing, and the processing method is the same as that of the Gram angle field image.

[0060] Step 6: The Gram angle field image and the image data converted from the corresponding geological parameter data are stored in correspondence to construct a database.

[0061] All geological parameter data stored in the database are traversed, and the production dynamic data corresponding to each set of geological parameter data is converted into a Gram angle field image or a Gram angle difference field image according to the processing method of step 5. After the conversion is completed, the geological parameter data, image data, production dynamic data, and the Gram angle field image or the Gram angle difference field image are stored one by one.

[0062] Step 7: Convert the geological parameter data to be predicted into image data to be predicted, input it into the geological parameter classification model for classification, and obtain the prediction classification number.

[0063] The geological parameter data to be predicted refers to the geological parameter data that needs to be predicted for a certain reservoir of the current oil and gas reservoir. The geological parameter data to be predicted is converted into corresponding image data according to the processing method of step 2, which is called image data to be predicted. The image data to be predicted is input into the geological parameter classification model trained in step 4 for classification, and the classification number of the classification to which the geological parameter data to be predicted belongs is obtained, which is called the prediction classification number.

[0064] Step 8: Based on the predicted classification number, the image data with the highest structural similarity under the same classification number is retrieved from the database as the target image data.

[0065] Compare the image data to be predicted with all the image data under the prediction classification number in the database, and find the image data with the highest structural similarity as the target image data. The structural similarity calculation method is as follows: (10); in, (11); (12); (13).

[0066] In the formula, and They represent the corresponding pixel points in the two image data to be compared, is the structural similarity of the two contrasting image data. , , Represent the comparison functions of brightness, contrast and structure respectively. , and It is a parameter used to adjust the importance of each part and is usually set to 1. and They are and The average brightness. and They are and The standard deviation of , indicating contrast. yes and , indicating the structural similarity between them. and They are all constants, used to avoid the denominator being zero, and are usually taken as . is the dynamic range of pixel values. is a constant, taking the value hour is 0.01, the value hour is 0.03. is also a constant, usually taken as .

[0067] Combining formulas (10) to (13), we can obtain: (14).

[0068] Step 9: Use the structural similarity method to calculate the difference image between the image data to be predicted and the target image data.

[0069] Using the structural similarity method, by comparing all corresponding pixels in the image data to be predicted and the target image data, the structural similarity data between the image data to be predicted and the target image data is calculated using formula (14). Figure 2 In an exemplary embodiment shown, the structural similarity between the pixel in the 4th row and the 5th column of the image data to be predicted and the target image data is 0.223. The structural similarity data calculated for all pixels are mapped to the range of [0-255] according to a processing method similar to step 2, expressed as the grayscale value of the grayscale image, and then converted into the corresponding image data, thereby constructing a difference image.

[0070] Step 10: Take the Gram angle field image corresponding to the target image data in the database as the image to be corrected, and correct the image to be corrected based on the difference image to obtain a corrected image.

[0071] The database is searched to find the Gram angle field image of the production dynamic data corresponding to the target image data as the image to be corrected. The image to be corrected is corrected using the difference image data as the correction variable. The difference image contains the feature difference between the image data to be predicted and the target image data. The color of the corresponding pixel block in the image to be corrected is adjusted based on the feature difference.

[0072] The step 10 specifically includes the following steps 10.1 and 10.2.

[0073] Step 10.1: Divide the image to be corrected into Pixel blocks.

[0074] The number of pixel blocks in the difference image The image to be corrected is divided into In one exemplary embodiment, ,like Figure 3 As shown, the image to be corrected is divided into Pixel blocks.

[0075] Step 10.2: Based on The grayscale value of the corresponding pixel in the difference image of size is The RGB channel color values ​​of the pixel blocks at corresponding positions in the pixel blocks are adjusted to obtain a corrected image.

[0076] The difference image is Line The gray value of the pixel in the column is The image to be corrected is divided into Each pixel block includes Then for a certain pixel block, the color is adjusted from top to bottom and from left to right. The adjustment method is described as follows.

