Shale gas reservoir parameter prediction model training, reservoir parameter prediction method and device
By employing random forest machine learning technology, and utilizing the transformation and rotational recombination of P-wave impedance, S-wave impedance, and density well bypass curves, the most correlated attribute curves are selected to train a random forest regression model. This solves the error and small sample problem in shale gas reservoir parameter prediction, and achieves high-precision quantitative prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PETROCHINA CO LTD
- Filing Date
- 2021-12-30
- Publication Date
- 2026-06-30
AI Technical Summary
Existing technologies based on rock physics models for predicting shale gas reservoir parameters have significant errors, making it difficult to achieve quantitative exploration. Furthermore, machine learning methods rely on a large amount of known attribute data, while data resources in the field of shale gas seismic exploration are limited, resulting in a small sample problem.
By using random forest machine learning technology, the P-wave impedance, S-wave impedance, and density well bypass curves are converted into base impedance curves and rotated and recombined. The final attribute curves with the highest correlation to the logging curves of the target reservoir parameters are selected, and a random forest regression model is trained to predict shale gas reservoir parameters.
Quantitative prediction of shale gas reservoir parameters was achieved under small sample conditions, improving prediction accuracy and overcoming dependence on the amount of input attribute data.
Smart Images
Figure CN116432508B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of geophysical exploration technology, and in particular to shale gas reservoir parameter prediction model training, reservoir parameter prediction methods and devices. Background Technology
[0002] Shale gas, as a crucial global resource, presents both a hot research topic and a significant challenge in its exploration and development. Shale gas reservoir parameters, such as brittleness, total organic carbon (TOC) content, porosity, total gas content, and effective pressure, are essential for shale gas reservoir prediction. One effective method for obtaining these parameters is pre-stack seismic inversion driven by rock physics models. Seismic data possesses good spatial continuity and plays a vital role in reservoir characterization. The core of pre-stack seismic inversion of shale gas reservoir parameters lies in establishing a deterministic relationship between seismic amplitude response and reservoir parameters based on a rock physics model. Summary of the Invention
[0003] The inventors discovered that due to the extremely complex conditions for shale gas accumulation, there is a highly nonlinear relationship and uncertainty between seismic response characteristics and reservoir parameters, making it difficult to accurately establish the inverse problem of rock physics. Conventional techniques for shale gas reservoir parameter estimation based on rock physics models have significant errors and fail to meet the requirements of quantitative exploration. The emergence of machine learning technology has made it possible to quantitatively predict shale gas reservoir parameters.
[0004] The inventors discovered that among numerous machine learning techniques, random forest machine learning stands out due to its high accuracy, robustness to noise, and minimal parameter tuning requirements, achieving significant application results in many fields. However, like other machine learning techniques, random forest machine learning requires a large amount of manually input known attribute data as training data, and the quantity and quality of these attributes directly determine the accuracy of the prediction results. In the field of shale gas seismic exploration, data resources are limited, with a severe lack of known attribute data, leading to a prominent small sample size problem. To at least partially address the technical problems of existing technologies, the inventors developed this invention, providing a shale gas reservoir parameter prediction model training method and apparatus through specific implementation methods. This method has low dependence on the amount of input attribute data and achieves high prediction accuracy.
[0005] In a first aspect, embodiments of the present invention provide a method for training a shale gas reservoir parameter prediction model, comprising:
[0006] Based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area, multiple basic impedance curves with different seismic wave incident angles were obtained through a preset basic impedance equation.
[0007] By combining the basic impedance curves in pairs, multiple pairs of basic impedance curves are obtained. The two basic impedance curves in each pair are rotated and then recombined to obtain multiple basic property curves with different rotation angles.
[0008] From the multiple basic attribute curves of each pair of basic impedance curves, the basic attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameters is selected as the final attribute curve.
[0009] A training set is formed by the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters. The training set is used to train a specified random forest regression model to obtain a regression model for predicting the target reservoir parameters in the study area.
