Coal reservoir gas-bearing property prediction method and system based on artificial intelligence, and medium

Abstract

The invention relates to a coal measure reservoir gas-bearing prediction method and system based on artificial intelligence and a medium, the coal measure reservoir gas-bearing prediction method based on artificial intelligence comprises the following steps: based on well logging interpretation data, adaptive rock physics modeling is carried out, and calculation results of longitudinal and transverse wave velocity and density are obtained; obtaining angular velocity elastic impedance based on the velocity and density of longitudinal and transverse waves obtained by rock physical modeling, training a deep neural network model for predicting elastic parameters, and performing elastic impedance inversion based on pre-stack seismic data to obtain an elastic impedance data body to realize elastic parameter inversion; based on the elastic parameters of the well logging interpretation data, the longitudinal wave impedance and the actually measured gas content intersection curve graph, determining a gas content prediction basis; and performing gas content prediction based on a gas content analysis basis, and giving a gas content favorable area and a gas content quantitative calculation result. According to the method, the prediction result matched with the actually measured gas value can be given, and the prediction precision of the coal seam reservoir can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of predicting the gas content of coal-bearing reservoirs. In particular, it relates to a method, system, and medium for predicting the gas content of coal-bearing reservoirs based on artificial intelligence. Background Technology

[0002] my country possesses abundant coal-bearing gas resources, accounting for over 60% of the country's total natural gas geological resources, forming a crucial resource foundation for ensuring national energy security. However, coal-bearing reservoirs are characterized by strong heterogeneity and the development of superimposed gas-bearing systems, rendering conventional seismic prediction techniques based on elastic parameter inversion unsuitable and severely hindering the improvement of coal-bearing gas extraction capacity. Therefore, improving the accuracy of gas-bearing prediction in coal-bearing reservoirs is an urgent technical problem to be solved. Summary of the Invention

[0003] This invention provides an artificial intelligence-based method, system, and medium for predicting the gas content of coal-bearing reservoirs, in order to address the limitation on coal-bearing gas extraction capacity caused by the low accuracy of gas content prediction in existing technologies.

[0004] To achieve the above objectives, in a first aspect, the present invention relates to an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs, comprising: adaptive rock physics modeling based on well logging interpretation data to obtain calculation results of P-wave and S-wave velocities and densities, wherein the well logging interpretation data is obtained by interpreting actual well logging data, and the well logging interpretation data includes at least density, porosity, water saturation and well logging mineral data;

[0005] Using the angular elastic impedance obtained from the rock physics model and the corresponding calculated P-wave and S-wave velocity parameters and density parameters, a deep neural network model for predicting elastic parameters is trained. Based on pre-stack seismic data, elastic impedance data volume is obtained by inverting elastic impedance. The trained network model is then applied to the inverted elastic impedance data volume to achieve elastic parameter inversion. The elastic parameters include density and P-wave and S-wave velocity ratio.

[0006] Based on the elastic parameters, longitudinal wave impedance, and measured gas content cross-plot of the actual well logging data, the basis for gas content prediction is determined.

[0007] Based on the aforementioned gas content analysis, gas content is predicted, and favorable gas content areas and quantitative calculation results of gas content are given.

[0008] To achieve the above objectives, in a second aspect, the present invention relates to an artificial intelligence-based coal-bearing reservoir gas-bearing prediction system, comprising: a velocity and density calculation module, used for adaptive rock physics modeling based on well logging interpretation data to obtain calculation results of P-wave and S-wave velocities and densities;

[0009] The inversion module is used to train a deep neural network model for predicting elastic parameters based on the angular elastic impedance obtained by calculating the P-wave and S-wave velocity parameters and density parameters using rock physics modeling. The module performs elastic impedance inversion based on pre-stack seismic data to obtain an elastic impedance data volume. The trained network model is then applied to the inverted elastic impedance data volume to achieve elastic parameter inversion. The elastic parameters include at least density and the P-wave and S-wave velocity ratio.

[0010] The gas content prediction and analysis module is used to determine the basis for gas content prediction based on the elastic parameters, longitudinal wave impedance, and measured gas content cross-plot of the actual well logging data.

