Bedding fracture prediction method, device, medium and product based on virtual well technology
Through virtual well technology and deep learning model, seismic body properties and well trajectory properties of shale reservoirs are extracted, and the cost and accuracy of stratigraphic joint recognition and prediction in the existing technology is solved, and more efficient stratigraphic workshops are achieved seam prediction accuracy and generalization ability.
Patent Information
- Application Number
- CN202411433028.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-15
- Publication Date
- 2025-01-03
- Estimated Expiration
- 2044-10-15
AI Technical Summary
The existing technology is costly and difficult to promote in large-scale applications when identifying and predicting stratification joints in shale reservoirs, and the accuracy is difficult to ensure when the actual number of wells in the well area is small. The model accuracy of traditional machine learning models is very accurate and extensive. Inadequate ability to transform.
Using a method based on virtual well technology, by obtaining seismic record data from the research work area, using SMI software to determine the time-depth relationship data, establish a velocity field, and deploying virtual wells under hierarchical interpretation constraints, extracting the trajectory attributes along the well, combining convolution with Neural networks and bidirectional long and short-term memory neural networks build layered joint prediction models to improve prediction accuracy.
The prediction accuracy of the layered joints is improved when the number of wells in the research area is small, which reduces costs, enhances the generalization ability and robustness of the model, and can more accurately identify and predict the distribution of the layered joints.
Smart Images
Figure CN118962802B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of geophysical exploration technology, and in particular to a bedding fracture prediction method, device, medium and product based on virtual well technology. Background Art
[0002] Cracks are generated under various geological structural effects, such as the deformation of folds, the influence of abnormal overpressure of fluids, fault activity, etc. They can serve as important channels for oil and gas migration and as storage spaces. Bedding fractures refer to natural fractures that split along the bedding of rocks during the sedimentation and diagenesis process. A large number of exploration practices and studies have shown that bedding fractures are commonly developed in clastic rocks, carbonate rocks, and shale reservoirs. In particular, the degree of research on the laws of bedding fractures and oil and gas migration and accumulation in unconventional sedimentary reservoirs restricts the exploration and discovery of oil and gas. Therefore, the identification and prediction of bedding fractures provides certain guidance for the next step of oil and gas exploration.
[0003] In 2023, the Journal of Paleogeography published Lai Jin's "Progress in Logging Geology in the Unconventional Oil and Gas Era", which pointed out that the development and distribution trend of the vertical upper bedding fractures in a single well can be identified by imaging logging and core observation, and then using the acoustic and resistivity logging of conventional logging sequences. In 2020, Petroleum & Natural Gas Geology published Lu Shuangquan's "Horizontal Bedding Fracture Rock Physics Model and Its Seismic Response Characteristics". This paper used rock physics modeling and seismic response analysis, combined with Voigt-Reuss-Hill (VRH) and equivalent self-consistent approximation (SCA) theory, introduced the Chapman multi-scale fracture rock physics model, and successfully developed a set of horizontal bedding fracture modeling methods. In 2022, China Offshore Oil and Gas published Wang Junchao's "Main Control Factors and Evaluation of Shale Fracture Development in the "Upper Sweet Spot" of the Lucaogou Formation in the Jimusar Sag". Based on multivariate linear regression, this paper uses a variety of fracture-sensitive seismic attributes and reservoir parameters as model evaluation parameters to construct a bedding fracture development evaluation model to predict the planar development of bedding fractures. In 2023, Earth Science published Zeng Lianbo's "Progress in the Study of Natural Fractures in Organic-Rich Shales", which pointed out that based on the development characteristics of bedding fractures, geology, geophysics and machine learning can be combined to develop methods for the classification, evaluation and prediction of bedding fractures. In 2018, Petroleum Geophysical Exploration published Liu Cai's "Characterization of Anisotropy of Shale Reservoirs Based on Rock Physics". This paper takes shale oil reservoir fractures as the research object, simulates seismic response, and proposes an anisotropic parameter AVO inversion method based on VTI medium reflection theory to predict the spatial distribution of horizontal bedding fractures.
