A method and device for predicting oil and gas fracturing production based on geological and engineering data

Abstract

This specification relates to a method and device for predicting oil and gas fracturing production based on geological-engineering data. The method comprises obtaining pumping data and well logging data of a target well; performing two-dimensional conversion on the pumping data and well logging data to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix; inputting the two-dimensional pumping feature matrix into a first convolutional neural network model for convolution and pooling calculations to obtain a construction quality evaluation index; inputting the two-dimensional well logging feature matrix into a second convolutional neural network model for convolution and pooling calculations to obtain formation property indicators and reservoir fluid indicators; and inputting the construction quality evaluation index, formation property indicators, and reservoir fluid indicators into a fully connected layer to output the production prediction result. This specification further uses well logging data along the well depth direction and pumping data based on time series to predict post-fracturing production, thereby ensuring the objectivity and accuracy of the fracturing data.

Description

Technical Field

[0001] This specification relates to the field of oil and gas exploration, and in particular to a method and device for predicting oil and gas fracturing production based on geological and engineering data. Background Art

[0002] Currently, most methods for predicting oil and gas fracturing production in undeveloped wells only use average geological data for prediction. However, the conditions that affect fracturing production in oil and gas fracturing technology are complex. Using only average geological data cannot accurately predict the oil and gas fracturing production of undeveloped wells, which will not be able to accurately guide the subsequent development work of undeveloped wells.

[0003] How to accurately predict the oil and gas fracturing production of unexploited wells is currently an urgent problem to be solved. Summary of the Invention

[0004] To solve the problems existing in the prior art, the embodiments of this specification provide a method and device for predicting oil and gas fracturing production based on geological and engineering data, and further use well logging data along the well depth direction and time-series-based pumping data to predict post-fracturing production, thereby ensuring the objectivity and accuracy of fracturing data.

[0005] In order to solve any of the above technical problems, the specific technical solutions of this specification are as follows:

[0006] The embodiments of this specification provide a method for predicting oil and gas fracturing production based on geological and engineering data, including:

[0007] Acquiring pumping data of a target well and well logging data of the formation where the target well is located;

[0008] Performing two-dimensional conversion on the pumping data and the well logging data respectively to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix;

[0009] Inputting the pumping two-dimensional feature matrix into the trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index;

[0010] Inputting the well logging two-dimensional feature matrix into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators;

[0011] The construction quality evaluation index, formation physical property index and reservoir fluid index are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the target well through the activation function.

[0012] Furthermore, the pumping data is subjected to a two-dimensional conversion to obtain a two-dimensional pumping feature matrix including:

[0013] Dividing the pumping data according to working conditions to obtain pumping data under multiple stages;

[0014] The pumping data corresponding to each stage are converted into two dimensions to obtain the two-dimensional pumping feature matrix corresponding to each stage.

[0015] Furthermore, the first convolutional neural network model includes multiple input channels;

[0016] The pumping two-dimensional feature matrix is input into the trained first convolutional neural network model for convolution and pooling calculations to obtain the construction quality evaluation index further including:

[0017] The two-dimensional pumping feature matrix corresponding to each stage is input into the first convolutional neural network model through the corresponding input channel of the first convolutional neural network model, and convolution and pooling calculations are performed to obtain the construction quality evaluation index.

[0018] Furthermore, performing a two-dimensional conversion on the well logging data to obtain a two-dimensional feature matrix of the well logging further includes:

[0019] Analyzing the logging data to obtain index parameters of the formation, wherein the index parameters include permeability, brittleness index, Young's modulus and oil-water classification;

[0020] The index parameters are converted into two dimensions to obtain a two-dimensional well logging characteristic matrix corresponding to the index parameters.

[0021] Furthermore, the second convolutional neural network model includes at least two input channels;

[0022] Inputting the two-dimensional feature matrix of well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators further includes:

[0023] The logging data and index parameters are respectively input into the second convolutional neural network model through the corresponding input channels of the second convolutional neural network model, and convolution and pooling calculations are performed to obtain the formation physical property index and reservoir fluid index.

[0024] Furthermore, the steps of training the first convolutional neural network model and the second convolutional neural network model include:

[0025] Acquire pumping data of a produced well as a training pumping data set, and acquire well logging data of a formation where the produced well is located as a training well logging data set;

[0026] Performing two-dimensional conversion on the data in the training pumping data set and the training well logging data set respectively to obtain a training pumping two-dimensional feature matrix and a training well logging two-dimensional feature matrix;

[0027] Inputting the training pump injection two-dimensional feature matrix into the first convolutional neural network model to be trained to perform convolution and pooling calculations to obtain a predicted construction quality evaluation index;

[0028] Inputting the training well logging two-dimensional feature matrix into the second convolutional neural network model to be trained to perform convolution and pooling calculations to obtain predicted formation physical property indicators and predicted reservoir fluid indicators;

[0029] The predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well through the activation function. The first convolutional neural network model and the second convolutional neural network model are iteratively trained according to the production prediction result of the produced well and the actual production of the produced well until the difference between the production prediction result and the actual production of the produced well meets the predetermined requirements, thereby obtaining the trained first convolutional neural network model and the second convolutional network model.