[0011] The gas content prediction module is used to predict the gas content based on the gas content analysis, and to provide the favorable gas content area and the quantitative calculation results of the gas content.

[0012] To achieve the above objectives, in a third aspect, the present invention also relates to a computer-readable storage medium storing instructions that, when executed, perform the aforementioned artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs.

[0013] The present invention relates to an artificial intelligence-based method, system, and medium for predicting the gas content of coal-bearing reservoirs, which has the following advantages compared to existing technologies:

[0014] An artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs can provide prediction results that match the measured gas content values, effectively improving the accuracy of coal seam reservoir prediction and helping to overcome the technical bottlenecks of coal-bearing reservoir co-exploration technology. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Example 1.

[0016] Figure 2 This is a calibration diagram of the Xinhucan 1 well, which is an example of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Example 1 of Embodiment 1.

[0017] Figure 3 This is a schematic diagram of the adaptive rock physics modeling process for Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Embodiment 1.

[0018] Figure 4This figure shows the prediction results of well logging data from a rock physics model for Example 1 of an artificial intelligence-based method for predicting gas content in coal-bearing reservoirs, as described in Example 1. In the figure, the "Depth" column represents the formation depth in meters. In the "Combined Trajectory" column, the left column represents the stratigraphic division, with B1-B7 as stratigraphic identifiers corresponding to 7 sets of strata; the right column represents the stratigraphic lithology, with different patterns representing different lithologies. The "Density" column represents density parameters, with the vertical axis representing formation depth in meters and the horizontal axis representing density in kilograms per cubic meter. The "P-wave Velocity" column represents P-wave velocity parameters, with the vertical axis representing formation depth in meters and the horizontal axis representing P-wave velocity in meters per second. The "S-wave Velocity" column represents S-wave velocity parameters, with the vertical axis representing formation depth in meters and the horizontal axis representing S-wave velocity in meters per second.

[0019] Figure 5 This is a flowchart of the pre-stack seismic inversion method of artificial intelligence in Example 1 of the artificial intelligence-based coal-series reservoir gas-bearing prediction method in Example 1.

[0020] Figure 6 This is a comparison of the P-wave impedance results of conventional inversion (left) and intelligent inversion (right) in Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs, as shown in Example 1. The horizontal axis of both the left and right graphs represents the seismic signal observation point number, and the vertical axis represents the double travel time of the seismic wave, in milliseconds.

[0021] Figure 7 The image shows the intelligent inversion results of shear wave impedance (left) and density (right) of Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Example 1. The horizontal axis of both images represents the seismic signal observation point number, and the vertical axis represents the double travel time of the seismic wave, in milliseconds.

[0022] Figure 8 Example 1 of the artificial intelligence-based method for predicting gas content in coal-bearing reservoirs in Embodiment 1 shows a cross-plot (right) of P-wave impedance, P-S / S / V velocity ratio, and gas content obtained from well logging curves (left). The left plot shows the well logging curves, with the vertical axis representing the formation depth. The horizontal axes of the five curves represent P-wave velocity, S / V velocity, density, P-S / S / V velocity ratio, and gas content, respectively, in meters per second, meters per second, grams per cubic meter, dimensionless, and dimensionless. The right plot shows the cross-plot of P-wave impedance, P-S / S / V velocity ratio, and gas content. The horizontal axis represents P-wave impedance in meters per second·gram per cubic meter, and the vertical axis represents the P-S / S / V velocity ratio, which is dimensionless. The shades of color in the plot represent changes in gas content, which is dimensionless.

[0023] Figure 9 This is a schematic diagram of the favorable reservoir area identified in Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs; the horizontal and vertical axes are... Figure 8The same applies to the middle and right figures; the area circled in the elliptical curve in the figure is the favorable area.