[0010] Secondly, embodiments of the present invention provide a method for predicting shale gas reservoir parameters, including:
[0011] Based on the set parameters of the selected final attribute curve, the longitudinal wave impedance, transverse wave impedance and density data volume of the study area are converted into the final attribute data volume.
[0012] The final attribute data volume is input into the regression model trained according to the above-mentioned shale gas reservoir parameter prediction model training method, and the target reservoir parameter data volume of the study area is obtained based on the output of the regression model.
[0013] Thirdly, embodiments of the present invention provide a shale gas reservoir parameter prediction model training device, comprising:
[0014] The base impedance curve determination module is used to obtain multiple base impedance curves with different seismic wave incident angles based on the P-wave impedance, S-wave impedance and density well bypass curve of the study area through a preset base impedance equation.
[0015] The basic property curve determination module is used to combine the basic impedance curves in pairs to obtain multiple pairs of basic impedance curves. The two basic impedance curves in the basic impedance curve pairs are rotated and then recombined to obtain multiple basic property curves with different rotation angles.
[0016] The final attribute curve module is used to select the base attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameters from multiple base attribute curves of each base impedance curve pair as the final attribute curve;
[0017] The training module is used to train a specified random forest regression model using a training set consisting of the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters, so as to obtain a regression model for predicting the target reservoir parameters in the study area.
[0018] Fourthly, embodiments of the present invention provide a shale gas reservoir parameter prediction device, comprising:
[0019] The final attribute data volume determination module is used to convert the longitudinal wave impedance, transverse wave impedance and density data volume of the study area into the final attribute data volume according to the set parameters of the selected final attribute curve.
[0020] The target reservoir parameter prediction module is used to input the final attribute data volume into the regression model trained according to the above-mentioned shale gas reservoir parameter prediction model training method, and obtain the target reservoir parameter data volume of the study area based on the output of the regression model.
[0021] Fifthly, embodiments of the present invention provide a computer program product with reservoir parameter prediction function, including a computer program / instruction, wherein when the computer program / instruction is executed by a processor, it implements the above-mentioned shale gas reservoir parameter prediction model training method, or implements the above-mentioned shale gas reservoir parameter prediction method.
[0022] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:
[0023] The shale gas reservoir parameter prediction model training method provided in this invention transforms the P-wave impedance, S-wave impedance, and density well bypass curves into multiple basic impedance curves with different seismic wave incident angles through a preset basic impedance equation. These basic impedance curves are then combined in pairs and rotated to obtain multiple basic attribute curves with different rotation angles. A final attribute curve is selected from these based on its correlation coefficient with the target reservoir parameter logging curves. The final attribute curves of each pair of basic impedance curves and the target reservoir parameter logging curves form a training set to train a specified random forest regression model. This method requires only P-wave impedance, S-wave impedance, density well bypass curves, and the target reservoir parameter logging curves, overcoming the quantitative dependence of input attribute data on target reservoir parameter prediction in existing technologies. It is suitable for predicting target reservoir parameters in shale gas reservoirs with small sample conditions. The trained model can achieve quantitative prediction of shale gas reservoir target parameters with high accuracy.
[0024] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0025] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:
[0026] Figure 1 This is a flowchart of the shale gas reservoir parameter prediction model training method in Embodiment 1 of the present invention;
[0027] Figure 2 This is a schematic diagram of the longitudinal wave impedance, transverse wave impedance, and density well bypass curves in Embodiment 1 of the present invention.
[0028] Figure 3 This is a schematic diagram of two base impedance curves in Embodiment 1 of the present invention;
[0029] Figure 4 for Figure 3 The basic property curve is obtained by rotating and combining the two basic impedance curves in the curve.
[0030] Figure 5 for Figure 4 The mid-base property curve is the curve after removing the sedimentary background;
[0031] Figure 6 This is a schematic diagram of the five curves in the final attribute curve after redundancy removal and the target reservoir parameter curve.