[0030] Furthermore, iteratively training the first convolutional neural network model and the second convolutional neural network model according to the production prediction result of the produced well and the actual production of the produced well further includes:

[0031] During any iterative training process, the model parameters of either the first convolutional neural network model or the second convolutional neural network model are frozen, the model parameters of the unfrozen convolutional neural network model are adjusted, and then the training pump injection two-dimensional feature matrix and the training logging two-dimensional feature matrix are input into the corresponding convolutional neural network model for calculation to obtain the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index of this iteration, and then the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index obtained in this iteration are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well in this iteration through the activation function.

[0032] On the other hand, the embodiments of this specification also provide an oil and gas fracturing production prediction device based on geological and engineering data, comprising:

[0033] A data acquisition unit, configured to acquire pumping data of a target well and well logging data of a formation where the target well is located;

[0034] a data two-dimensional conversion unit, configured to perform two-dimensional conversion on the pumping data and the well logging data to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix;

[0035] A first calculation unit is used to input the pumping two-dimensional feature matrix into a trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index;

[0036] A second computing unit inputs the two-dimensional feature matrix of well logging into a trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators;

[0037] The fully connected layer calculation unit is used to input the construction quality evaluation index, formation physical property index and reservoir fluid index into the fully connected layer, and the fully connected layer outputs the production prediction result of the target well through the activation function.

[0038] On the other hand, an embodiment of this specification further 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 implements the above method when executing the computer program.

[0039] On the other hand, an embodiment of this specification further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0040] Using the embodiments of this specification, first, the pumping data of the target well for oil and gas fracturing production prediction and the well logging data of the formation where the target well is located are obtained. The pumping data is time-series based and refers to the process of injecting liquids and / or solids into the ground. Pumping data, as long as it records the data from the start and end of liquid and / or solid pumping, reflects the quality of the construction process. Well logging data is depth-based and measures the geology and reservoir fluid properties of the formation along the depth of the wellbore, reflecting the oil production potential of the formation. Then, the pumping data and well logging data are two-dimensionally transformed to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix, thereby achieving dimensional expansion and feature extraction. Because the pumping data is time-series based and the well logging data is depth-based, the embodiments of this specification establish two convolutional neural network models to calculate the pumping data and well logging data respectively, achieving layered input. The pumping data and well logging data are calculated in parallel in their respective convolutional neural network models. Finally, the output results of the two convolutional neural network models are input into a fully connected layer, which uses an activation function to obtain the target well's production prediction result.

[0041] The embodiments of this specification implement layered input and parallel calculation of two types of data (pumping data and logging data) that have different properties but can both reflect the oil production of the target well, and ultimately output the predicted oil and gas fracturing production of the target well through the fully connected layer, thereby achieving accurate prediction of the oil and gas production of the target well (i.e., unproduced well).

[0042] The innovation of the embodiments of this specification lies in: providing the input of logging and pumping data into an intelligent model. On the basis of conventional prediction using only average geological data, the logging and pumping data along the well depth direction and time series-based pumping data are further used to predict the post-fracturing production, thereby ensuring the objectivity and accuracy of the fracturing data. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 The figure shows a schematic diagram of an implementation system of a method for predicting oil and gas fracturing production based on geological and engineering data in an embodiment of this specification;

[0045] Figure 2 The figure shows a flow chart of a method for predicting oil and gas fracturing production based on geological and engineering data in an embodiment of this specification;

[0046] Figure 3 The figure shows a flow chart of performing two-dimensional conversion on the pumping data to obtain a two-dimensional pumping feature matrix in an embodiment of this specification;

[0047] Figure 4 The figure shows a flow chart of performing two-dimensional conversion on the well logging data to obtain a two-dimensional feature matrix of the well logging data in an embodiment of this specification;

[0048] Figure 5 The figure shows a schematic diagram of the structure of an oil and gas fracturing production prediction device based on geological and engineering data in an embodiment of this specification;

[0049] Figure 6 The figure shows a schematic diagram of the structure of a computer device in an embodiment of this specification.

[0050]

Description of the accompanying drawings

[0051] 101. Terminal;

[0052] 102. Server;

[0053] 501. Data acquisition unit;

[0054] 502. Data two-dimensional conversion unit;

[0055] 503. First computing unit;

[0056] 504, second computing unit;

[0057] 505, fully connected layer computing unit;

[0058] 602. Computer equipment;

[0059] 604, processor;

[0060] 606, memory;

[0061] 608, driving mechanism;

[0062] 610, input / output module;

[0063] 612. Input devices;

[0064] 614. Output device;

[0065] 616. Presentation equipment;

[0066] 618. Graphical User Interface;

[0067] 620, network interface;

[0068] 622, communication link;

[0069] 624. Communication bus. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the embodiments described are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this specification.

[0071] It should be noted that the terms "first," "second," and the like in the description and claims of this specification and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of this specification described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0072] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0073] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of relevant laws and regulations.