[0024] Figure 10 This is a schematic diagram of the density-gas content-P-wave impedance cross-plot obtained from the well logging curves (left) in Example 1 of the artificial intelligence-based coal-bearing reservoir gas content prediction method in Embodiment 1. The left figure shows the well logging curves, with the vertical axis representing the formation depth. The horizontal axes of the five curves correspond to P-wave velocity, S-wave velocity, density, P-wave / S-wave velocity ratio, and gas content, respectively, with units of meters per second, meters per second, grams per cubic meter, dimensionless, and dimensionless. The right figure is a cross-plot of density, gas content, and P-wave impedance, where the horizontal axis represents density in grams per cubic centimeter, and the vertical axis represents gas content, dimensionless. The color intensity represents the P-wave impedance in meters per second·gram per cubic meter.

[0025] Figure 11 This is a schematic diagram of the density-gas content relationship curve of Example 1 of the artificial intelligence-based coal-series reservoir gas content prediction method in Example 1; the horizontal axis of both the left and right graphs is density in grams per cubic meter, and the vertical axis is gas content, which is dimensionless; the box in the left graph shows the gas content when the longitudinal wave impedance is less than 4800, and the box in the right graph shows the gas content when the impedance is greater than or equal to 4800.

[0026] Figure 12 This is a schematic diagram of the distribution range of favorable areas on the cross section of Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Example 1; the horizontal axis represents the seismic signal observation point number, and the coordinates represent the double travel time of seismic waves in milliseconds.

[0027] Figure 13 This is a schematic diagram of the gas content prediction results (through-well line) on the profile of Example 1 of an artificial intelligence-based coal-bearing reservoir gas content prediction method in Example 1; the horizontal axis represents the seismic signal observation point number, and the coordinate axis represents the double travel time of the seismic wave in milliseconds.

[0028] Figure 14 This is a schematic diagram of the gas content prediction results (through-well line) on the profile of Example 1 of an artificial intelligence-based coal-bearing reservoir gas content prediction method in Example 1; the horizontal axis represents the seismic signal observation point number, and the coordinate axis represents the double travel time of the seismic wave in milliseconds.

[0029] Figure 15 This is a schematic diagram comparing the measured gas content at the well point (left) with the predicted value of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs in Example 1 of Embodiment 1. The vertical axis of the left figure represents the burial depth, and the horizontal axis represents the gas content. The horizontal axis of the right figure represents the measured gas content, and the predicted gas content is shown in the right figure. All values ​​are dimensionless.

[0030] Figure 16The four sets of comprehensive planar maps showing the gas content distribution of coal-bearing strata in Example 1 of an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs are shown (top left: T6; top right: T4; bottom left: T2; bottom right: T1). The horizontal and vertical axes of the four maps are the x and y coordinates of geodetic space, and the unit is meters.

[0031] Figure 17 This is a schematic diagram of the structure of an artificial intelligence-based coal-bearing reservoir gas-bearing prediction system according to Embodiment 2 of the present invention. Detailed Implementation

[0032] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention and not the entire structure.

[0033] Example 1

[0034] Please refer to the following: An artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs. Figure 1-16 As shown, the present invention provides a method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence, comprising the following steps: S101 to S104.

[0035] S101 uses adaptive rock physics modeling based on well logging interpretation data to obtain the calculation results of P-wave and S-wave velocities and densities. The well logging interpretation data is obtained by interpreting actual well logging data and includes at least density, porosity, water saturation, and well logging mineral data.

[0036] Based on the P-wave and S-wave velocities and densities obtained from rock physics modeling, S102 obtains angular velocity elastic impedance. Using angular elastic impedance, P-wave and S-wave velocities, and density, a deep neural network model for predicting elastic parameters is trained. Based on pre-stack seismic data, elastic impedance inversion is performed to obtain elastic impedance data volume. The trained network model is applied to the inverted elastic impedance data volume to realize elastic parameter inversion. Elastic parameters include density and P-wave / S-wave velocity ratio.

[0037] S103 determines the basis for gas-bearing prediction based on the elastic parameters, P-wave impedance, and measured gas content cross-plot of actual well logging data.

[0038] Based on the gas content analysis, S104 predicts the gas content and provides the favorable gas content areas and quantitative calculation results of the gas content.