[0004] In summary, the identification and prediction of bedding fractures play an important role in the shale oil and gas exploration industry. Although the development and distribution of bedding fractures can be identified through core observation and imaging logging, the cost is too high and not suitable for large-scale promotion; bedding fractures can be better identified through rock physics models combined with seismic response analysis, but small-scale fractures and microfractures are difficult to evaluate; the accuracy of the evaluation model can be quantitatively characterized by combining fracture-sensitive seismic attributes and reservoir parameters, but when the actual number of wells in the well area is small, the accuracy is difficult to guarantee; it is difficult to evaluate the effectiveness and accuracy of bedding fractures through related seismic attribute bodies such as coherence bodies and ant bodies; through traditional machine learning models or simple neural network models combined with bedding fracture development characteristics, although the density of vertical bedding fractures in a single well can be predicted to a certain extent, the accuracy and generalization ability of the model still need to be improved. Summary of the invention
[0005] The purpose of the present application is to provide a bedding fracture prediction method, device, medium and product based on virtual well technology, which improves the prediction accuracy of bedding fractures.
[0006] To achieve the above objectives, this application provides the following solutions.
[0007] A bedding fracture prediction method based on virtual well technology, comprising:
[0008] Acquire seismic record data of the research area; the seismic record data includes: post-stack three-dimensional seismic data and well location data of each real well, well trajectory data, layer interpretation data, fault polygon data and core statistical bedding fracture data;
[0009] Using SMI software, based on the seismic record data, determine the time-depth relationship data of each real well, perform velocity field modeling based on each time-depth relationship data, and obtain a seismic velocity field;
[0010] Based on the range coordinate difference of the research area, coordinate difference interval is adopted to evenly deploy multiple virtual wells in the research area; the number of the virtual wells is greater than the number of the real wells;
[0011] Extracting the seismic volume attributes of the study area, and extracting the well trajectory attributes of each virtual well according to the seismic volume attributes based on the horizon interpretation constraints and the seismic velocity field; the well trajectory attributes include: variance, root mean square amplitude, adjacent channel coherence, gradient modulus, similarity, maximum positive curvature and minimum negative curvature attributes;
[0012] The well trajectory attributes of each virtual well are respectively input into the bedding fracture prediction model to obtain corresponding bedding fracture prediction information; the bedding fracture prediction model is determined based on an initial network, and the initial network includes: a convolutional neural network and a bidirectional long short-term memory neural network.
[0013] Optionally, the process of determining the bedding fracture prediction model includes:
[0014] Extracting seismic volume attributes based on the seismic velocity field, and extracting along-well trajectory attributes of each real well according to the seismic volume attributes based on the constraints of the stratigraphic interpretation;
[0015] Constructing an initial network based on the convolutional neural network and the bidirectional long short-term memory neural network;
[0016] The initial network is trained using the well trajectory attributes of each of the real wells and the corresponding core statistical bedding fracture data to obtain the bedding fracture prediction model.
[0017] Optionally, the initial network is trained using the well trajectory attributes of each of the real wells and the corresponding core statistical bedding fracture data to obtain the bedding fracture prediction model, including:
[0018] Pearson correlation analysis is used to optimize the features of each well trajectory attribute to obtain the target attribute;
[0019] Normalize each target attribute to obtain a normalized target attribute;
[0020] The normalized target attributes of each real well are used as input, and the corresponding core statistical bedding fracture data are used as output, and the initial network is trained to obtain the bedding fracture prediction model.
[0021] Optionally, after inputting the well trajectory attributes of each virtual well into the bedding fracture prediction model to obtain the corresponding bedding fracture prediction information, the method further includes:
[0022] The bedding fracture prediction information of each virtual well is plotted using a double-fox plane map to obtain a virtual well fracture plane map.
[0023] Optionally, after the bedding fracture prediction information of each virtual well is plotted in a double-fox plane map to obtain the virtual well fracture plane map, the method further includes:
[0024] The core statistical bedding fracture data of each real well is projected onto the virtual well fracture plane map to determine whether the bedding fracture prediction information of each virtual well is consistent with the bedding fracture development of each real well.
