Method for predicting fracture parameters based on waveform classification

Abstract

The invention discloses a method for predicting fracture parameters based on waveform classification. The method comprises the following steps: S1, performing calculation and data reconstruction processing on'center angle-attribute 'of each data point of each center angle; s2, performing curve fitting on the'central angle-attribute 'data pair in a two-dimensional coordinate axis, and taking a fitted curve as a crack response waveform of the data point; and S3, designing the classification number of crack response waveforms and model channels, carrying out waveform classification on the model channels on a plane, and determining the crack strength and azimuth angle value of each type of model channels. The method is a technology for predicting different types of fracture development conditions by classifying fracture response waveforms of'central angle-attribute 'data pairs of data points, and can accurately and effectively predict different types of fracture distribution conditions.

Description

Technical Field

[0001] The present invention relates to the field of seismic exploration technology, belongs to seismic data interpretation technology, and is a method for predicting fracture development parameters by performing waveform classification on fracture response waveforms of "central angle-attribute" data pairs of each data point (CDP point). Background Art

[0002] Fractures are important channels for underground oil and gas accumulation and migration. In most cases, fractures are mainly structural fractures: fractures attributed to or associated with local tectonic events, including fracture systems related to faults, fracture systems related to uplift and arching, fracture systems related to folds, etc. The terrestrial Xujiahe Formation tight sandstone gas reservoirs in the Sichuan Basin are closely related to the development of fractures. Large fractures damage sandstone reservoirs, while medium and small fractures play a positive role in the later fracturing of sandstone reservoirs and the communication of gas-bearing reservoirs. It can be seen that different types of fractures play different roles in tight sandstone reservoirs.

[0003] At present, there are two ways to use seismic data to predict underground fractures, and both are related to pre-stack and post-stack seismic data. In addition, there are also geological empirical analysis techniques such as finite element analysis and tectonic stress field analysis to predict fractures. One of them uses azimuth seismic stacking data, guided by the theory of azimuth anisotropy, to determine an ellipse with a central coordinate of (0, 0) and an arbitrary angle θ with the X-axis. Three parameters need to be determined: the major axis radius a, the minor axis radius b and θ, so at least three known points are required to determine an ellipse. The ratio of the major axis and the minor axis of these fitted ellipses is calculated to calculate the fracture intensity and direction. Some geophysical workers have also achieved relatively more results in predicting fracture reservoirs using seismic reflection waveform classification. Among them, the core algorithm "self-organizing neural network" in waveform classification is a neural network with learning function. It is actually a nonlinear mapping that can reflect the topological relationship of each mode in the signal space almost unchanged on the network output. It is mainly composed of two parts: input and output. The input is one-dimensional, the output can be multi-dimensional, and the output node is widely connected with other nodes in the neighborhood. Each input neuron has a feedforward connection with each output neuron. Through adaptive, self-organizing learning and continuous adjustment of weights, all neurons in each neighborhood of the neural network have the same output for a certain input when the neural network is stable.

[0004] There are already quite a few patented technical documents on the prediction of different types of cracks, which shows that the detection of cracks has always been a hot topic of research and exploration. Different types of cracks may have different seismic reflection waveforms, but some, such as medium and small cracks, have similar seismic reflection waveforms, which make it difficult to distinguish and identify them from the seismic reflection waveforms. Some patents and technical documents, such as the establishment of fracture distribution model and its seismic response proposed by Fan Guozhang and Mou Yongguang ("Changes in P-wave velocity in fracture anisotropic media and its influence on the stacking of common center point gathers" published by Fan Guozhang and Mou Yongguang in "Petroleum Geophysical Exploration" in 2002; the invention of "A method and device for seismic waveform analysis and reservoir prediction" (patent number: 201210135544.X) proposes to use the seismic reflection waveform of the target layer to establish multiple model traces, classify different waveforms according to the self-organizing neural network, generate seismic phase classification map, and further form sedimentary phase map, and use sedimentary phase map for reservoir prediction; the invention of "fracture prediction method and device" (patent number: 201010205983.4) patent proposes to use the picked target layer time window to obtain the reflection amplitude of each seismic trace, and perform ellipse fitting through the obtained azimuth and reflection amplitude, so as to predict the direction and density of fractures. However, these methods are quite difficult to implement fracture prediction, and are easily affected by various factors in operation. There are mainly the following problems:

[0005] (1) The prediction results of conventional fracture prediction technology for different types of fractures are often not consistent with well data, and the prediction difficulty is relatively large.