[0039] In this embodiment S101, as Figure 2 Adaptive rock physics modeling, specifically including: S111-S117.

[0040] S111 performs regularization processing on the acquired well logging interpretation mineral data, expanding the well logging mineral data at each depth point to include all minerals in the work area.

[0041] Among them, the value of minerals that do not exist at the corresponding depth point is set to 0, and they are processed into standard logging mineral data that includes all mineral types in the work area.

[0042] S112 Establishing the preferred rock physics model: For each logging depth point in the well logging interpretation data, calculate the priority parameters of various rock physics models in the set of candidate rock physics models.

[0043] In this embodiment, specifically:

[0044] Based on the regularized well logging mineral data, priority parameters of various rock physics models are calculated for each well logging depth point, and the priority of the rock physics models is sorted according to the priority parameters. The calculation method of priority parameters of rock physics models at any depth point is as follows: Suppose that the rock physics model has m variables that require input, the penalty value βj for matching value of the j-th variable is 2, and the penalty value βj for non-matching value is 0.5, where Ad is the priority parameter.

[0045] S113: The reservoir type selection model is based on a preset priority calculation criterion. In the set of candidate rock physical models, the main component analysis of minerals at each logging depth is performed to obtain the priority parameter of each rock physical model. The priority parameter calculation criterion is as follows: if the time logging data does not provide the required data, the rock physical model corresponding to the set of rock physical models is removed from the set, and the remaining ones are the candidate rock physical models. Each rock physical model first compares with the required input data. If the required input data does not exist in the logging data, the priority parameter Ad value is 0.

[0046] S114 calculates the P-wave and S-wave velocity parameters based on the rock physics model corresponding to the highest priority parameters.

[0047] In this embodiment, specifically: based on the default mineral modulus of the highest priority rock physics model, the rock matrix modulus is calculated, and then according to the default pore type and pore aspect ratio, the P-wave and S-wave velocity parameters and density parameters under the logging fluid state are obtained through the Gassmann formula.

[0048] S115 inputs the reservoir type into the preset rock physics optimization model, and obtains the rock physics model based on the formation properties of the reservoir type.

[0049] S116 uses P-wave and S-wave velocities and densities as objective functions to determine the calculation errors of reservoir type optimization models and rock physics optimization models relative to well logging interpretation data.

[0050] S117 sets an error threshold. If the calculation error is greater than the error threshold, the reservoir type optimization model and the rock physics optimization model are reselected. If the calculation error is less than the error threshold, the adaptive rock physics modeling is completed.

[0051] In this embodiment S102, as Figure 5 As shown, it includes: S121-S123.

[0052] S121 extracts three incident angle gathers from pre-stack seismic data and inverts elastic impedance data volumes at three angles.

[0053] S122 calculates elastic impedance data at three angles based on the P-wave and S-wave velocities and densities obtained from rock physics modeling. It then uses the elastic impedance at the three angles and the P-wave and S-wave velocities and densities to train a deep neural network model for predicting elastic parameters.

[0054] S123 uses the trained deep neural network model to invert the elastic impedance data volume to obtain the P-wave velocity ratio and density.

[0055] In this embodiment, S103 includes: S131-S132;

[0056] S131 identifies favorable reservoir areas based on the P-wave impedance characteristics and P-wave / S-wave velocity ratio characteristics of favorable reservoir areas on the P-wave impedance-P-wave / S-wave velocity ratio-gas content cross-plot.

[0057] Based on the negative correlation between density and gas content curves from actual well logging, S132 divides the region into two areas with a P-wave impedance value of 4800 as the threshold, and fits density-gas content relationship curves to each region as a quantitative prediction method for gas content. Where GC is the gas content, ρ is the density, and I ρ This is the longitudinal wave impedance.

[0058] In this embodiment, S104 includes: S141-S142:

[0059] S141 uses the gas content value corresponding to the favorable reservoir area given in the cross plot of P-wave impedance-P-S-wave velocity ratio-gas content to separate the corresponding favorable gas content area in the inversion data, and obtain the corresponding favorable reservoir area of ​​the target reservoir in the entire working area.