[0025] Optionally, the post-stack three-dimensional seismic data is information about underground rock and strata determined based on echoes; the echoes are waves that propagate back to the surface by exciting vibrations on the ground during seismic exploration;
[0026] The well location data are coordinates used to describe the geographical location of the well in oil and gas exploration;
[0027] The well trajectory data is the drilling path or trajectory of the well underground;
[0028] The horizon interpretation data is the interpretation information of the underground stratum structure realized by extending the geological information revealed by the logging and drilling data to the through-well seismic profile;
[0029] The fault polygon data is a graph displayed by projecting the upper and lower plates of the fault in space onto a plane;
[0030] The core statistical bedding fracture data is the bedding fracture development information recorded by actual core observation.
[0031] A computer device comprises: 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 any of the above-mentioned bedding fracture prediction methods based on virtual well technology.
[0032] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements any of the above-mentioned bedding fracture prediction methods based on virtual well technology.
[0033] A computer program product comprises a computer program, which, when executed by a processor, implements any of the above-mentioned bedding fracture prediction methods based on virtual well technology.
[0034] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0035] The present application discloses a bedding fracture prediction method, device, medium and product based on virtual well technology, which evenly deploys virtual wells in the research area and extracts the attributes of real wells and virtual wells along the well trajectory, and makes supervised learning model sample training data based on the bedding fracture data statistically observed by real well core observation and the extracted attributes along the well trajectory, and combines convolutional neural network and bidirectional long short-term memory neural network to establish a bedding fracture prediction model to carry out intelligent plane prediction of bedding fracture virtual wells. It can effectively solve the problems of small number of wells in the research area that are difficult to quantitatively characterize, too high cost, general prediction accuracy, generalization ability and model robustness. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] 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.
[0037] Figure 1 A schematic flow chart of a bedding fracture prediction method based on virtual well technology provided in Example 1 of the present application.
[0038] Figure 2 Schematic diagram of the intelligent prediction process of bedding fractures using virtual well technology.
[0039] Figure 3 Schematic diagram of virtual well deployment.
[0040] Figure 4 This is a correlation heat map.
[0041] Figure 5 Schematic diagram of the CNN-BiLSTM principle.
[0042] Figure 6 Schematic diagram for comparing the loss functions of the model test set.
[0043] Figure 7 This is a schematic diagram of the predicted plane development of bedding fractures in the second sub-segment.
[0044] Figure 8 This is a schematic diagram of the predicted plane development of bedding fractures in the third sub-section.
[0045] Fig. 9 This is a schematic diagram of the predicted plane development of bedding fractures in the fourth sub-section.
[0046] Fig.10 This is a schematic diagram of the predicted plane development of bedding fractures in the fifth sub-section. DETAILED DESCRIPTION
[0047] 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.
[0048] The purpose of this application is to provide a bedding fracture prediction method, device, medium and product based on virtual well technology, aiming to improve the prediction accuracy of bedding fractures.
[0049] 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.
[0050] Example 1 provides a bedding fracture prediction method based on virtual well technology.
[0051] like Figure 1 and Figure 2 As shown, the bedding fracture prediction method based on virtual well technology in this embodiment includes the following steps.
[0052] Step 1: Obtain earthquake record data in the study area.
[0053] Among them, the seismic record data include: post-stack three-dimensional seismic data and the well location data of each real well, well trajectory data, layer interpretation data, fault polygon data and core statistical bedding fracture data.
[0054] As an optional implementation, the post-stack 3D seismic data is information about underground rock and strata determined based on echoes; echoes are waves that propagate back to the surface by exciting vibrations on the ground during seismic exploration.
[0055] Well location data are coordinates used in oil and gas exploration to describe the geographic location of a well.
[0056] Well trajectory data is the drilling path or trajectory of a well in the ground.
[0057] The stratigraphic interpretation data is the information on the underground stratigraphic structure that is obtained by extending the geological information revealed by logging and drilling data to the seismic profile through the well.