[0074] like Figure 1 The diagram shows a system for implementing a method for predicting oil and gas fracturing production based on geological and engineering data in an embodiment of the present specification, including a terminal 101 and a server 102. The terminal 101 and the server 102 can communicate with each other via a network, which can include a local area network (LAN), a wide area network (WAN), the Internet, or a combination thereof, and is connected to a website, a user device (e.g., a computing device), and a back-end system.

[0075] The user first inputs the pumping data, well logging data, and oil and gas fracturing production of the mined wells into server 102 via terminal 101. Server 102 then trains a first convolutional neural network model and a second convolutional neural network model based on the pumping data, well logging data, and oil and gas fracturing production of the mined wells. When predicting the oil and gas fracturing production of unmined wells, the user inputs the pumping data and well logging data of the unmined wells into server 102 via terminal 101. Server 102 uses the trained first convolutional neural network model and the trained second convolutional neural network model to calculate the pumping data and well logging data of the unmined wells, respectively, to obtain a predicted oil and gas fracturing production result for the unmined wells. The server then provides the predicted oil and gas fracturing production result for the unmined wells to the user via terminal 101 for reference, thereby guiding the user to perform oil and gas fracturing production on the unmined wells.

[0076] Alternatively, the server 102 may be a node of a cloud computing system (not shown), or each server may be a separate cloud computing system including multiple computers interconnected by a network and operating as a distributed processing system.

[0077] In addition, it should be noted that Figure 1 What is shown is only one application environment provided by the embodiment of this specification. In actual application, other application environments may also be included, and this specification does not limit them.

[0078] Pumping data refers to the process of injecting liquids and / or solids into the ground. Pumping data, which records data from the start to the end of liquid and / or solid pumping, reflects the quality of the operation process. Well logging data is depth-based and measures the geology and reservoir fluid properties of the formation near the wellbore along the depth of the wellbore, reflecting the formation's oil production potential. Both pumping data and well logging data can reflect the oil and gas fracturing production of undeveloped wells. However, because pumping data is time-series based and well logging data is depth-based, current oil and gas fracturing production prediction methods cannot predict oil and gas fracturing production of undeveloped wells based on both pumping data and well logging data.

[0079] In order to solve the problems existing in the prior art, the embodiments of this specification provide an oil and gas fracturing production prediction method based on geological and engineering data, and further use well logging data along the well depth direction and time series-based pumping data to predict the post-fracturing production, thereby ensuring the objectivity and accuracy of the fracturing data. Figure 2 The figure shows a flow chart of a method for predicting oil and gas fracturing production based on geological and engineering data in an embodiment of this specification. This figure describes the process of predicting oil and gas fracturing production based on pumping data and logging data, but it may include more or fewer operating steps based on conventional or non-creative work. The order of steps listed in the embodiment is only one way of executing the steps among many, and does not represent the only execution order. When the system or device product is actually executed, it can be executed in sequence or in parallel according to the method shown in the embodiment or the accompanying drawings. Specifically, Figure 2 As shown, the method may be performed by the server 102, and may include:

[0080] Step 201: Acquire pumping data of a target well and well logging data of the formation where the target well is located;

[0081] Step 202: performing two-dimensional conversion on the pumping data and the well logging data to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix;

[0082] Step 203: inputting the pumping two-dimensional feature matrix into the trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index;

[0083] Step 204: inputting the two-dimensional feature matrix of well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators;

[0084] Step 205: Input the construction quality evaluation index, formation physical property index, and reservoir fluid index into the fully connected layer, and the fully connected layer outputs the production prediction result of the target well through an activation function.

[0085] Using the embodiments of this specification, first, the pumping data of the target well for oil and gas fracturing production prediction and the well logging data of the formation where the target well is located are obtained. The pumping data is time-series based and refers to the process of injecting liquids and / or solids into the ground. Pumping data, as long as it records the data from the start and end of liquid and / or solid pumping, reflects the quality of the construction process. Well logging data is depth-based and measures the geology and reservoir fluid properties of the formation along the depth of the wellbore, reflecting the oil production potential of the formation. Then, the pumping data and well logging data are two-dimensionally transformed to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix, thereby achieving dimensional expansion and feature extraction. Because the pumping data is time-series based and the well logging data is depth-based, the embodiments of this specification establish two convolutional neural network models to calculate the pumping data and well logging data respectively, achieving layered input. The pumping data and well logging data are calculated in parallel in their respective convolutional neural network models. Finally, the output results of the two convolutional neural network models are input into a fully connected layer, which uses an activation function to obtain the target well's production prediction result.

[0086] The embodiments of this specification implement layered input and parallel calculation of two types of data (pumping data and logging data) that have different properties but can both reflect the oil production of the target well, and ultimately output the predicted oil and gas fracturing production of the target well through the fully connected layer, thereby achieving accurate prediction of the oil and gas production of the target well (i.e., unproduced well).