[0060] S142 calculates the gas content of the target reservoir within its favorable zone using the density-gas content relationship curve and inversion data.

[0061] To better illustrate the solution of the present invention, an example is given below, such as... Figure 2-16 As shown, it includes the following steps:

[0062] This case study primarily utilizes well logging data from the Xishanyao Formation of the Xinhucan 1 well in the Zhunnan Coalfield of Xinjiang, 3D pre-stack seismic time migration data, and stratigraphic interpretation data from four sets of strata (T6, T4, T2, and T1) to investigate the seismic attributes and prediction methods for sweet spots in coal-bearing reservoirs. The study mainly comprises four steps: adaptive rock physics modeling, artificial intelligence seismic inversion, gas-bearing prediction analysis, and reservoir gas-bearing prediction.

[0063] Figure 2 This is a comprehensive calibration map of the Xinhucan 1 well located within the work area. The map shows that the Xishanyao Formation in this area mainly consists of seven coal seams, B7-B1, etc. The synthetic seismic records produce four sets of strong reflection interfaces, which also correspond to four sets of strong reflection interfaces on the seismic profile: the B7-B5 coal seams are relatively close and synthesize into the same reflection interface T6 on the seismic profile; the B4-B3 coal seams are also relatively close and synthesize into the same reflection interface on the seismic profile, without forming a separate reflection interface T4; the B2 coal seam corresponds to the reflection interface T2 on the seismic profile; the two B1 coal seams are too close and share a single phase axis on the seismic profile, i.e., T1. Therefore, the seismic interpretation data for this project mainly consists of four strata: T6, T4, T2, and T1.

[0064] Adaptive rock physics modeling: Figure 3 This is a schematic diagram of the adaptive rock physics modeling process. First, based on well logging interpretation data and using mineral principal component analysis, the reservoir type is automatically selected. Second, based on the determined reservoir type and its formation attributes, a corresponding rock physics model is automatically selected. Third, using P-wave velocity and density as objective functions, the calculation errors of the first two automatically selected models are given. Finally, a threshold is given; if the error exceeds the threshold, a new model is selected; otherwise, the calculation ends, completing the modeling process. The final calculation results are as follows: Figure 4 As shown in the figure, the lithology calculation results and well logging interpretation results have a high degree of agreement, and the density and P-wave velocity calculation results have a high degree of agreement with the well logging curves. Therefore, based on the established model, high-precision prediction of well logging curves can be achieved, and the predicted curve of shear wave velocity can be given in the end.

[0065] Artificial Intelligence Seismic Inversion: Based on the elastic impedance inversion method described above, elastic impedance data corresponding to three incident angles (3°, 8°, 13°) can be extracted from pre-stack seismic data. A neural network model is trained based on known well information and seismic data to invert and obtain the P-wave and S-wave impedance and density. Figure 6 and Figure 7 The longitudinal wave impedance inversion results obtained by conventional methods and artificial intelligence methods are compared (e.g.) Figure 6 As shown in the figure, the results indicate that conventional inversion results have low resolution and cannot accurately identify the differences in elastic parameters within the coal seam. Intelligent inversion results have better resolution and are conducive to conducting more detailed research on the characteristics of coal-bearing strata reservoirs.

[0066] Gas content prediction analysis: Figure 8 A cross-plot of P-wave impedance, P-S / ... Figure 9 Therefore, favorable reservoir regions can be effectively identified on the P-wave impedance-P-S / S velocity ratio-gas content cross-plot, such as... Figure 10 As shown.

[0067] ② Analysis of density-gas content-longitudinal wave impedance cross plot

[0068] The density curve and the gas content curve show a clear negative correlation. Figure 11 The negative correlation trend between density and gas content is basically the same in different coal seams (not shown in the attached figure). Using a longitudinal wave impedance value of 4800 as a threshold, two regions were divided, and density-gas content relationships were fitted to each region separately, serving as the basis for subsequent quantitative prediction of gas content. Figure 12 ).

[0069] Where GC is the gas content, ρ is the density, and I ρ This is the longitudinal wave impedance.