[0006] (2) Waveform classification technology is usually used in seismic phase identification, but it is relatively difficult to use this technology to further distinguish the seismic reflection waveforms of different types of fractures. Summary of the invention

[0007] In order to overcome the shortcomings of the above conventional technical methods and solve the problem of fracture strength prediction and evaluation, the present invention provides a corresponding technical process to solve the prediction, which can use the existing three-dimensional longitudinal wave seismic acquisition data to conveniently, simply and accurately predict the fracture strength of the target layer. The purpose of the present invention is to overcome the above and other shortcomings in the existing conventional fracture prediction technology, and make full use of the characteristics that the azimuth information of the pre-stack gather is relatively sensitive to the fracture response. To this end, the present invention provides a method for classifying and predicting the development parameters of different types of fractures based on the fracture response waveform of the "central angle-attribute" data pair of the data point (CDP point) processed by the pre-stack data. Summary of the invention:

[0009] S1. Reconstruct the data of the same data point in each central angle seismic attribute data volume and extract the attribute data of the target layer segment to form a "central angle-attribute" data pair for each data point.

[0010] S1-1. Use appropriate azimuth division parameters to divide the CDP gather data after dynamic correction into corresponding azimuth ranges to obtain relevant center angle gather data. The median of the azimuth range in the actual seismic data is the center angle, and the direction of the center angle represents the response direction of the seismic data in the azimuth range, that is, the center angle gather data is obtained, and it is required that the maximum offset distance in the center angle gather data should not be greater than 30° for the incident angle of the target layer. The azimuth of the seismic data is set to 0° in the north direction and 360° in the clockwise direction. Under normal circumstances, in principle, the number of divided center angles should be greater than or equal to five, and the division requirement of the azimuth range is that the number of superpositions of the divided center angle data should be basically uniform, so that the accuracy of the extracted attributes and attribute fitting curves is relatively high and is conducive to the calculation of the next step.

[0011] S1-2. Use post-stack 3D seismic data to perform "well-seismic" calibration to determine the top and bottom interfaces of the target layer. Perform manual interpretation, interpolation and smoothing on the top and bottom interfaces to obtain the relevant interpretation horizon of the target layer and facilitate subsequent extraction and calculation of target layer attributes.

[0012] S1-3, stack and offset the data of each center angle gather to obtain the corresponding center angle seismic data volume, reconstruct and extract attributes to form the "center angle-attribute" data pair of each data point. Specifically:

[0013] S1-3-1. Reconstruct each central angle seismic data channel according to the same data point (CDP point) to form the channel gather data of the reconstructed data point (CDP point).

[0014] S1-3-2. Extract the seismic attributes of the target layer segment based on the interpreted horizon data of the target layer segment and the gather data of the reconstructed data points, and obtain the "central angle-attribute" data pair of the target layer segment of each data point. The types of seismic attributes may include attribute data such as frequency, amplitude, double-layer travel time, velocity, etc. caused by fractures and related to anisotropy calculations, and one of the attribute data is preferably used for attribute extraction calculation, requiring that the attribute can effectively highlight the response of the fracture, and the difference in attribute data values ​​between fractures and non-fractures is relatively obvious.

[0015] S2. Perform curve fitting on the “central angle-attribute” data pair of each data point to obtain the crack response waveform of each data point.

[0016] S2-1. Establish a two-dimensional coordinate system (θ, f), where the abscissa is the angle value θ and the ordinate is the attribute value f.

