A collaborative information feedback prediction method for crack identification

Abstract

The present invention relates to the field of seismic data interpretation technology, and discloses a collaborative information feedback prediction method for crack identification. The method comprises: preprocessing raw seismic data to obtain a two-dimensional seismic profile; extracting features from the two-dimensional seismic profile to obtain crack-sensitive seismic attributes; fusing the crack-sensitive seismic attributes to obtain a preliminary crack identification result; correcting the preliminary crack identification result to obtain a crack training sample; and inputting the crack training sample into a preset network for training to obtain a crack identification model. The present invention can accurately predict the actual crack distribution in seismic data profiles and planes. Compared with traditional crack identification methods, the present method improves crack identification efficiency by approximately 5-10 times and crack identification accuracy by more than 30%, showing good practicality and effectiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of seismic data interpretation, and in particular to a collaborative information mutual feedback prediction method for crack identification. Background Art

[0002] Fractures are important reservoir spaces and migration pathways for oil and gas, significantly impacting reservoir oil and gas productivity. Accurately understanding the spatial distribution of underground fractures is crucial for studying fractured oil and gas reservoirs, improving exploration success rates, and enhancing development efficiency. Therefore, identifying and predicting the development and distribution of underground fractures has long been a research priority in the oil and gas field, both domestically and internationally.

[0003] Currently, fracture identification methods primarily include geological methods, well logging methods, and seismic methods. Geological methods are the most direct and simplest, enabling quantitative predictions. However, they cannot meet the high-precision requirements of oil and gas field development and are not widely used in reservoir prediction. Well logging methods exhibit anomalous responses to fractures and can qualitatively identify them, but are severely affected by environmental factors, frequently suffer from multiple solutions, and are difficult to mass-produce. Seismic methods offer continuous observations in both vertical and horizontal directions and are relatively low-cost, making them highly effective in predicting reservoir fractures. Seismic methods are the primary method used in reservoir prediction. Seismic methods primarily utilize low-cost 3D post-stack seismic data, using either manual seismic interpretation of profiles or seismic attributes (such as coherence, curvature, variance, and ant volumes) to predict fracture development areas. Manual seismic interpretation is highly subjective, time-consuming, and, due to the limited resolution of the seismic data, struggles to accurately identify small and medium-sized fractures. The seismic attribute prediction method is relatively time-saving, but the type of seismic attributes and the related parameter settings of each attribute have a significant impact on the fracture prediction results. Overall, it has problems such as low accuracy, high multi-solution, and limited prediction reliability. Summary of the Invention

[0004] The purpose of the present invention is to provide a collaborative information mutual feedback prediction method for crack identification, which solves the problems of low accuracy, high multi-solution, and limited prediction reliability of existing crack prediction methods.

[0005] To achieve the above-mentioned object, the technical solution adopted by the present invention is: a collaborative information feedback prediction method for crack identification, the method comprising:

[0006] Preprocess the original seismic data to obtain a two-dimensional seismic profile;

[0007] Extract features from 2D seismic profiles to obtain fracture-sensitive seismic attributes;

[0008] Attribute fusion is performed on fracture-sensitive seismic attributes to obtain preliminary fracture identification results;

[0009] Correct the preliminary crack identification results to obtain crack training samples;

[0010] The crack training samples are input into the preset network for training to obtain the crack recognition model.

[0011] Preferably, the method further comprises: verifying and optimizing the crack identification model based on an information feedback scoring mechanism.

[0012] Preferably, the original seismic data is three-dimensional seismic data.

[0013] Preferably, the preprocessing includes normalization processing and standardization processing.

[0014] Preferably, the preset network is a Res-Unet network.

[0015] Preferably, the fracture-sensitive seismic attributes include: coherence, variance, curvature and ant volume.

[0016] Preferably, the preliminary crack identification results are corrected to obtain crack training samples, including:

[0017] Process the acquired geological and well logging data to obtain the fracture development pattern of the work area;

[0018] Based on the crack development pattern of the work area, the value range of the crack parameters is obtained;

[0019] The preliminary crack identification results are constrained according to the value range of the crack parameters to obtain constrained cracks;

[0020] A fracture label library is constructed using the 2D seismic profile as input and the constrained fractures corresponding to the 2D seismic profile as labels;

[0021] Determine the crack training samples based on the crack label library.

[0022] Preferably, it also includes:

[0023] Obtain earthquake sample data to be predicted;

[0024] The earthquake sample data to be predicted is identified based on the fracture identification model to obtain the fracture identification result.