[0070] Reservoir gas content prediction: First, using the gas content values ​​corresponding to the favorable gas content areas given in the P-wave impedance-P-S-wave velocity ratio-gas content cross-plot, the corresponding favorable areas in the P-wave impedance and density inversion results are separated, thus obtaining the distribution range of favorable areas of the corresponding target layer within the entire work area. Figure 12 ).

[0071] Then, within the favorable reservoir area, the gas content of the reservoir is calculated using the density-gas content relationship curve, based on the P-wave and S-wave impedance and density inversion data. The inversion results via the well line are as follows: Figure 13 and 14 As shown. Through comparison of data from well logging locations ( Figure 15 As can be seen, the predicted value and the measured value have a high degree of agreement. Figure 17 The study presents a comprehensive plan view of gas content distribution in four coal-bearing strata. All well points are located in areas with high gas content, indicating that the quantitative prediction method for gas content has high accuracy.

[0072] In summary, this invention relates to an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs, comprising four main steps: adaptive rock physics modeling to obtain high-precision calculation results for P-wave and S-wave velocities and densities; artificial intelligence seismic inversion to obtain high-resolution P-wave and S-wave impedance and density inversion profiles; gas content prediction analysis to provide the basis for gas content prediction; and gas content prediction to provide favorable gas-bearing areas and quantitative calculation results for gas content. Finally, comparison with well logging data in the examples shows that this method can provide prediction results that match the measured gas content values, effectively improving the accuracy of coal seam reservoir prediction and helping to overcome the technical bottlenecks of coal-bearing reservoir co-exploration technology.

[0073] Example 2

[0074] An artificial intelligence-based coal-bearing reservoir gas-bearing prediction system is provided for predicting the gas content of coal-bearing reservoirs. It is implemented using electronic hardware with a central processing unit, such as a personal computer, smart terminal, local area network, or server. For implementation details in this example, please refer to [link to relevant documentation]. Figure 17 It includes a velocity and density calculation module 61, an inversion module 62, a gas content prediction and analysis module 63, and a gas content prediction module 64.

[0075] The velocity and density calculation module 61 is used for adaptive rock physics modeling based on well logging interpretation data to obtain the calculation results of P-wave and S-wave velocity and density;

[0076] Inversion module 62 is used to obtain the P-wave and S-wave velocities and densities obtained from rock physics modeling, to obtain the angular elastic impedance, and to train a deep neural network model to predict elastic parameters using the angular elastic impedance, P-wave and S-wave velocities and densities. Based on pre-stack seismic data, elastic impedance is inverted to obtain the elastic impedance data volume. The trained network model is applied to the inverted elastic impedance data volume to realize the elastic parameter inversion. The elastic parameters include at least the density and the P-wave and S-wave velocity ratio.

[0077] The gas content prediction and analysis module 63 is used to determine the basis for gas content prediction based on the elastic parameters, P-wave impedance and measured gas content cross-plot of well logging interpretation data.

[0078] The gas content prediction module 64 is used to predict the gas content based on the gas content analysis, and to provide the favorable gas content area and the quantitative calculation results of the gas content.

[0079] In this embodiment, the velocity and density calculation module 61 is specifically used for:

[0080] The acquired well logging interpretation mineral data is processed into a regularized form, expanding the well logging mineral data at each depth point to include all minerals in the work area. Minerals that do not exist at the corresponding depth point are set to 0, thus processing the data into standard well logging mineral data that encompasses all mineral types in the work area.

[0081] Establish the preferred rock physics model: For each logging depth point of the standard logging interpretation data, calculate the priority parameters of various rock physics models in the set of candidate rock physics models;

[0082] The reservoir type selection model is based on a preset priority calculation criterion. In the set of candidate rock physics models, the minerals at each logging depth point are analyzed for major components to obtain the priority parameters of each rock physics model. The reservoir type is obtained based on the rock physics model corresponding to the highest priority parameter, and the P-wave and S-wave velocities are calculated.