[0032] Figure 7 This is a flowchart of the shale gas reservoir parameter prediction method in Embodiment 2 of the present invention;
[0033] Figure 8 This is a comparison chart of the predicted total gas content curve of a single well and the actual curve in Embodiment 2 of the present invention;
[0034] Figure 9 This is a two-dimensional profile in the target reservoir parameter data volume predicted in Embodiment 2 of the present invention;
[0035] Figure 10 This is a schematic diagram of the structure of the shale gas reservoir parameter prediction model training device in an embodiment of the present invention;
[0036] Figure 11 This is a schematic diagram of the shale gas reservoir parameter prediction device in an embodiment of the present invention. Detailed Implementation
[0037] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0038] It should be understood that the terminology used in this invention is merely for describing particular embodiments and is not intended to limit the invention. Furthermore, with respect to numerical ranges in this invention, it should be understood that each intermediate value between the upper and lower limits of the range is also specifically disclosed. Every smaller range between any stated value or intermediate value within a stated range, and any other stated value or intermediate value within said range, is also included in this invention. The upper and lower limits of these smaller ranges may be independently included or excluded from the range.
[0039] Unless otherwise stated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. While only preferred methods and materials have been described herein, any methods and materials similar or equivalent to those described herein may be used in the implementation or testing of this invention. All references to this specification are incorporated by way of citation to disclose and describe methods and / or materials associated with those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.
[0040] In the description of this invention, it should be noted that the terms "comprising," "including," "having," "containing," etc., are all open-ended terms, meaning that they include but are not limited to. Furthermore, the terms "first," "second," and "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0041] To address the significant errors in existing rock physics model-driven shale gas reservoir parameter prediction technologies, which fail to meet the requirements of quantitative exploration, and the dependence of machine learning methods on sample size, this invention provides a shale gas reservoir parameter prediction model training, reservoir parameter prediction method, and apparatus. The training process has low dependence on the amount of input attribute data, and the model prediction accuracy is high.
[0042] Example 1
[0043] Embodiment 1 of the present invention provides a training method for a shale gas reservoir parameter prediction model, the process of which is as follows: Figure 1 As shown, it includes the following steps:
[0044] Step S11: Based on the P-wave impedance, S-wave impedance and density well bypass curve of the study area, obtain multiple base impedance curves with different seismic wave incident angles through the preset base impedance equation.
[0045] In some optional embodiments, it may include obtaining multiple basic impedance curves with different seismic wave incident angles based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area using the following basic impedance equation:
[0046]
[0047] Where BI(θ) represents the fundamental impedance of the seismic wave at an incident angle θ, which is dimensionless, and the unit of θ is degrees; a(θ) = cos(θ) + sin(θ), b(θ) = -8Ksin(θ), c(θ) = cos(θ) - 4Ksin(θ), and K is the ratio of the shear wave velocity to the p-wave velocity; Z p Z represents the longitudinal wave impedance, measured in meters per second * kilograms per cubic meter. sρ is the shear wave impedance, measured in meters per second * kilograms per cubic meter; ρ is the formation density, measured in kilograms per cubic meter. ρ0 and ρ0 represent the average P-wave impedance, S-wave impedance, and density of the target layer in the study area, respectively; M is a dynamic normalization parameter used to determine the target layer's impedance based on the target layer's impedance. The order of magnitude, from different angles Corrected to the same order of magnitude, dimensionless.
[0048] The acquisition of P-wave impedance, S-wave impedance, and density well-bypass curves can include obtaining P-wave impedance, S-wave impedance, and density data volumes of the study area using conventional pre-stack seismic inversion techniques, and extracting P-wave impedance, S-wave impedance, and density well-bypass curves from the P-wave impedance, S-wave impedance, and density data volumes, respectively.
[0049] See Figure 2 The figures shown are schematic diagrams of longitudinal wave impedance, transverse wave impedance, and density well bypass curves, respectively.
[0050] Furthermore, the incident angle of the aforementioned seismic waves can be determined in the following manner:
[0051] Multiple seismic wave incident angles are determined by setting the range and step size of the seismic wave incident angle variation.
[0052] For example, the range of the incident angle of the seismic wave can be [-90°, 90°], with a step size of 5°.
[0053] See Figure 3 As shown, the base impedance curves BI(15°) and BI(65°) are obtained for seismic wave incident angles of 15° and 65°, respectively. It can be seen that the trends of the two curves are quite different, representing different reservoir information. By calculating the base impedance curves at different angles, reservoir information can be enriched.