[0077] For example, for the image to be corrected Line Column pixel blocks, for which pixels, set the adjustment value of the pixel in row 1 and column 1 to , will Line The adjustment value of the pixel in the column is set to 0, and the remaining pixels are interpolated in the order from top to bottom and from left to right to obtain the adjustment values ​​at different pixel positions, which are recorded as Then, for a certain pixel, adjust its color value according to the following formula: (15) in, , , are the original red, green, and blue channel color values ​​of a pixel in the image to be corrected; is the adjustment value of each channel at the pixel position. , , That is, the color value of the red, green, and blue channels of the pixel after adjustment, that is, the red, green, and blue channel color value of the corrected image at the pixel.

[0078] The color of each pixel block in the image to be corrected is adjusted until all pixel blocks are corrected to obtain the corrected image.

[0079] Step 11: Inversely transform the corrected image into predicted production dynamic data.

[0080] The corrected image, that is, the corrected Gram angle field image or the Gram angle difference field image is inversely converted into time series data of predicted production data through the inverse operation of step 5, that is, the production dynamic data predicted based on the geological parameter data to be predicted is obtained.

[0081] Specifically, the corrected image is first converted into RGB color values, then normalized, the polar angle is obtained through the inverse operation of trigonometric functions, the polar diameter is obtained through the corresponding time step and the size of the production data, and then the polar coordinate system data is converted into rectangular coordinate system data to obtain the tuple data of the production data and the corresponding time. It should be noted that during the inverse transformation process, is the yield data of the corresponding pixel block, and Determined by the darkest and lightest pixel blocks in the Gram angle field image or the Gram angle difference field image.

[0082] The above-mentioned dynamic prediction method for the production of fractured oil and gas wells proposed in this application can greatly reduce the calculation time required for production prediction. Compared with the numerical simulation method, the calculation efficiency can be improved by more than 2 orders of magnitude. Moreover, the required calculation methods are all free, and there is no need to pay expensive software usage fees, which greatly reduces the prediction cost.

[0083] In an exemplary embodiment, the present application also provides a computer device, which may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface, and a communication interface. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the method for dynamically predicting the production of a fractured oil and gas well is implemented.

[0084] In an exemplary embodiment, the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for dynamically predicting the production of fractured oil and gas wells is implemented.

[0085] In an exemplary embodiment, the present application also provides a computer program product, including a computer program, which implements the method for dynamically predicting the production of fractured oil and gas wells when executed by a processor.

[0086] It can be understood by a person skilled in the art that all or part of the processes in the above-mentioned embodiment method can be completed by hardware related to computer program instructions, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the process of the embodiment of the above-mentioned method. Among them, any reference to the memory or other medium in each embodiment provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0087] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0088] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0089] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for dynamically predicting the production of a fractured oil and gas well, characterized in that: include: Obtain multiple sets of geological parameter data and corresponding production dynamic data of different types of reservoirs in different oil and gas reservoirs; The dynamic production data refers to production data at different times; Convert each set of geological parameter data into corresponding image data; The image data is classified using an unsupervised clustering method, and the classification results are optimized, and the optimized clusters are numbered as the final classifications; A data set is constructed based on the image data and the corresponding classification number, and a deep learning model is trained. After the training is completed, the model is used as a geological parameter classification model; the input of the geological parameter classification model is the image data, and the output is the corresponding classification number; The production dynamic data is processed using the Gram angle field to generate a Gram angle field image; The Gram angle field image and the image data converted from the corresponding geological parameter data are stored correspondingly to construct a database; The geological parameter data to be predicted is converted into image data to be predicted, and input into the geological parameter classification model for classification to obtain a prediction classification number; Based on the predicted classification number, the image data with the highest structural similarity under the same classification number is retrieved from the database as the target image data; Calculate the difference image between the image data to be predicted and the target image data using a structural similarity method; The Gram angle field image corresponding to the target image data in the database is used as the image to be corrected, and the image to be corrected is corrected based on the difference image to obtain a corrected image; The corrected image is reversely converted into predicted production dynamics data.