[0017] S2-2, the "central angle (θ i )—seismic attributes (fi )” data pairs are projected into the coordinate system, and the least squares binary curve fitting is performed on these data pairs at this point to obtain the relevant fitting curve. The fitting curve is used as the fracture response waveform (θi, fi, θi∈(0°,180°)) of the target layer segment at this data point, and proceeds to the next step.

[0018] S2-3, and so on, to complete the calculation of the crack response waveform of each data point.

[0019] S3. Classify the crack response waveforms of each data point and predict the distribution range of different types of cracks.

[0020] S3-1. Set the number of fracture classifications and create multiple fracture response waveform model channels. The number of fracture response waveform model channels is the same as the number of fracture classifications. In general, the number of fracture response waveform model channels is 3-5 times the number of sedimentary phases in the study area. In actual operation, the number of fracture waveform classifications and model channels can be determined based on relevant waveform classification test conditions, expert experience, fracture prediction accuracy, etc.

[0021] S3-2. Crack response waveforms of different data points are classified according to the self-organizing neural network to distinguish different types of cracks. The self-organizing neural network is trained on the actual crack response waveform. After several iterations, the self-organizing neural network constructs a synthetic crack response waveform, which is then compared with the actual crack response waveform. Through adaptive experiments and error processing, the synthetic waveform is changed after each iteration to find a better correlation between the model channel and the actual crack response waveform. Generally, the number of classifications used will generate the number of model channels. Generally, the larger the number of classifications, the higher the accuracy of predicting different types of cracks, and the longer the calculation time. The smaller the number of classifications, the lower the prediction accuracy and the shorter the calculation time. The calculation steps of the self-organizing neural network are as follows:

[0022] Assume the number of neurons is N, and the total neuron vector is:

[0023] The input modes are:

[0024] The implementation steps are:

[0025] ① Initialize the weight vector When selecting random values, select the initial learning rate α(0) and the neighborhood function ω m (0);

[0026] ② The sample vector Input to the input layer;

[0027] ③Select the weight vector that best matches The neuron is the winning neuron, the winning neuron is labeled m, and satisfies

[0028]

[0029] ④ Train the weight vector to move the neurons in the active bubble range toward the input vector direction:

[0030]

[0031] ⑤Update the learning rate α(t);

[0032] ⑥Reduce the neighborhood function: ω m (t); Finally check the end condition and exit when the weight no longer changes significantly, otherwise continue to step ②.

[0033] Through competitive learning, the neighboring neuron weight vectors can be made close to the winning neighborhood neuron weight vector, and the winning neuron weight vector is closest to a certain learning sample vector. Therefore, if there is an obvious classification clustering relationship in the learning sample vector, then the distribution of neuron weight vectors in the competitive learning neural network presents a partitioning phenomenon. The neuron weight vectors in the same zone are close to each other and close to a certain class of sample vectors, thereby achieving the purpose of classification.

[0034] S3-3. Calculate the crack intensity and azimuth values ​​according to the crack response waveform model, so as to obtain the crack intensity and direction of each crack classification. Specifically, the maximum and minimum attribute values ​​in the crack response waveform model are determined. After the two values ​​are calculated by the relevant crack intensity calculation formula, the data value obtained is used as the crack intensity value of the crack response waveform model; in addition, the azimuth angle corresponding to the maximum value of the attribute or adding or subtracting 90° can be used as the crack azimuth of the model. Whether to add or subtract 90° depends on the nature of the attribute response to the crack and expert experience. In principle, the crack azimuth can be understood as the horizontal angle from the north direction line of a point on the crack to the crack strike line in a clockwise direction. The azimuth and relative strength of the crack strike predicted in the seismic rose diagram design are symmetrically distributed at 180°, and the azimuth value is in the range of 0°-180°. Among them, the crack strength calculation formula is as follows:

[0035] P i =|P k | / |P l | (1)

[0036] In formula (1), P i is the crack strength value of the model channel, P k is the maximum attribute value of the model, P l The minimum attribute value of the model.