[0058] Fault polygon data is the graphics displayed by projecting the upper and lower plates of the fault in space onto a plane.
[0059] The core statistical bedding fracture data is the bedding fracture development information recorded by actual core observation.
[0060] Step 2: Using SMI software, based on seismic record data, determine the time-depth relationship data of each real well, perform velocity field modeling based on each time-depth relationship data, and obtain the seismic velocity field.
[0061] Step 3: Based on the range coordinate difference of the study area, coordinate difference interval is used to evenly deploy multiple virtual wells in the study area; the number of virtual wells is greater than the number of real wells.
[0062] Step 4: Extract the seismic volume attributes of the study area, and extract the along-well trajectory attributes of each virtual well according to the seismic volume attributes based on the horizon interpretation constraints and seismic velocity field.
[0063] Among them, the attributes along the well trajectory include: variance, root mean square amplitude, adjacent channel coherence, gradient modulus, similarity, maximum positive curvature and minimum negative curvature attributes.
[0064] Step 5: Input the well trajectory attributes of each virtual well into the bedding fracture prediction model to obtain the corresponding bedding fracture prediction information.
[0065] Among them, the bedding fracture prediction model is determined based on the initial network, which includes: convolutional neural network and bidirectional long short-term memory neural network.
[0066] As an optional implementation, the process of determining the bedding fracture prediction model includes steps 51 to 53.
[0067] Step 51: extracting seismic volume attributes based on the seismic velocity field, and extracting along-well trajectory attributes of each real well based on the seismic volume attributes and on the basis of the stratigraphic interpretation constraints.
[0068] Step 52: Construct an initial network based on the convolutional neural network and the bidirectional long short-term memory neural network.
[0069] Step 53: The initial network is trained using the well trajectory attributes of each real well and the corresponding core statistical bedding fracture data to obtain a bedding fracture prediction model.
[0070] As an optional implementation, step 53 includes step 531 and step 532.
[0071] Step 53: Pearson correlation analysis is used to perform feature optimization on the attributes along the well trajectory to obtain target attributes.
[0072] Step 531: normalize each target attribute to obtain a normalized target attribute.
[0073] Step 532: Taking the normalized target attributes of each real well as input and the corresponding core statistical bedding fracture data as output, the initial network is trained to obtain a bedding fracture prediction model.
[0074] As an optional implementation manner, after step 5, the method further includes:
[0075] Step 6: Draw a double-fox plane map of the bedding fracture prediction information of each virtual well to obtain a virtual well fracture plane map.
[0076] As an optional implementation manner, after step 6, the method further includes:
[0077] Step 7: Project the core statistical bedding fracture data of each real well onto the virtual well fracture plane map to determine whether the bedding fracture prediction information of each virtual well is consistent with the bedding fracture development of each real well.
[0078] In a specific embodiment, the method in Example 1 is explored by taking the Gaoyou Depression study area as an example.
[0079] Step 1: Use "SMI" geological software to establish the research area, read the post-stack 3D seismic data of Gaoyou Sag, the well location data of 4 real wells (represented by HY1, HY3, H2C, and HX28 respectively), well trajectory data, stratigraphic interpretation data of the four sub-segments of the Fu 2nd Member, fault polygon data, and core statistical bedding fracture data.
[0080] Among them, the "SMI" software, also known as seismic waveform indication inversion software, is a high-precision seismic phase-controlled reservoir prediction software independently developed by Beijing Zhongheng Lihua Petroleum Technology Research Institute. The software uses a Chinese interface and has a three-dimensional visualization function. It is suitable for high-precision reservoir prediction of thin interlayers, tight reservoirs and unconventional reservoirs.