[0087] In this embodiment, a first convolutional neural network model is used to calculate a two-dimensional feature matrix of pumping data after two-dimensionalization to obtain construction evaluation indicators. A second convolutional neural network model is used to calculate a two-dimensional feature matrix of well logging data after two-dimensionalization to obtain formation physical property indicators and reservoir fluid indicators. In this embodiment, pumping data includes but is not limited to oil pressure, casing pressure, displacement, and sand ratio, while well logging data includes but is not limited to natural potential, natural gamma, acoustic transit time, compensated neutron, shale content, lithologic logging, and gas logging. The target well's production prediction result is a cumulative production indicator, where cumulative production types include oil production, water production, gas production, and liquid production, and cumulative production days include but are not limited to 30 days, 60 days, 180 days, and 330 days. The cumulative production indicator can be converted into a dimensionless indicator using a specific geological engineering parameter as a unit, including but not limited to section length, permeability, oil saturation, sand addition amount, and construction cost. Bucket classification is used to categorize the production indicator into high, medium, and low production intervals to reduce the number of labels and improve prediction accuracy.

[0088] According to one embodiment of this specification, the steps of training the first convolutional neural network model and the second convolutional neural network model include:

[0089] Acquire pumping data of a produced well as a training pumping data set, and acquire well logging data of a formation where the produced well is located as a training well logging data set;

[0090] Performing two-dimensional conversion on the data in the training pumping data set and the training well logging data set respectively to obtain a training pumping two-dimensional feature matrix and a training well logging two-dimensional feature matrix;

[0091] Inputting the training pump injection two-dimensional feature matrix into the first convolutional neural network model to be trained to perform convolution and pooling calculations to obtain a predicted construction quality evaluation index;

[0092] Inputting the training well logging two-dimensional feature matrix into the second convolutional neural network model to be trained to perform convolution and pooling calculations to obtain predicted formation physical property indicators and predicted reservoir fluid indicators;

[0093] The predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well through the activation function. The first convolutional neural network model and the second convolutional neural network model are iteratively trained according to the production prediction result of the produced well and the actual production of the produced well until the difference between the production prediction result and the actual production of the produced well meets the predetermined requirements, thereby obtaining the trained first convolutional neural network model and the second convolutional network model.

[0094] Furthermore, during model training, since the input includes two inputs, the parallel convolutional neural networks can be trained differently. For example, the convolutional neural network for well logging data input can be frozen, and only the convolutional neural network for pumping data input can be trained to achieve the purpose of modular input, so that the two convolutional neural networks can be trained separately or simultaneously. Specifically, iteratively training the first convolutional neural network model and the second convolutional neural network model based on the production prediction results of the produced well and the actual production of the produced well further includes:

[0095] During any iterative training process, the model parameters of either the first convolutional neural network model or the second convolutional neural network model are frozen, the model parameters of the unfrozen convolutional neural network model are adjusted, and then the training pump injection two-dimensional feature matrix and the training logging two-dimensional feature matrix are input into the corresponding convolutional neural network model for calculation to obtain the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index of this iteration, and then the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index obtained in this iteration are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well in this iteration through the activation function.

[0096] In the embodiments of this specification, before the two-dimensional conversion of the pumping data and the logging data, the pumping data and the logging data are also subjected to data smoothing, data noise reduction, data interpolation, precision unification, parameter conversion, result interpretation, code conversion and other processing to standardize the data and provide a data basis for subsequent steps.

[0097] Methods for preprocessing the pumping data and logging data include but are not limited to data smoothing, data noise reduction, data interpolation, precision unification, parameter conversion, result interpretation, and code conversion, so as to obtain a characteristic parameter set of the target well, wherein the characteristic parameter set includes values of multiple characteristic parameters, including but not limited to: reservoir thickness, initial formation pressure, matrix permeability, matrix porosity, permeability of the fractured volume area, porosity of the fractured volume area, fracture permeability, fracture segment length, number of segments, number of clusters, actual segment spacing, actual cluster spacing, front fluid ratio, and slickwater ratio. The characteristic parameters are geological parameters and engineering parameters that affect the predicted production. The data sources include but are not limited to field-collected data, data simulation data, etc. The collected data is marked as data set D. The fracturing data in data set D includes geological data, engineering data, and production indicators.

[0098] In the embodiments of this specification, methods including but not limited to the Gram angular field method, Markov transition field (MTF), recursive graph, short-time Fourier transform (STFT), etc. are used to perform two-dimensional conversion on the pumping data, and convert the sequence data of a single parameter into a two-dimensional matrix image. The conversion parameters include but are not limited to oil pressure, casing pressure, displacement, sand ratio, etc.

[0099] According to one embodiment of this specification, Figure 3 As shown, the pumping data is two-dimensionally transformed to obtain a two-dimensional pumping feature matrix including:

[0100] Step 301: Divide the pumping data according to working conditions to obtain pumping data in multiple stages;

[0101] Step 302: Perform two-dimensional conversion on the pumping data corresponding to each stage to obtain a two-dimensional pumping feature matrix corresponding to each stage.