[0037] If there are well data in some model channels, and the relevant well data can determine the fracture intensity and azimuth of the model channel, then the fracture intensity and azimuth of the well can be used as the fracture intensity and fracture direction of each data point in the relevant model channel; other model channels without well data are calculated using formula (1) to obtain the fracture intensity and azimuth of the relevant model channel. In addition, the fracture intensity calculation method of the present invention and the fracture intensity value on the well can be combined to calculate the fracture intensity on the well and the fracture intensity value obtained by the fracture intensity calculation of the present invention at the corresponding data point, thereby calculating the fracture intensity value of each model channel (this method requires two or more well data for fitting calculation). Another method for calculating the fracture intensity and azimuth of a model channel is to use ellipse fitting calculation to obtain the fracture intensity and azimuth of a certain data point in the model channel, and the obtained fracture parameters are used as the fracture intensity and direction of the model channel. Similarly, the calculation of the fracture intensity and azimuth of each model channel is completed, so that the fracture intensity and direction can be interpreted on the plane after waveform classification.

[0038] S4. The display method of the crack waveform classification result is to obtain a waveform classification plane diagram, and three display annotation bars are set in the blank space on the right side of the diagram. One display annotation bar is the classification model color, the second display annotation bar is the crack intensity corresponding to the classification color, and the third display annotation bar is the crack azimuth value corresponding to the classification color.

[0039] Preferably, the present application also provides an electronic device, comprising: a memory storing executable instructions; a processor, the processor running the executable instructions in the memory to implement the method for predicting cracks based on curve waveform classification.

[0040] Preferably, the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the method for predicting cracks based on curve waveform classification is implemented.

[0041] Beneficial effects of the present invention

[0042] The present invention is a technology that uses the fracture response waveform of the "central angle-attribute" data pair of data points to classify and predict the development of different types of fractures, which can accurately and effectively predict the distribution of different types of fractures. By extracting the "central angle-attribute" data pair of the target layer of each data point and establishing a two-dimensional coordinate system to obtain the relevant attribute fitting curve; the fitting curve is used as the fracture response waveform of the point; secondly, a reasonable number of classifications is designed to classify the fracture response waveform; and the fracture intensity and azimuth of the model channel are calculated using the relevant calculation formula; the relevant data is displayed on a plane map to obtain the plane distribution of different types of fracture development parameters, which can greatly improve the exploration economic benefits of fracture reservoirs.

[0043] The technology of the present invention predicts the distribution areas of different types of fractures more clearly, accurately and clearly than other methods. Compared with traditional fracture prediction methods, it can reduce multiple solutions and accurately detect different types of fracture reservoirs. Taking a certain exploration block in the Sichuan Basin as an example, when drilling encounters a tight sandstone reservoir with medium and small fractures, a medium and high-yield industrial gas flow can be obtained through fracturing testing. Through the known drilling in the study area, the results of the invention are well consistent with the drilling data, and the distribution areas of different types of fractures can be accurately predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the technical flow chart of this invention.

[0045] Figure 2 It is a schematic diagram of the relationship between the azimuth angle and the central angle in the technical example of this invention. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto:

[0047] Example 1

[0048] According to the technical process of this invention ( Figure 1 ), formulate work steps, and an example is to predict the distribution area of ​​different types of fracture reservoirs in the marine shale reservoir section of a certain three-dimensional work area.

[0049] In step S1, the seismic gather data is stacked and offset in different azimuth ranges ( Figure 2), with central angles of 18°, 54°, 90°, 126°, 160°, etc., totaling five central angle seismic data bodies - the azimuth ranges are 0°~36°, 36°~72°, 72°~108°, 108°~144°, 144°~180° respectively. According to the designed target layer buried depth of about 3400m, the stacked shot distance range is limited to 380m~3200m, so that the maximum incident angle of the target layer is limited to within 30°, and the stacked data of each central angle are offset respectively to return the reflection wave group to its position.