[0054] In some embodiments, after obtaining multiple fundamental impedance curves with different seismic wave incident angles, the method may further include correcting all the obtained fundamental impedance curves to the same order of magnitude.
[0055] Step S12: Combine the base impedance curves in pairs to obtain multiple sets of base impedance curve pairs. Rotate the two base impedance curves in each base impedance curve pair and then recombine them to obtain multiple base property curves with different rotation angles.
[0056] The two fundamental impedance curves in the fundamental impedance curve pair are rotated and recombined. Specifically, this can include recombining the two fundamental impedance curves in the fundamental impedance curve pair with a seismic wave incident angle of θ. i The fundamental impedance curve BI(θ) i ) and the incident angle of the seismic wave is θ j The fundamental impedance curve BI(θ) j Rotate ω as follows: kAfter recombining the angles, a rotation angle of ω is generated. k The base property curve A(ω) k ):
[0057] A(ω k )=BI(θ i )·sin(ω k )+BI(θ j )·cos(ω k ).
[0058] Furthermore, the rotation angle is determined as follows:
[0059] Multiple rotation angles can be determined by setting the rotation range and rotation step size.
[0060] Figure 4 for Figure 3 The two base impedance curves are rotated by 30° to form the base property curve A(30°). By comparing it with the original base impedance curve, it can be seen that the shape of the base property curve has changed, and the rotation operation has further added new information.
[0061] Step S13: From the multiple basic attribute curves of each pair of basic impedance curves, select the basic attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameters as the final attribute curve.
[0062] In some embodiments, before step S13 is performed, the sedimentary background is first removed from the target reservoir parameter logging curves and multiple basic attribute curves of each pair of basic impedance curves in the following manner:
[0063] R(X)=ln(X)-t(ln(X))
[0064] Where R(X) represents the curve X after background removal, ln(X) represents the logarithmic operation on curve X, and t(ln(X)) represents the Chebyshev filtering operation on ln(X).
[0065] See Figure 5 As shown, Figure 4 The mid-base property curve after removing the sedimentary background.
[0066] After removing the deposition background from the above curves, a significant amount of attribute information can be added.
[0067] Step S14: A training set is formed by the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters. The training set is used to train a specified random forest regression model to obtain a regression model for predicting the target reservoir parameters in the study area.
[0068] Specifically, the final attribute curves in the training set are deredundant using a specified random forest regression model; the random forest regression model is then trained using the deredundant training set.
[0069] Furthermore, a random forest pre-training process is performed on the final attribute curves in the training set using a specified random forest regression model to obtain a final attribute curve tree; the final attribute curves with the same node position in the tree are deredundant, and only one of the final attribute curves is retained.
[0070] To address the issue that shale gas reservoir property predictions often involve small samples, this paper utilizes the redundancy-removed final attribute curves and the target reservoir parameter logging curves as the forest training dataset to formally train the random forest regression model. When dividing the nodes of the redundancy-removed final attribute curves, no criteria are relied upon, and the division points are randomly selected to increase the generalization ability of small sample training.
[0071] Figure 6 Five curves from the final attribute curves selected after redundancy removal, as well as the target reservoir parameter (target reservoir physical property parameter) curve (taking the prediction of total gas content as an example).
[0072] The shale gas reservoir parameter prediction model training method provided in Embodiment 1 of this invention transforms the P-wave impedance, S-wave impedance, and density well bypass curves into multiple basic impedance curves with different seismic wave incident angles through a preset basic impedance equation. These basic impedance curves are then combined in pairs and rotated to obtain multiple basic attribute curves with different rotation angles. A final attribute curve is selected from these based on its correlation coefficient with the target reservoir parameter logging curves. The final attribute curves of each pair of basic impedance curves and the target reservoir parameter logging curves form a training set to train a specified random forest regression model. This method requires only P-wave impedance, S-wave impedance, density well bypass curves, and the target reservoir parameter logging curves, overcoming the dependence of existing technologies on the quantity of input attribute data for predicting target reservoir parameters through machine learning. It is suitable for predicting target reservoir parameters in shale gas reservoirs with small sample conditions. The trained model can achieve quantitative prediction of shale gas reservoir target parameters with high accuracy.