[0025] The present invention further provides a collaborative information mutual feedback prediction device for crack identification, which is used to implement the collaborative information mutual feedback prediction method for crack identification. The device includes:

[0026] An acquisition module is used to pre-process the original seismic data to obtain a two-dimensional seismic profile;

[0027] Extraction module, used to extract features from two-dimensional seismic profiles and obtain fracture-sensitive seismic attributes;

[0028] The fusion module is used to fuse the fracture-sensitive seismic attributes to obtain the preliminary fracture identification results;

[0029] A correction module is used to correct the initial crack identification results to obtain crack training samples;

[0030] The training module is used to input crack training samples into the preset network for training to obtain a crack recognition model.

[0031] The present invention also provides an electronic 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-mentioned collaborative information mutual feedback prediction method for crack identification when executing the computer program.

[0032] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned collaborative information mutual feedback prediction method for crack identification.

[0033] The beneficial effects of the present invention are concentrated in:

[0034] The present invention can accurately predict the actual fracture distribution of seismic data profiles and planes. Compared with traditional fracture identification methods, the method of the present invention improves the fracture identification efficiency by about 5-10 times and the fracture identification accuracy by more than 30%, showing good practicality and effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the collaborative information mutual feedback prediction method for crack identification of the present invention;

[0036] Figure 2 is a schematic diagram of training labels of the present invention;

[0037] Figure 3 is the distribution of the seismic data profile predicted by the present invention;

[0038] Figure 4 The distribution of cracks in the seismic profile of the research area of ​​the present invention;

[0039] Figure 5 This is a schematic diagram of the information feedback scoring mechanism used by experts in the present invention;

[0040] Figure 6 It is a block diagram of the collaborative information mutual feedback prediction device for crack identification of the present invention. DETAILED DESCRIPTION

[0041] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] like Figure 1 As shown, this embodiment provides a collaborative information mutual feedback prediction method for crack identification, the method comprising:

[0043] Step S101: pre-processing the original seismic data to obtain a two-dimensional seismic profile;

[0044] In this embodiment, the original seismic data comes from the three-dimensional seismic data of the study area. First, it is necessary to select seismic sections with relatively good data quality from the original seismic data, that is, select seismic data with a signal-to-noise ratio greater than 4 as sections with good seismic data quality, and then preprocess the sections, which includes normalization and standardization.

[0045] Specifically, the seismic profiles are standardized and cropped into 512*512 pixels. The cropped profiles are then normalized using the following formula:

[0046]

[0047] Where x' is the two-dimensional seismic profile, x max is the maximum value of the original seismic data, x min is the average value of the original seismic data, and x is the amplitude value of the data point.

[0048] After the above preprocessing, a sufficient number of two-dimensional seismic profiles can be obtained. In this study, a total of 500 512*512 seismic profiles were produced for subsequent artificial intelligence network training.

[0049] Step S102: extracting features from the two-dimensional seismic profile to obtain fracture-sensitive seismic attributes. In this embodiment, the fracture-sensitive seismic attributes include coherence, variance, curvature, and ant volume.

[0050] Step S103: performing attribute fusion on fracture-sensitive seismic attributes to obtain preliminary fracture identification results;

[0051] In this embodiment, the three attributes (coherence, curvature, ant volume, etc.) of each two-dimensional seismic profile are fused and reconstructed by the RGB attribute fusion method to obtain the preliminary fracture identification results of each two-dimensional seismic profile.

[0052] Step S104: correcting the preliminary crack identification results to obtain crack training samples;

[0053] In this example, the initial fracture identification results based on attribute fusion are inaccurate in some details due to the limitations of the algorithm and the seismic data itself, and require further correction and optimization based on the characteristics of the work area. Specifically, the initial fracture identification results are corrected to obtain fracture training samples, including:

[0054] Step a01: Extract information from the acquired geological and well logging data. In this example, geological evolutionary background data indicates that the study area was influenced by multiple periods of tectonic movement, including the Caledonian, Yanshanian, and Himalayan periods. The fractures in the study area are primarily composed of four types: Caledonian EW-trending, Yanshanian NE-trending, and Himalayan NNE and NS-trending fractures. Combined with well logging data, the occurrence characteristics of each type of fracture are further determined, and the fracture development scale is summarized. The fracture development stage, occurrence characteristics, and corresponding scale together constitute the fracture development pattern in the study area.

[0055] Step a02: Based on the fracture development pattern in the work area obtained in the previous step, combined with logging and actual drilling data, the value ranges of various fracture parameters are obtained. In this embodiment, the fracture parameters include: fracture dip, azimuth, length, and aspect ratio. In other words, the value ranges of parameters such as fracture dip, azimuth, length, and aspect ratio need to be determined. Specifically, the fracture dip is between 20 degrees and 70 degrees, the azimuth is between 25 degrees and 270 degrees, the length varies from 5 meters to 100 meters, and the aspect ratio is between 3:1 and 10:1.