[0050] Well logging data and three-dimensional post-stack seismic data are used for "well-seismic" calibration to determine the target layer segment, and the top and bottom interfaces of the target layer are interpreted in the horizon. In the example, the top and bottom interface reflections of the shale target layer are selected as the trough and the peak, respectively, and the two horizons are tracked and interpreted. The horizon interpretation parameters start from 1280 lines and 200 channels, and end at 3100 lines and 1600 channels. According to the actual situation of the seismic data, the interpretation grid is set to 10 lines x 10 channels. After the interpretation, the horizon is interpolated and smoothed to 1 line x 1 channel.

[0051] In this example, based on well data and expert experience, the starting attenuation frequency is selected as the attribute data. After reconstructing the data of the five central angle seismic data channels at the same data point, the starting attenuation frequency attribute value of the target layer segment is extracted to form a "central angle-attribute" data pair at each data point (CDP point).

[0052] In step S2, a two-dimensional coordinate system is established, with the horizontal axis being the angle value and the vertical axis being the attribute value of the starting attenuation frequency. The five "central angle-attribute" data pairs on a single data point are projected into the coordinate system, and the five attribute data pairs are fitted by the least square method to obtain the relevant fitting curve. The fitting curve is used as the crack response waveform of the data point. Similarly, the crack response waveform calculation of each data point is completed.

[0053] In step S3, the number of classifications is designed and the fracture response waveform is classified using a self-organizing neural network. According to the accuracy to be predicted and the experience of experts, the number of fracture classifications is set to 22 in the study area, that is, different fracture types in the study area can be divided into 22 model channels - different fracture types have different fracture development directions and fracture strengths, and the fracture strength and direction are calculated according to the relevant data of the model channels. In the example, different types of fracture response model channels are set to display colors, that is, different types of fractures are set to display different colors. Since the attribute calculated for the fracture direction is the starting attenuation frequency, the maximum value of the "central angle-attribute" is perpendicular to the fracture direction - the corresponding central angle value is added or subtracted by 90° to obtain the fracture direction. According to the plane diagram of the distribution of different types of fracture waveforms calculated by the present invention, the well data of subsequent drilling in the study area show that the prediction results of the present invention technology are consistent with the fracture development direction and strength in the subsequent drilling wells, and the compliance rate is relatively better than the results of conventional fracture prediction methods, proving that the invention technology is effective.

[0054] Example 2

[0055] The present invention discloses an electronic device, which includes: a memory storing executable instructions; a processor running the executable instructions in the memory to implement the above-mentioned method for predicting cracks based on curve waveform classification.

[0056] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.

[0057] The memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0058] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.

[0059] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0060] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0061] Example 3

[0062] An embodiment of the present disclosure provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for predicting cracks based on curve waveform classification is implemented.

[0063] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the above-mentioned methods of each embodiment of the present disclosure are executed.

[0064] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).

[0065] Those skilled in the art should understand that the purpose of the above description of the embodiments of the present invention is only to exemplarily illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any given examples.

[0066] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for predicting crack parameters based on waveform classification, characterized in that: It includes the following steps: S1. Calculate and reconstruct the "central angle-attribute" of each data point of each central angle; S2. Perform curve fitting on the "central angle-attribute" data pair in the two-dimensional coordinate axis, and use the fitting curve as the crack response waveform of the data point; S3. Design the classification number and model channels of the crack response waveform and classify the waveform on the plane, and determine the crack intensity and azimuth value of each type of model channel.

2. The method according to claim 1, characterized in that It also includes the steps: S4, displaying the waveform classification results in a plane diagram, setting three display annotation bars in the blank space of the diagram, the first display annotation bar is the classification model color, the second display annotation bar is the crack intensity corresponding to the classification color, and the third display annotation bar is the crack azimuth value corresponding to the classification color.