[0073] Example 2
[0074] Embodiment 2 of the present invention provides a method for predicting shale gas reservoir parameters, the process of which is as follows: Figure 7 As shown, it includes the following steps:
[0075] Step S71: Based on the set parameters of the selected final attribute curve, convert the longitudinal wave impedance, transverse wave impedance and density data volume of the study area into the final attribute data volume.
[0076] The settings parameters include the following:
[0077] The first and second seismic wave incident angles of the two base impedance curves that make up the selected final attribute curve, and the rotation angle of the selected final attribute curve.
[0078] Furthermore, the final attribute curve with the highest correlation coefficient to the logging curve of the target reservoir parameters can be selected as the final attribute curve.
[0079] Furthermore, a final attribute curve is selected from the redundancy-free final attribute curves obtained in Example 1.
[0080] The specific implementation of step S71 may include: obtaining a first basic impedance data volume with a first seismic wave incident angle and a second basic impedance data volume with a second seismic wave incident angle by using a preset basic impedance equation to obtain the P-wave impedance, S-wave impedance and density data volume of the study area; and recombining the first basic impedance data volume and the second basic impedance data volume by rotating them according to the rotation angle to obtain the final attribute data volume.
[0081] Step S72: Input the final attribute data volume into the regression model, and obtain the target reservoir parameter data volume of the study area based on the output of the regression model.
[0082] The regression model is a regression model trained according to the shale gas reservoir parameter prediction model training method in Example 1 above.
[0083] The target reservoir parameter can be any of the following parameters:
[0084] Brittleness, total organic carbon, porosity, total gas content, and effective pressure.
[0085] Optionally, other types of parameters that can be used for reservoir prediction may also be used, and the specific parameter types are not limited in this embodiment.
[0086] Figure 8 The image shows a comparison of the predicted and actual curves for the target reservoir properties of a single well, using total gas content as an example. It can be seen that the predicted curve (dashed line) and the actual curve (solid line) are very close.
[0087] Figure 9 This is a two-dimensional profile within the predicted target reservoir parameter data volumes (porosity data volume, total gas content data volume, and TOC content data volume, respectively). The dashed elliptical lines in the figure indicate favorable shale gas reservoir development zones determined based on the predicted target reservoir parameter values.
[0088] The above embodiments illustrate that the shale gas reservoir parameter prediction model training and reservoir parameter prediction method of the present invention have flexible scalability and transferability.
[0089] Based on the inventive concept of this invention, embodiments of this invention also provide a shale gas reservoir parameter prediction model training device, the structure of which is as follows: Figure 10 As shown, it includes:
[0090] The base impedance curve determination module 101 is used to obtain multiple base impedance curves with different seismic wave incident angles based on the P-wave impedance, S-wave impedance and density well bypass curve of the study area through a preset base impedance equation.
[0091] The basic attribute curve determination module 102 is used to combine the basic impedance curves in pairs to obtain multiple pairs of basic impedance curves. The two basic impedance curves in the basic impedance curve pairs are rotated and then recombined to obtain multiple basic attribute curves with different rotation angles.
[0092] The final attribute curve module 103 is used to select the base attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameter from multiple base attribute curves of each base impedance curve pair as the final attribute curve.
[0093] Training module 104 is used to train a specified random forest regression model using a training set consisting of the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters, so as to obtain a regression model for predicting the target reservoir parameters in the study area.
[0094] Based on the inventive concept of this invention, embodiments of this invention also provide a shale gas reservoir parameter prediction device, the structure of which is as follows: Figure 11 As shown, it includes:
[0095] The final attribute data volume determination module 111 is used to convert the longitudinal wave impedance, transverse wave impedance and density data volume of the study area into the final attribute data volume according to the set parameters of the selected final attribute curve.