[0102] Before performing two-dimensional conversion, the pumping data needs to be further divided according to the working conditions. The division stages include but are not limited to ball injection, sealing, temporary plugging, crack expansion, etc. The pumping data are converted into independent pumping two-dimensional feature matrices according to different working conditions and different parameters.

[0103] When transforming the pumping data using methods such as the Gram angular field method, a two-dimensional matrix image is obtained, where each dimension can represent a different feature or attribute (time, frequency, pumping data feature or attribute, etc.). The choice of these dimensions depends on the purpose of the analysis and the specific problem area of interest. These features or attributes include:

[0104] 1. Time dimension: If the pumping data is sampled in time, then one of the dimensions may represent time, that is, each pixel corresponds to the data at a specific time point.

[0105] 2. Frequency dimension: If frequency domain transformation or feature extraction techniques are used, then another dimension may represent frequency, that is, each pixel corresponds to data at a specific frequency.

[0106] 3. Pumping data features: Pumping data features may include pressure, flow rate, temperature, etc. Each dimension may represent a different pumping data feature, for example, one dimension may represent pressure and another dimension may represent flow rate.

[0107] 4. Spatial or location information of pumping data: If the pumping data are collected in space, then it may also contain information about the location or spatial distribution of the data collection.

[0108] 5. Additional features or attributes: Other features or attributes related to the pumping data may also be considered, such as weather conditions, geological properties of the well, etc.

[0109] In the embodiment of this specification, the first convolutional neural network model includes multiple input channels;

[0110] The pumping two-dimensional feature matrix is input into the trained first convolutional neural network model for convolution and pooling calculations to obtain the construction quality evaluation index further including:

[0111] The two-dimensional pumping feature matrix corresponding to each stage is input into the first convolutional neural network model through the corresponding input channel of the first convolutional neural network model, and convolution and pooling calculations are performed to obtain the construction quality evaluation index.

[0112] In the embodiments of this specification, methods including but not limited to the Gram angle field method, Markov transition field (MTF), recursion graph, short-time Fourier transform (STFT) and the like are used to perform two-dimensional conversion on the logging data, converting the sequence data of a single parameter into a two-dimensional matrix image. The conversion parameters include but are not limited to natural potential, natural gamma, acoustic wave time difference, compensated neutron, mud content and lithologic logging, gas logging, etc.

[0113] The logging data are further converted and interpreted to obtain the results of permeability, brittleness index, Young's modulus, oil-water classification, etc., and the results are also converted into two dimensions. Specifically, Figure 4 As shown, performing two-dimensional conversion on the well logging data to obtain a two-dimensional feature matrix of the well logging further includes:

[0114] Step 401: Analyze the logging data to obtain index parameters of the formation, wherein the index parameters include permeability, brittleness index, Young's modulus, and oil-water classification;

[0115] Step 402: Perform two-dimensional conversion on the index parameters to obtain a two-dimensional well logging characteristic matrix corresponding to the index parameters.

[0116] When a one-dimensional vector of well log data is converted into a two-dimensional matrix using the Gram angular field method, the values of different well logs are arranged in rows or columns, forming a matrix. In such an image, each dimension typically represents a different well log or attribute, and its meaning may vary depending on the application context and requirements. The following are some possible meanings and application scenarios:

[0117] 1. Dimensional expansion: Converting a one-dimensional vector into a two-dimensional matrix can achieve dimensional expansion. This can be helpful for analyzing and solving certain problems. For example, in some data processing and machine learning tasks, converting one-dimensional data into a two-dimensional matrix can provide more information or more flexible representations.

[0118] 2. Feature Extraction: By converting a one-dimensional vector into a two-dimensional matrix using the Gram Angular Field method, we can exploit the orthogonal properties of the matrix to extract features. This orthogonal property helps capture structural and correlation information in the data. This is very useful for certain pattern recognition, image processing, and signal processing tasks.

[0119] 3. Data compression: After a one-dimensional vector is converted to a two-dimensional matrix, the sparsity or other properties of the matrix can be exploited to achieve data compression. This can save space and bandwidth when storing and transmitting data.

[0120] 4. Data Visualization: Converting one-dimensional vectors into two-dimensional matrices makes data visualization easier. In data visualization tasks, two-dimensional matrices can more easily represent and display relationships and patterns between data.

[0121] 5. Data Analysis and Modeling: After converting a one-dimensional vector into a two-dimensional matrix, you can leverage the properties of matrices to perform more complex data analysis and modeling. For example, you can apply matrix decomposition, matrix operations, and statistical methods for data processing, pattern recognition, or machine learning.

[0122] Furthermore, well logging data describes the unique formation physical properties and reservoir fluid characteristics of each well, and pumping data is an important influencing factor on the expansion of fractures caused by construction and the communication of fracture networks. However, the data volume is lengthy and noisy, so it is necessary to reduce noise and increase dimension of the data, and make it self-orthogonal to highlight the important features of certain parts.