[0081] Step 2: First, based on step 1, use the built-in module of synthetic records in the SMI software to make synthetic seismic records of the well, and you can get the time-depth relationship of the real well. The time-depth relationship is the result obtained after making synthetic seismic records in the SMI software. Because most seismic data are depths in the time domain, and the depth on the well logging is the depth in the depth domain. Therefore, the two do not match and need to be converted. After saving the time-depth relationship, the seismic data near the well can be converted from the time domain to the depth domain. The time-depth relationship is a series of data pairs obtained after the synthetic seismic record is made, and one time value corresponds to one depth value. Secondly, using the real well with an established time-depth relationship, the interpolation principle is used inside the SMI software to calculate the layer velocity of each layer of the real well and obtain the seismic velocity field. Once again, read the coordinates of the three points in the study area (range coordinates), take the coordinates of any two points as the basis, use the control variable method to control the same value of the coordinates of the two points, select two coordinate points in the same direction, calculate the distance between the two points, and distribute the calculated distance between the two points at equal intervals, that is, divide this distance into several equal parts, so as to expand the original two-point coordinates in the same direction into several coordinate points, and use these expanded coordinate points as the coordinates of the virtual well. In the "DF-GVision" software, import the coordinates of the study area and the coordinates of the equally spaced coordinates mentioned above to obtain the coordinates of the work area range and the virtual well. In this software, the virtual well coordinates are continuously copied and typeset to obtain the virtual well coordinates evenly distributed in the study area, and are evenly spaced to obtain a large amount of virtual well coordinate data. A total of 876 virtual wells were deployed, and the obtained virtual well location plane distribution is shown as follows: Figure 3 shown.
[0082] Step 3: Based on step 1 and step 2, use SMI software to extract seismic body attributes, add velocity field-based time-depth conversion and interpretation of horizon constraints on the basis of seismic body attributes, and extract the seismic attributes along the well trajectory of real wells and virtual wells in the four sub-segments of the Fu 2nd segment in the study area. The seismic attributes along the well trajectory include: variance, root mean square amplitude, adjacent channel coherence, gradient modulus, similarity, maximum positive curvature and minimum negative curvature attributes.
[0083] Step 4: Based on step 3, the seismic attribute data along the well trajectory of the two real wells HY1 and HY3 extracted are used as the feature data of the supervised learning model, and the bedding fracture information in the four sub-segments of the Fu 2nd Section obtained from the core observation of these two wells is used as the label data of the supervised learning model. The feature data is combined with the label data to make training sample data. The attribute heat map is drawn using Pearson correlation analysis so that the feature data can be optimized and the feature variables with low correlation with the target variable can be removed to achieve the purpose of dimensionality reduction and redundancy reduction. Figure 4 The following is a heat map showing the correlation between seismic attributes and bedding fractures. Figure 4 It can be seen that the variance in the data is most correlated with the core statistical bedding fracture data, and the correlation coefficient can reach 0.41. The maximum positive curvature and gradient modulus are correlated with the core statistical bedding fracture data by 0.31 and 0.24, respectively. The attribute with the worst correlation is the minimum negative curvature curvature attribute, which is only 0.0055, which is at the end of the numerical value of the correlation between the seismic attributes along the well trajectory and the core statistical bedding fracture data. The feature selection process should select the feature attributes with high correlation with the label data, rather than the attributes with high feature autocorrelation. By selecting the top 80% of the attributes represented by the correlation value with the label data, that is, selecting the attributes with a correlation with the bedding fracture data greater than or equal to 0.1. The minimum negative curvature is finally discarded among the above attributes, thus completing the feature data selection process. Based on the training sample data after feature optimization, data normalization operation is performed to constrain the scale difference between the data. And the virtual well is used to use data mining technology to retain the virtual well trajectory attribute file after the attribute column is removed, which is used as the data to be predicted.
[0084] Specifically, the principle of Pearson correlation is as follows:
[0085] .
[0086] in, Represents seismic attributes along the well trajectory , The linear correlation coefficient between them; m represents the number of samples of seismic attributes along the well trajectory; Represents seismic attributes along the well trajectory The value of the i-th sample; Represents seismic attributes along the well trajectory The average value of the i-th sample; Represents seismic attributes along the well trajectory The value of the i-th sample; Represents seismic attributes along the well trajectory The mean of the ith sample.