[0096] The target reservoir parameter prediction module 112 is used to input the final attribute data body into the regression model trained according to the above-mentioned shale gas reservoir parameter prediction model training method, and obtain the target reservoir parameter data body of the study area based on the output of the regression model.
[0097] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0098] Based on the inventive concept of this invention, embodiments of this invention also provide a computer program product with reservoir parameter prediction function, including a computer program / instruction, wherein when the computer program / instruction is executed by a processor, it implements the above-mentioned shale gas reservoir parameter prediction model training method, or implements the above-mentioned shale gas reservoir parameter prediction method.
[0099] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.
[0100] In the detailed description above, various features are combined together in a single embodiment to simplify this disclosure. This approach to disclosure should not be construed as reflecting an intention that embodiments of the claimed subject matter require more features than are explicitly stated in each claim. Rather, as reflected in the appended claims, the invention is presented with fewer features than all of the features in a single disclosed embodiment. Therefore, the appended claims are hereby explicitly incorporated into the detailed description, with each claim representing a separate preferred embodiment of the invention.
[0101] The foregoing description includes examples of one or more embodiments. It is certainly impossible to describe all possible combinations of components or methods in order to describe the above embodiments, but those skilled in the art will recognize that further combinations and arrangements of the various embodiments are possible. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. Furthermore, the term "comprising" as used in the specification or claims is interpreted in a manner similar to the term "including," as interpreted when used as a conjunction in the claims. Additionally, the use of any term "or" in the specification of the claims is intended to mean "non-exclusive or."
Claims
1. A method for training a shale gas reservoir parameter prediction model, characterized in that, include: Based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area, multiple basic impedance curves with different seismic wave incident angles were obtained through a preset basic impedance equation. By combining the basic impedance curves in pairs, multiple pairs of basic impedance curves are obtained. The two basic impedance curves in each pair are rotated and then recombined to obtain multiple basic property curves with different rotation angles. From the multiple basic attribute curves of each pair of basic impedance curves, the basic attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameters is selected as the final attribute curve. A training set is formed by the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters. The training set is used to train a specified random forest regression model to obtain a regression model for predicting the target reservoir parameters in the study area. Based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area, multiple basic impedance curves with different seismic wave incident angles are obtained through a preset basic impedance equation, specifically including: Based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area, multiple basic impedance curves with different seismic wave incident angles are obtained through the following basic impedance equation: ； in, The angle of incidence of the seismic wave is indicated by The fundamental impedance; , , , It is the ratio of the transverse wave velocity to the longitudinal wave velocity; For longitudinal wave impedance; ρ is the transverse wave impedance; ρ is the density. , and These represent the average P-wave impedance, S-wave impedance, and density of the target layer in the study area, respectively; M is a dynamic normalization parameter. The two base impedance curves in the base impedance curve pair are rotated and then recombined, specifically including: The incident angle of the seismic wave in the base impedance curve is The basic impedance curve and the incident angle of the seismic wave is The basic impedance curve Rotate as follows After recombining the angles, a rotation angle of is generated. base property curves : ； Before selecting the base attribute curve with the highest correlation coefficient to the logging curve of the target reservoir parameters from multiple base attribute curves in each base impedance curve pair as the final attribute curve, the process also includes: For the target reservoir parameter logging curves and multiple basic attribute curves of each pair of basic impedance curves, the sedimentary background is removed in the following manner: ； in, Represents curve The curve after removing the deposition background. Indicates the curve Perform logarithmic operations. Indicates to Perform Chebyshev filtering.
2. The method as described in claim 1, characterized in that, After obtaining multiple basic impedance curves with different seismic wave incident angles based on the P-wave impedance, S-wave impedance, and density well bypass curves of the study area through a preset basic impedance equation, the process also includes: Correct the fundamental impedance curves of multiple seismic waves with different incident angles to the same order of magnitude.
3. The method as described in claim 1 or 2, characterized in that, The incident angle of the seismic wave is determined according to the following method: Multiple seismic wave incident angles are determined by setting the range and step size of the seismic wave incident angle variation.
4. The method as described in claim 1, characterized in that, The rotation angle is determined according to the following method: Multiple rotation angles can be determined by setting the rotation range and rotation step size.