[0056] Step a03: constraining the preliminary crack identification results according to the value range of the crack parameters to obtain constrained cracks;

[0057] After constraining the preliminary crack identification results according to the value range of crack parameters, cracks that exceed the parameter range can be eliminated, further optimizing the identification effect of cracks of various scales.

[0058] Step a04: constructing a fracture label library using the two-dimensional seismic profile as input and the constrained fractures corresponding to the two-dimensional seismic profile as labels;

[0059] In this embodiment, each 2D seismic profile is used as input, and the corresponding constrained fracture of each 2D seismic profile after processing in the above steps is used as a label to form a pair of training labels, such as Figure 2 As shown, 500 sets of training labels can be obtained at this time, and the 500 sets of training labels are used as a big data crack label library that conforms to the crack characteristics of the work area.

[0060] Step a05: determining crack training samples based on the crack label library;

[0061] In this embodiment, the above 500 sets of training labels are allocated as training set and validation set in a ratio of 4:1, wherein 400 sets of training labels are used as crack training samples, and the remaining 100 sets of training labels are used for prediction result verification.

[0062] Step S105: inputting the crack training samples into a preset network for training to obtain a crack recognition model.

[0063] In this embodiment, the preset network is a Res-Unet network. The Res-Unet network adds a Res residual module to the traditional Unet intelligent network used for crack identification tasks. Its core idea is to introduce residual edges, directly connecting the input to the output edge. When the learning effect of the newly added training network layer is very poor, the weight parameters of the training layer are set to 0, so that the input of the network layer is equal to the output, becoming an identity mapping, to prevent the occurrence of gradient vanishing, gradient explosion, network degradation, etc. during training. The mathematical expression of the residual learning unit is as follows:

[0064] y1=h(x l )+F(x l ,W l )#(2)

[0065] x l+1 =f(y l )#(3)

[0066] Among them, x l and x l+1 Represent the input and output of the Lth residual learning unit, F is the residual function, representing the learned residual, W l is the training weight of the Lth layer, h(x l )=x l Represents the identity mapping, and f represents the ReLU activation function. Through recursion, the feature expression of any deep unit L can be obtained:

[0067]

[0068] From formula (4), we can see that the characteristic x of the unit L of any number of layers is L The feature x of the shallow unit l can be used l The sum of the residual function F indicates that there is a residual feature between any deep unit and the shallow unit. For any deep unit in the network, its feature is the sum of all the residual function outputs before, while the feature of any deep unit in the network without residual units is composed of the product of a series of vectors. In terms of computational complexity, summation is much smaller than product, and the residual structure significantly reduces the computational complexity. From the perspective of backpropagation, assuming that the loss function is represented by ε, for x l Taking partial derivatives we get:

[0069]

[0070]

[0071] Formula (5) divides the gradient into and passed through the weight layer Ensure that the signal can be directly transmitted back to the shallow x l , ensuring that the gradient vanishing problem does not occur, further suppressing the decrease of the gradient. On the other hand, the addition is highly stable, thus preventing the gradient explosion. In summary, the residual module (Res) not only effectively prevents training problems such as gradient vanishing and gradient explosion, but also significantly reduces the amount of computation and saves training time.

[0072] As a further optimization of this embodiment, after the Res-Unet network training is completed, it is necessary to perform expert information feedback on the obtained crack identification model for verification and optimization. The experts use the information feedback scoring mechanism as follows: Figure 6 As shown, the training results are further screened and optimized. The scoring mechanism divides the recognition effect of a single profile into three levels: accurate, relatively accurate, and inaccurate. The crack recognition accuracy of more than 90% is accurate, and the result does not need to be modified. The accuracy of 80%-90% is relatively accurate, and it is necessary to return to the intelligent network for further optimization until the accuracy is increased to 90% and reaches the accuracy level. The accuracy rate is lower than 80% for inaccurate, and the original training labels need to be reprocessed and re-entered into the intelligent network for training. If the accuracy rate of the re-training is still lower than 80%, it is regarded as a problem sample and discarded directly. It can be seen from actual operation that the optimization mechanism of mutual feedback scoring can effectively improve the quality of crack recognition results, and ultimately obtain a crack recognition model with an identification accuracy rate of up to 94.5%. As a further optimization of this embodiment, the method also includes:

[0073] Step b01: Obtain earthquake sample data to be predicted;

[0074] Step b02: Identify the earthquake sample data to be predicted based on the fracture identification model to obtain a fracture identification result.