[0087] Among them, the normalization principle is as follows:
[0088] .
[0089] in, represents the normalized data; represents the data to be normalized; Represents the minimum value of the data to be normalized; Indicates the maximum value of the data to be normalized.
[0090] Step 5: Based on step 4, the deep learning library Keras is used as the background to use the architecture model that combines the Convolutional Neural Network (CNN) and the Bi-directional Long Short-Term Memory (BiLSTM) network. The convolutional neural network is used to extract features from the training samples. After the process of convolution, pooling, and nonlinear transformation (activation function), a processed data matrix is obtained, and this data matrix is sent to the BiLSTM network to continue the nonlinear transformation of the data, and then output. CNN is mainly used to extract features from the data, and BiLSTM is mainly used to selectively forget excessive data. The model principle is as follows: Figure 5 The model is trained based on the seismic attributes along the well trajectory and the core statistical bedding fracture data after the feature optimization of the real well in step 4. After the model is adjusted, the specific parameters are shown in Table 1.
[0091] Table 1 Model parameters
[0092]
[0093] Table 2 Comparison of model test set prediction values and evaluation indicators
[0094]
[0095] Based on the parameters in Table 1, the model was trained and compared with the prediction accuracy of the conventional neural network model, as shown in Table 2. It can be seen that the evaluation index of the MLP model is still far behind other models, and there is also a large deviation between the predicted value and the true value; the CNN-BiLSTM model has the highest accuracy among them, and the model's determination coefficient is 0.910, which is about 3 percentage points higher than the CNN-LSTM model. In order to clearly observe the trend of the model loss function, the loss diagram of the first 50 rounds is shown, as shown in Figure 6 The figure shows the loss function diagram of each model on the test set. It can be seen that the CNN-BiLSTM model has the lowest test set loss, indicating that the error between the predicted value and the true value is the smallest. In Table 2, Rmse is the root mean square error, Mae is the mean absolute error, and R2 is the determination coefficient.
[0096] Step 6: According to the CNN-BiLSTM model optimized in step 5, predict the bedding fracture development value of each virtual well in each sub-segment. Import the bedding fracture development value and coordinates at the virtual well point into the "DF-GVision" software to draw the bedding fracture development value plane contour map within the study area, and project the four core observation wells HY1, HY3, H2C, and HX28 in the study area onto the virtual well fracture prediction plane map for observation. Among them, H2C and HX28 are not involved in the modeling and are verification wells. The bedding fracture plane development map predicted by the virtual well is shown in the figure below: Figure 7-10 The actual well core observation record bedding fracture development results are shown in Table 3. By sorting the development degree, the sub-section location of each well can be known. By comparing the two, it can be seen that the real statistical results on the well are basically consistent with the results of the virtual well prediction plane map, which shows the excellent generalization ability and prediction accuracy of the CNN-BiLSTM model, and provides theoretical and practical guidance for the next step of oil and gas exploration.
[0097] Table 3 Bedding fracture development in actual well longitudinal core observation
[0098]
[0099] Among them, the levels of bedding fracture development in a single well are poorly developed, moderately developed, well developed, and best developed, and the corresponding numbers are 1, 2, 3, and 4.
[0100] Embodiment 2 provides a computer device, including: 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 bedding fracture prediction method based on virtual well technology in Embodiment 1.
[0101] Embodiment 3 provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the bedding fracture prediction method based on the virtual well technology in Embodiment 1 is implemented.
[0102] Embodiment 4 provides a computer program product, including a computer program, which, when executed by a processor, implements the bedding fracture prediction method based on virtual well technology in Embodiment 1.
[0103] Embodiment 5 provides a computer device, which may be a database. The computer device includes a processor, a memory, an input / output interface (I / O for short) 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 database of the computer device is used to store pending transactions. 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 bedding fracture prediction method based on virtual well technology in Embodiment 1 is implemented.
[0104] It should be noted that the object information (including but not limited to object device information, object 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 object or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0105] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, 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 processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this 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. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0106] 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.