[0123] It should be noted that the Gram angle field method can be used to generate 2D images of well logging data, and the Markov transition field method can be used to generate 2D images of injection data. However, different 2D conversion methods are used for different situations and data types, including but not limited to those mentioned in the examples of this specification.

[0124] Exemplarily, the steps of the Gram's angle field method include:

[0125] (1) Aggregate the time series to reduce the size by taking the average of each M point;

[0126] (2) Normalize the data to [-1, 1]. Normalization methods include but are not limited to min-max normalization, z-score normalization, and decimal scaling normalization.

[0127] Min-max is also called deviation standardization, which is a linear transformation of the original data so that the result falls into the interval [0,1], where max is the maximum value of the sample data and min is the minimum value of the sample data. n Perform the transformation:

[0128]

[0129] Then the new sequence y1,y2,……,y n ∈[0,1] and dimensionless. General data can be normalized first when needed.

[0130] Z-score normalization, also known as standard deviation normalization, is the most common normalization method. Its advantages are simple algorithm, no impact on data magnitude, and easy comparison of results. The specific steps are as follows:

[0131] (1) Calculate the arithmetic mean (mathematical expectation) x of each variable (indicator) i and Standards i ;

[0132] (2) Standardization: ij =(x ij -x i ) / s i

[0133] Where: z ij is the standardized variable value; x ijis the actual variable value.

[0134] (3) Swap the positive and negative signs before the inverse indicator. The normalized variable value fluctuates around 0. A value greater than 0 indicates that it is above the average level, and a value less than 0 indicates that it is below the average level.

[0135] The z-score normalization formula is as follows:

[0136] For the sequence x1,x2,……,x n Perform the transformation:

[0137]

[0138] in,

[0139] Then the new sequence y1,y2,……,y n has a mean of 0 and a variance of 1 and is dimensionless.

[0140] After normalizing the data to [-1, 1], assuming the data type is X = {x1, x2, ..., x N}, the standardized value is recorded as The normalized Gram matrix is:

[0141]

[0142] Here φi,j represents the angle between vector i and vector j.

[0143] (4) Switch to the polar coordinate system and provide the angle value as the cosine of the angle:

[0144]

[0145]

[0146] Among them, t i ∈N represents the point x i timestamp, and N is the number of all time points contained in the time series data.

[0147] It should be noted that different normalized scaling scales have different angular ranges after mapping to the polar coordinate system. The [0,1] normalization corresponds to the cosine function range of [0,Π / 2], while the [-1,1] normalization corresponds to the cosine function range of [0,Π];

[0148] (5) Generate GASF / GADF. The Gram angle field can be understood as an inner product with a penalty term. Each pair of values is added (subtracted), and then the cosine value is taken and summed.

[0149]

[0150]

[0151] Where I is the unit row vector [1,1,…,1], yes The derivative of GASF and GADF is used to directly image the values into grayscale images or map them into pseudo-color images, that is, to realize the conversion of one-dimensional time series signals into images;

[0152] The Markov migration field steps include:

[0153] (1) First, the sequence data (length n) is divided into Q bins (similar to quantiles) according to its value range. Each data point i belongs to a unique qi (∈{1,2,…,Q});

[0154] (2) Construct a Markov transfer matrix with a matrix size of [Q,Q], where the Markov transfer matrix W[i,j] is determined by the frequency with which the data in qi is adjacent to the data in qj, and its calculation formula is:

[0155]

[0156] Among them, w i,j Represents an element in the Markov transition matrix.

[0157] (3) Construct a Markov transition field M, where the value of M[i,j] is W[qi,qj]:

[0158]

[0159] To improve efficiency, we try to reduce the size of M, grid M, and then replace the subgraphs in each grid with their average value.

[0160] In this embodiment of the present specification, the second convolutional neural network model includes at least two input channels;

[0161] Inputting the two-dimensional feature matrix of well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators further includes:

[0162] The logging data and index parameters are respectively input into the second convolutional neural network model through the corresponding input channels of the second convolutional neural network model, and convolution and pooling calculations are performed to obtain the formation physical property index and reservoir fluid index.

[0163] In the embodiment of this specification, the two-dimensional matrix obtained by converting the well logging and its index parameters is input one, and the two-dimensional matrix obtained by converting the pumping data and its operating condition annotations is input two. Input one and input two are respectively passed through two different convolutional neural networks, merged and output to the same fully connected layer, and the fully connected layer outputs the production prediction result through the activation layer. The fixed indicator for production prediction is the cumulative oil production per unit section in predetermined days, where the predetermined days include but are not limited to 30 days, 60 days, 180 days, and 330 days.

[0164] It can be understood that for well logging data and pumping data converted into two-dimensional matrix images using methods including but not limited to the Gram angular field method, Markov transition field (MTF), recursion graph, short-time Fourier transform (STFT), etc., each well will produce multiple images, and these two-dimensional matrix images are difficult to merge at the current stage, and it is not possible to achieve one two-dimensional matrix image for each well. In this case, the convolutional neural network supports simultaneous input of multiple channels into the model for model training and production prediction. Multi-channel input effectively solves the problem of simultaneously inputting multiple two-dimensional matrix images corresponding to a well into the model.