3. The method according to claim 1, characterized in that The specific steps of S1 are: S1-1. The CDP gather data after dynamic correction is divided into corresponding azimuth ranges using appropriate azimuth division parameters to obtain relevant center angle gather data; the median of the azimuth range in the actual seismic data is the center angle, and the direction of the center angle represents the response direction of the seismic data in the azimuth range, that is, the center angle gather data is obtained, and the maximum offset distance in the center angle gather data is required to be no more than 30o for the incident angle of the target layer; the azimuth of the seismic data is set to 0o in the due north direction and 360o in the clockwise direction; S1-2. Use post-stack 3D seismic data to perform "well-seismic" calibration to determine the top and bottom interfaces of the target layer segment; perform manual interpretation, interpolation and smoothing on the top and bottom interfaces to obtain the relevant interpretation horizon of the target layer segment and facilitate the subsequent extraction and calculation of the target layer segment attributes; S1-3. Superimpose and offset the data of each central angle gather to obtain the corresponding central angle seismic data volume, reconstruct and extract attributes to form the "central angle-attribute" data pair of each data point.

4. The method according to claim 3, characterized in that In S1-3, the specific steps are: S1-3-1, reconstructing each central angle seismic data channel according to the same data point to form the channel gather data of the reconstructed data point; S1-3-2. Extract seismic attributes of the target layer segment based on the interpreted stratigraphic data of the target layer segment and the gather data of the reconstructed data points, and obtain the "central angle-attribute" data pairs of the target layer segment at each data point; the types of attributes include frequency, amplitude, double-layer travel time, and velocity attribute data related to anisotropy calculation caused by fractures, and select one of the attribute data for attribute extraction calculation.

5. The method according to claim 1, characterized in that The specific steps of S2 are: S2-1, establish a two-dimensional coordinate system (θ, f), the abscissa of the system is the angle value θ, and the ordinate is the attribute value f; S2-2, the "central angle-attribute" data pair (θ i —f i ) is projected into the coordinate system, and the data pairs of the point are fitted by the least squares binary curve to obtain a relevant fitting curve; the fitting curve is used as the fracture response waveform of the target layer segment of the data point; S2-3. Repeat step S2-2 to complete the calculation of the crack response waveform for each data point.

6. The method according to claim 1, characterized in that The specific steps of S3 are: S3-1, setting the number of crack classifications and creating multiple crack response waveform model channels, the number of crack response waveform model channels being the same as the number of crack classifications; S3-2, classifying the crack response waveforms of different data points according to the self-organizing neural network to distinguish different types of cracks; S3-3. Calculate the crack intensity and azimuth value according to the crack response waveform model, so as to obtain the crack intensity and direction of each crack classification.

7. The method according to claim 6, characterized in that In S3-2, the calculation steps of the self-organizing neural network are as follows: Assume the number of neurons is N, and the total neuron vector is: The input modes are: The implementation steps are: ① Initialize the weight vector When selecting random values, select the initial learning rate α(0) and the neighborhood function ω m (0); ② The sample vector Input to the input layer; ③Select the weight vector that best matches The neuron is the winning neuron, the winning neuron is labeled m, and satisfies ④ Train the weight vector to move the neurons in the active bubble range toward the input vector direction: ⑤Update the learning rate α(t); ⑥Reduce the neighborhood function: ω m (t); Finally check the end condition and exit when the weight no longer changes significantly, otherwise continue to step ②.

8. The method according to claim 6, characterized in that In S3-3, the calculated crack strength and azimuth values ​​are obtained by one of the following methods: ① Directly obtain fracture strength and azimuth through well data of the model track; ② The crack strength is obtained by formula (1): P i =|P k | / |P l | (1) Among them, P i is the crack strength value of the model, P k is the maximum attribute value of the model, P l is the minimum attribute value of the model; Obtaining the azimuth angle: The azimuth angle corresponding to the maximum value of the attribute plus or minus 90° is used as the fracture azimuth angle of the model; plus or minus 90° depends on the nature of the attribute response to the fracture and expert experience; ③ Obtain fracture strength and azimuth values ​​through well data fitting; ④ Use ellipse fitting to calculate the crack strength and azimuth angle of a data point in the model.

9. An electronic device, comprising: A memory storing executable instructions; A processor, characterized in that the processor runs the executable instructions in the memory to implement the method for predicting cracks based on curve waveform classification.

10. A computer-readable storage medium storing a computer program, characterized in that When the computer program is executed by a processor, the method for predicting cracks based on curve waveform classification is implemented.

Images