[0107] 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 bedding fracture prediction method based on virtual well technology, characterized in that: The method comprises: Acquire seismic record data of the research area; the seismic record data includes: post-stack three-dimensional seismic data and well location data of each real well, well trajectory data, layer interpretation data, fault polygon data and core statistical bedding fracture data; Using SMI software, based on the seismic record data, determine the time-depth relationship data of each real well, perform velocity field modeling based on each time-depth relationship data, and obtain a seismic velocity field; Based on the range coordinate difference of the research area, coordinate difference interval is adopted to evenly deploy multiple virtual wells in the research area; the number of the virtual wells is greater than the number of the real wells; Extracting the seismic volume attributes of the study area, and extracting the well trajectory attributes of each virtual well according to the seismic volume attributes based on the horizon interpretation constraints and the seismic velocity field; the well trajectory attributes include: variance, root mean square amplitude, adjacent channel coherence, gradient modulus, similarity, maximum positive curvature and minimum negative curvature attributes; Inputting the well trajectory attributes of each virtual well into the bedding fracture prediction model to obtain corresponding bedding fracture prediction information; the bedding fracture prediction model is determined based on an initial network, and the initial network includes: a convolutional neural network and a bidirectional long short-term memory neural network; The process of determining the bedding fracture prediction model includes: Extracting seismic volume attributes based on the seismic velocity field, and extracting along-well trajectory attributes of each real well according to the seismic volume attributes based on the constraints of the stratigraphic interpretation; Constructing an initial network based on the convolutional neural network and the bidirectional long short-term memory neural network; The initial network is trained using the well trajectory attributes of each of the real wells and the corresponding core statistical bedding fracture data to obtain the bedding fracture prediction model.
2. The bedding fracture prediction method based on virtual well technology according to claim 1 is characterized in that: The initial network is trained using the well trajectory attributes of each of the real wells and the corresponding core statistical bedding fracture data to obtain the bedding fracture prediction model, including: Pearson correlation analysis is used to optimize the features of each well trajectory attribute to obtain the target attribute; Normalize each target attribute to obtain a normalized target attribute; The normalized target attributes of each real well are used as input, and the corresponding core statistical bedding fracture data are used as output, and the initial network is trained to obtain the bedding fracture prediction model.
3. The bedding fracture prediction method based on virtual well technology according to claim 1 is characterized in that: After inputting the well trajectory attributes of each virtual well into the bedding fracture prediction model to obtain the corresponding bedding fracture prediction information, the method further includes: The bedding fracture prediction information of each virtual well is plotted using a double-fox plane map to obtain a virtual well fracture plane map.
4. The bedding fracture prediction method based on virtual well technology according to claim 3 is characterized in that: After the bedding fracture prediction information of each virtual well is drawn into a double-fox plane map to obtain the virtual well fracture plane map, the following steps are also included: The core statistical bedding fracture data of each real well is projected onto the virtual well fracture plane map to determine whether the bedding fracture prediction information of each virtual well is consistent with the bedding fracture development of each real well.
5. The bedding fracture prediction method based on virtual well technology according to claim 1 is characterized in that: The post-stack 3D seismic data is information about underground rock and strata determined based on echoes; the echoes are waves that propagate back to the surface by exciting vibrations on the ground during seismic exploration; The well location data are coordinates used to describe the geographical location of the well in oil and gas exploration; The well trajectory data is the drilling path or trajectory of the well underground; The horizon interpretation data is the interpretation information of the underground stratum structure realized by extending the geological information revealed by the logging and drilling data to the through-well seismic profile; The fault polygon data is a graph displayed by projecting the upper and lower plates of the fault in space onto a plane; The core statistical bedding fracture data is the bedding fracture development information recorded by actual core observation.
6. 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 bedding fracture prediction method based on virtual well technology as described in any one of claims 1 to 5.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the bedding fracture prediction method based on virtual well technology described in any one of claims 1 to 5 is implemented.
8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the bedding fracture prediction method based on virtual well technology described in any one of claims 1 to 5 is implemented.