2. The method for dynamic prediction of production of fractured oil and gas wells according to claim 1, characterized in that: The method of obtaining multiple sets of geological parameter data of different types of reservoirs in different oil and gas reservoirs specifically includes: If the reservoir type is an oil layer, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, oil, gas and water saturation, oil, gas and water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability and relative permeability curve data; PVT data includes pressure, volume and temperature data; If the reservoir type is a tight gas reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability, and relative permeability curve data; If the reservoir type is a shale gas reservoir or a coalbed methane reservoir, each set of geological parameter data obtained includes thickness, porosity, permeability, formation pressure, gas-water saturation, gas-water PVT data, well spacing, fracture penetration ratio, fracture equivalent permeability, relative permeability curve, Langmuir volume and Langmuir pressure data.

3. The method for dynamic prediction of production of fractured oil and gas wells according to claim 2, characterized in that: The step of converting each set of geological parameter data into corresponding image data specifically includes: According to the number of types of geological parameter data contained in each set of geological parameter data Determines the width of the image data to be converted and height ;in and , To fix the filling is an integer; Each geological parameter data contained in each set of geological parameter data is normalized to a value range between [0-1]; Map the normalized data to the range of [0-255], expressed as the grayscale value of the grayscale image; Arrange the mapped grayscale values ​​from top to bottom and from left to right and reconstruct them into OK A two-dimensional array of columns, each position in the two-dimensional array represents a pixel position, the grayscale value of each position represents the grayscale value of the pixel, and positions without grayscale values ​​are filled with zeros to form a grayscale image as the corresponding image data.

4. The method for dynamic prediction of production of fractured oil and gas wells according to claim 3, characterized in that: The unsupervised clustering method is used to classify the image data, and the classification results are optimized, and the optimized clusters are numbered as the final classification, which specifically includes: Use the Gaussian kernel function to calculate the similarity between every two image data and construct a similarity matrix; Use the k-nearest neighbor graph to construct a similarity graph corresponding to the similarity matrix, and construct a Laplacian matrix based on the similarity graph; Perform eigendecomposition on the Laplace matrix and calculate the pre- feature vector, this The feature vectors are composed as columns The matrix U, each row in the matrix U is a dimensional row vector, total row vectors; To this The row vectors are clustered using the K-means clustering method to obtain the distribution of clusters, where each cluster represents a classification; The clusters contained in the classification results are optimized based on the DB index, and the optimized clusters are numbered as the final classifications.

5. The method for dynamic prediction of production of fractured oil and gas wells according to claim 1, characterized in that: The production dynamic data is processed using the Gram angle field to generate a Gram angle field image, specifically including: The dynamic data of production is expressed in the form of tuples as ; where each pair of values Refers to time The production data at ; , is the number of sampling points; The production data Zoom to Within the range, get the scaled production data ; Representing scaled production data in polar coordinates ,get Polar angle in polar coordinates and polar diameter ; Based on polar angle and polar diameter Generate a Gram angular field image.

6. The method for dynamic prediction of production of fractured oil and gas wells according to claim 1, characterized in that: The method of using the structural similarity method to calculate the difference image between the image data to be predicted and the target image data specifically includes: Using a structural similarity method to calculate structural similarity data between the image data to be predicted and the target image data; The structural similarity data is converted into corresponding image data as a difference image.

7. The method for dynamic prediction of production of fractured oil and gas wells according to claim 3, characterized in that: The method of correcting the image to be corrected based on the difference image to obtain the corrected image specifically includes: Divide the image to be corrected into Pixel blocks; based on The grayscale value of the corresponding pixel in the difference image of size is The RGB channel color values ​​of the pixel blocks at corresponding positions in the pixel blocks are adjusted to obtain a corrected image.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for dynamically predicting the production of a fractured oil and gas well according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for dynamically predicting the production of a fractured oil and gas well according to any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for dynamically predicting the production of a fractured oil and gas well according to any one of claims 1 to 7 is implemented.

Images