[0165] During the model training phase, the model is iterated multiple times using machine learning training methods to compare the differences between the predicted yield indicators and the actual yield indicators. A loss function is constructed to correct the internal weights of the model. Multiple sets of training are performed simultaneously to search for the optimal hyperparameter combination.

[0166] Based on the same inventive concept, the embodiments of this specification also provide an oil and gas fracturing production prediction device based on geological and engineering data, such as Figure 5 As shown, including:

[0167] The data acquisition unit 501 is used to acquire the pumping data of the target well and the well logging data of the formation where the target well is located;

[0168] A data two-dimensional conversion unit 502 is used to perform two-dimensional conversion on the pumping data and the well logging data to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix;

[0169] A first calculation unit 503 is used to input the pumping two-dimensional feature matrix into the trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index;

[0170] The second calculation unit 504 inputs the two-dimensional feature matrix of the well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators;

[0171] The fully connected layer calculation unit 505 is used to input the construction quality evaluation index, formation physical property index and reservoir fluid index into the fully connected layer, and the fully connected layer outputs the production prediction result of the target well through the activation function.

[0172] The beneficial effects achieved by the above-mentioned device are consistent with the beneficial effects achieved by the above-mentioned method, and will not be described in detail in the embodiments of this specification.

[0173] like Figure 6 As shown, a computer device provided in an embodiment of the present invention is provided. The apparatus described in this specification may be a computer device in this embodiment, which executes the method described above. The computer device 602 may include one or more processors 604, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The computer device 602 may also include any memory 606 for storing any type of information, such as code, settings, data, etc. For example, without limitation, the memory 606 may include any one or more combinations of the following: any type of RAM, any type of ROM, a flash memory device, a hard disk, an optical disk, etc. More generally, any memory may use any technology to store information. Furthermore, any memory may provide volatile or non-volatile retention of information. Furthermore, any memory may represent a fixed or removable component of the computer device 602. In one embodiment, when the processor 604 executes associated instructions stored in any memory or combination of memories, the computer device 602 may perform any operation of the associated instructions. The computer device 602 also includes one or more drive mechanisms 608, such as a hard disk drive mechanism, an optical disk drive mechanism, etc., for interacting with any memory.

[0174] The computer device 602 may also include an input / output module 610 (I / O) for receiving various inputs (via input devices 612) and for providing various outputs (via output devices 614). A specific output mechanism may include a presentation device 616 and an associated graphical user interface (GUI) 618. In other embodiments, the input / output module 610 (I / O), input devices 612, and output devices 614 may not be included, and the computer device 602 may simply be a computer device in a network. The computer device 602 may also include one or more network interfaces 620 for exchanging data with other devices via one or more communication links 622. One or more communication buses 624 couple the components described above together.

[0175] The communication link 622 may be implemented in any manner, for example, via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 622 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0176] The embodiments of this specification also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0177] The embodiments of this specification also provide a computer-readable instruction, wherein when a processor executes the instruction, the program therein causes the processor to execute the above method.

[0178] It should be understood that in the various embodiments of this specification, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this specification.

[0179] It should also be understood that in the embodiments of this specification, the term "and / or" is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Furthermore, the character " / " in this specification generally indicates that the associated objects are in an "or" relationship.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this specification can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this specification.

[0181] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0182] In the several embodiments provided in this specification, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or can be an electrical, mechanical or other form of connection.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this specification.

[0184] In addition, the functional units in the various embodiments of this specification may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0185] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this specification is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of this specification. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0186] Specific embodiments are used in this specification to illustrate the principles and implementation methods of this specification. The description of the above embodiments is only used to help understand the methods and core ideas of this specification. At the same time, for those skilled in the art, based on the ideas of this specification, there will be changes in the specific implementation methods and application scope. In summary, the contents of this specification should not be understood as limiting this specification.

Claims

1. A method for predicting oil and gas fracturing production based on geological and engineering data, characterized in that: include: Acquiring pumping data of a target well and well logging data of the formation where the target well is located; Performing two-dimensional conversion on the pumping data and the well logging data respectively to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix; Inputting the pumping two-dimensional feature matrix into the trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index; Inputting the well logging two-dimensional feature matrix into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators; Inputting the construction quality evaluation index, formation physical property index, and reservoir fluid index into a fully connected layer, which outputs the production prediction result of the target well through an activation function; The pumping data is subjected to a two-dimensional conversion to obtain a two-dimensional pumping feature matrix including: Dividing the pumping data according to working conditions to obtain pumping data under multiple stages; Perform two-dimensional transformation on the pumping data corresponding to each stage to obtain the two-dimensional pumping feature matrix corresponding to each stage; The first convolutional neural network model includes multiple input channels; The pumping two-dimensional feature matrix is input into the trained first convolutional neural network model for convolution and pooling calculations to obtain the construction quality evaluation index further including: Inputting the two-dimensional pumping feature matrix corresponding to each stage into the first convolutional neural network model through the corresponding input channel of the first convolutional neural network model, and performing convolution and pooling calculations to obtain the construction quality evaluation index; Performing a two-dimensional conversion on the well logging data to obtain a two-dimensional feature matrix of the well logging further includes: Analyzing the logging data to obtain index parameters of the formation, wherein the index parameters include permeability, brittleness index, Young's modulus and oil-water classification; Performing a two-dimensional conversion on the index parameters to obtain a two-dimensional well logging characteristic matrix corresponding to the index parameters; The second convolutional neural network model includes at least two input channels; Inputting the two-dimensional feature matrix of well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators further includes: The logging data and index parameters are respectively input into the second convolutional neural network model through the corresponding input channels of the second convolutional neural network model, and convolution and pooling calculations are performed to obtain the formation physical property index and reservoir fluid index.