5. The method as described in claim 1, characterized in that, The step of training a specified random forest regression model using the training set specifically includes: Redundancy removal is performed on the final attribute curves in the training set using a specified random forest regression model. The random forest regression model is trained using the deredundant training set.
6. The method as described in claim 5, characterized in that, The process of removing redundancy from the final attribute curves in the training set using a specified random forest regression model specifically includes: The final attribute curves in the training set are pre-trained using a specified random forest regression model to obtain a final attribute curve tree. Redundancy is removed from the final attribute curves with the same node position in the tree, and only one of the final attribute curves is retained.
7. A method for predicting shale gas reservoir parameters, characterized in that, include: Based on the set parameters of the selected final attribute curve, the longitudinal wave impedance, transverse wave impedance and density data volume of the study area are converted into the final attribute data volume. The final attribute data body is input into the regression model trained according to the shale gas reservoir parameter prediction model training method according to any one of claims 1 to 6, and the target reservoir parameter data body of the study area is obtained according to the output result of the regression model.
8. The method as described in claim 7, characterized in that, The setting parameters include the following parameters: The first and second seismic wave incident angles of the two base impedance curves that make up the selected final property curve, and the rotation angle of the selected final property curve.
9. The method as described in claim 8, characterized in that, The process of converting the longitudinal wave impedance, transverse wave impedance, and density data volume of the study area into a final attribute data volume based on the set parameters of the selected final attribute curve specifically includes: The P-wave impedance, S-wave impedance and density data volumes of the study area are used to obtain the first basic impedance data volume at the first seismic wave incident angle and the second basic impedance data volume at the second seismic wave incident angle through the preset basic impedance equation. The first and second base impedance data bodies are rotated and recombined according to the rotation angle to obtain the final attribute data body.
10. The method according to any one of claims 7 to 9, characterized in that, The selected final attribute curve is selected in the following manner: The final attribute curve with the highest correlation coefficient with the logging curve of the target reservoir parameters is selected as the final attribute curve.
11. The method according to any one of claims 7 to 9, characterized in that, The target reservoir parameter is any one of the following parameters: Brittleness, total organic carbon, porosity, total gas content, and effective pressure.
12. A training device for a shale gas reservoir parameter prediction model, characterized in that, The apparatus is used to perform the method of claim 1, the apparatus comprising: The base impedance curve determination module is used to obtain multiple base impedance curves with different seismic wave incident angles based on the P-wave impedance, S-wave impedance and density well bypass curve of the study area through a preset base impedance equation. The basic property curve determination module is used to combine the basic impedance curves in pairs to obtain multiple pairs of basic impedance curves. The two basic impedance curves in the basic impedance curve pairs are rotated and then recombined to obtain multiple basic property curves with different rotation angles. The final attribute curve module is used to select the base attribute curve with the largest correlation coefficient with the logging curve of the target reservoir parameters from multiple base attribute curves of each base impedance curve pair as the final attribute curve; The training module is used to train a specified random forest regression model using a training set consisting of the final attribute curves of each pair of basic impedance curves and the logging curves of the target reservoir parameters, so as to obtain a regression model for predicting the target reservoir parameters in the study area.
13. A shale gas reservoir parameter prediction device, characterized in that, The apparatus is used to perform the method of claim 7, the apparatus comprising: The final attribute data volume determination module is used to convert the longitudinal wave impedance, transverse wave impedance and density data volume of the study area into the final attribute data volume according to the set parameters of the selected final attribute curve. The target reservoir parameter prediction module is used to input the final attribute data body into the regression model trained according to the shale gas reservoir parameter prediction model training method according to any one of claims 1 to 6, and obtain the target reservoir parameter data body of the study area based on the output of the regression model.
14. A computer program product with reservoir parameter prediction function, comprising a computer program / instructions, wherein, When the computer program / instruction is executed by the processor, it implements the shale gas reservoir parameter prediction model training method according to any one of claims 1 to 6, or the shale gas reservoir parameter prediction method according to any one of claims 7 to 11.