[0075] The present invention can accurately predict the actual fracture distribution of seismic data profiles and planes. Compared with traditional fracture identification methods, the method of the present invention improves the fracture identification efficiency by about 5-10 times and the fracture identification accuracy by more than 30%, showing good practicality and effectiveness.

[0076] like Figure 3-4 As shown, Figure 3 and Figure 4The actual fracture distribution of actual seismic data sections and planes is shown. Compared with traditional fracture identification methods, the proposed method improves fracture identification efficiency by approximately 6 times, and the fracture identification accuracy rate increases from approximately 60% to approximately 94.5%, an increase of approximately 35%, strongly demonstrating the practicality and effectiveness of the proposed method.

[0077] like Figure 6 As shown, the present invention also provides a collaborative information mutual feedback prediction device for crack identification, which is used to implement the collaborative information mutual feedback prediction method for crack identification. The device includes:

[0078] An acquisition module is used to pre-process the original seismic data to obtain a two-dimensional seismic profile;

[0079] Extraction module, used to extract features from two-dimensional seismic profiles and obtain fracture-sensitive seismic attributes;

[0080] The fusion module is used to fuse the fracture-sensitive seismic attributes to obtain the preliminary fracture identification results;

[0081] A correction module is used to correct the initial crack identification results to obtain crack training samples;

[0082] The training module is used to input crack training samples into the preset network for training to obtain a crack recognition model.

[0083] The present invention also provides an electronic 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-mentioned collaborative information mutual feedback prediction method for crack identification when executing the computer program.

[0084] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned collaborative information mutual feedback prediction method for crack identification.

[0085] The present invention can accurately predict the actual fracture distribution of seismic data profiles and planes. Compared with traditional fracture identification methods, the method of the present invention improves the fracture identification efficiency by about 5-10 times and the fracture identification accuracy by more than 30%, showing good practicality and effectiveness.

[0086] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0087] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0088] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0090] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0091] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0092] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0093] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0094] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A collaborative information feedback prediction method for crack identification, characterized by: The method comprises: Preprocess the original seismic data to obtain a two-dimensional seismic profile; Extract features from 2D seismic profiles to obtain fracture-sensitive seismic attributes; Attribute fusion is performed on fracture-sensitive seismic attributes to obtain preliminary fracture identification results; Process the acquired geological and logging data to obtain the fracture development pattern of the work area; Based on the crack development pattern of the work area, the value range of the crack parameters is obtained; The preliminary crack identification results are constrained according to the value range of the crack parameters to obtain constrained cracks; A fracture label library is constructed using the 2D seismic profile as input and the constrained fractures corresponding to the 2D seismic profile as labels; Determine crack training samples based on the crack label library; The crack training samples are input into the preset network for training to obtain the crack recognition model.

2. The collaborative information mutual feedback prediction method for crack identification according to claim 1, characterized in that: Also includes: The crack identification model is verified and optimized based on the information feedback scoring mechanism.

3. The collaborative information mutual feedback prediction method for crack identification according to claim 1, characterized in that: The original seismic data is three-dimensional seismic data.

4. The collaborative information mutual feedback prediction method for crack identification according to claim 1, characterized in that: The preprocessing includes normalization processing and standardization processing.

5. The collaborative information mutual feedback prediction method for crack identification according to claim 1, characterized in that: The preset network is the Res-Unet network.

6. The collaborative information mutual feedback prediction method for crack identification according to claim 1, characterized in that: The fracture-sensitive seismic attributes include: coherence volume, variance volume, curvature volume and ant volume.

7. The collaborative information mutual feedback prediction method for crack identification according to any one of claims 1 to 5, characterized in that: Also includes: Obtain earthquake sample data to be predicted; The earthquake sample data to be predicted is identified based on the fracture identification model to obtain the fracture identification result.

8. A collaborative information mutual feedback prediction device for crack identification, configured to implement the collaborative information mutual feedback prediction method for crack identification according to any one of claims 1 to 7, the device comprising: An acquisition module is used to pre-process the original seismic data to obtain a two-dimensional seismic profile; Extraction module, used to extract features from two-dimensional seismic profiles and obtain fracture-sensitive seismic attributes; The fusion module is used to fuse the fracture-sensitive seismic attributes to obtain the preliminary fracture identification results; A correction module is used to correct the initial crack identification results to obtain crack training samples; The training module is used to input crack training samples into the preset network for training to obtain a crack recognition model.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the collaborative information mutual feedback prediction method for crack identification according to any one of claims 1 to 7 is implemented.

Images