2. The method according to claim 1, characterized in that The steps of training the first convolutional neural network model and the second convolutional neural network model include: Acquire pumping data of a produced well as a training pumping data set, and acquire well logging data of a formation where the produced well is located as a training well logging data set; Performing two-dimensional conversion on the data in the training pumping data set and the training well logging data set respectively to obtain a training pumping two-dimensional feature matrix and a training well logging two-dimensional feature matrix; Inputting the training pump injection two-dimensional feature matrix into the first convolutional neural network model to be trained to perform convolution and pooling calculations to obtain a predicted construction quality evaluation index; Inputting the training well logging two-dimensional feature matrix into the second convolutional neural network model to be trained to perform convolution and pooling calculations to obtain predicted formation physical property indicators and predicted reservoir fluid indicators; The predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well through the activation function. The first convolutional neural network model and the second convolutional neural network model are iteratively trained according to the production prediction result of the produced well and the actual production of the produced well until the difference between the production prediction result and the actual production of the produced well meets the predetermined requirements, thereby obtaining the trained first convolutional neural network model and the second convolutional network model.

3. The method according to claim 2, characterized in that Iteratively training the first convolutional neural network model and the second convolutional neural network model according to the production prediction result of the produced well and the actual production of the produced well further includes: During any iterative training process, the model parameters of either the first convolutional neural network model or the second convolutional neural network model are frozen, the model parameters of the unfrozen convolutional neural network model are adjusted, and then the training pump injection two-dimensional feature matrix and the training logging two-dimensional feature matrix are input into the corresponding convolutional neural network model for calculation to obtain the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index of this iteration, and then the predicted construction quality evaluation index, predicted formation physical property index and predicted reservoir fluid index obtained in this iteration are input into the fully connected layer, and the fully connected layer outputs the production prediction result of the produced well in this iteration through the activation function.

4. An oil and gas fracturing production prediction device based on geological and engineering data, characterized in that: include: A data acquisition unit, configured to acquire pumping data of a target well and well logging data of a formation where the target well is located; a data two-dimensional conversion unit, configured to perform two-dimensional conversion on the pumping data and the well logging data to obtain a two-dimensional pumping feature matrix and a two-dimensional well logging feature matrix; A first calculation unit is used to input the pumping two-dimensional feature matrix into a trained first convolutional neural network model to perform convolution and pooling calculations to obtain a construction quality evaluation index; A second computing unit inputs the two-dimensional feature matrix of well logging into a trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators; A fully connected layer calculation unit is used to input the construction quality evaluation index, formation physical property index and reservoir fluid index into the fully connected layer, and the fully connected layer outputs the production prediction result of the target well through an activation function; The pumping data is subjected to a two-dimensional conversion to obtain a two-dimensional pumping feature matrix including: Dividing the pumping data according to working conditions to obtain pumping data under multiple stages; Perform two-dimensional transformation on the pumping data corresponding to each stage to obtain the two-dimensional pumping feature matrix corresponding to each stage; The first convolutional neural network model includes multiple input channels; The pumping two-dimensional feature matrix is input into the trained first convolutional neural network model for convolution and pooling calculations to obtain the construction quality evaluation index further including: Inputting the two-dimensional pumping feature matrix corresponding to each stage into the first convolutional neural network model through the corresponding input channel of the first convolutional neural network model, and performing convolution and pooling calculations to obtain the construction quality evaluation index; Performing a two-dimensional conversion on the well logging data to obtain a two-dimensional feature matrix of the well logging further includes: Analyzing the logging data to obtain index parameters of the formation, wherein the index parameters include permeability, brittleness index, Young's modulus and oil-water classification; Performing a two-dimensional conversion on the index parameters to obtain a two-dimensional well logging characteristic matrix corresponding to the index parameters; The second convolutional neural network model includes at least two input channels; Inputting the two-dimensional feature matrix of well logging into the trained second convolutional neural network model to perform convolution and pooling calculations to obtain formation physical property indicators and reservoir fluid indicators further includes: The logging data and index parameters are respectively input into the second convolutional neural network model through the corresponding input channels of the second convolutional neural network model, and convolution and pooling calculations are performed to obtain the formation physical property index and reservoir fluid index.

5. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images