[0083] Input the reservoir type into the preset rock physics optimization model, and obtain the rock physics model based on the formation properties of the reservoir type;

[0084] Using P-wave and S-wave velocity and density as objective functions, the calculation errors of the reservoir type optimization model and the rock physics optimization model relative to the well logging interpretation data are determined;

[0085] Set an error threshold. If the calculation error is greater than the error threshold, then reselect based on the reservoir type optimization model and the rock physics optimization model. If the calculation error is less than the error threshold, then complete the adaptive rock physics modeling.

[0086] The calculation results of P-wave and S-wave velocities and densities were obtained based on adaptive rock physics modeling.

[0087] In this embodiment, the inversion module 62 is specifically used for:

[0088] Three incident angle gathers are extracted from pre-stack seismic data, and elastic impedance data volumes at three angles are inverted.

[0089] Based on the P-wave and S-wave velocities and densities obtained from rock physics modeling, elastic impedance data at three angles are calculated. The elastic impedance at the three angles and the P-wave and S-wave velocities and densities are used to train a deep neural network model to predict elastic parameters.

[0090] The trained deep neural network model is used to invert the P-wave velocity ratio and density of the elastic impedance data volume.

[0091] In this embodiment, the gas content prediction and analysis module 63 is specifically used to: identify the favorable reservoir area on the cross-plot of P-wave impedance-P-S / S / S / S / G content based on the P-wave impedance characteristics and P-wave / S / S / S / G content characteristics of the favorable reservoir area.

[0092] Based on the negative correlation between density and gas content curves from actual well logging, two regions are divided with a P-wave impedance value of 4800 as the threshold. Density-gas content relationship curves are fitted to each region separately, serving as a quantitative prediction method for gas content. Where GC is the gas content, ρ is the density, and Iρ is the longitudinal wave impedance.

[0093] In this embodiment, the gas content prediction module 64 is specifically used for:

[0094] Using the gas content value corresponding to the favorable reservoir area given in the cross plot of P-wave impedance-P-S-wave velocity ratio-gas content, the corresponding favorable gas content area in the inversion data is separated to obtain the corresponding favorable reservoir area of ​​the target reservoir in the entire work area.

[0095] Within the favorable area of ​​the target reservoir, the gas content of the reservoir is calculated based on the inversion data using the density-gas content relationship curve.

[0096] The artificial intelligence-based coal-bearing gas content prediction system of this embodiment is the same as the artificial intelligence-based coal-bearing gas content prediction method described in Embodiment 1 in terms of implementation process, method and effect, and will not be repeated here.

[0097] Example 3

[0098] This invention relates to a computer-readable storage medium storing instructions that, when executed, perform an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs according to Embodiment 1. The execution process and effects are the same as those described in Embodiment 1, and will not be repeated here.

[0099] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0100] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence, characterized in that, include: Adaptive rock physics modeling based on well logging interpretation data yields the calculation results of P-wave and S-wave velocities and densities. The well logging interpretation data is obtained by interpreting actual well logging data. Based on the P-wave and S-wave velocities and densities obtained from the rock physics modeling, angular velocity elastic impedance is obtained. A deep neural network model for predicting elastic parameters is trained using the angular elastic impedance, the P-wave and S-wave velocities, and the density. Elastic impedance data volume is obtained by elastic impedance inversion based on pre-stack seismic data. The trained network model is applied to the inverted elastic impedance data volume to realize elastic parameter inversion. The elastic parameters include density and the P-wave and S-wave velocity ratio. Based on the elastic parameters, longitudinal wave impedance, and measured gas content cross-plot of the actual well logging data, the basis for gas content prediction is determined. Based on the aforementioned gas content analysis, gas content is predicted, and favorable gas content areas and quantitative calculation results of gas content are given.

2. The method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence according to claim 1, characterized in that, The adaptive rock physics modeling specifically includes: The acquired well logging interpretation mineral data is processed into a regularization process, and the well logging mineral data at each depth point is expanded to include all minerals in the work area. The value of minerals that do not exist at the corresponding depth point is set to 0, and the data is processed into standard well logging mineral data that includes all mineral types in the work area. Establish the preferred rock physics model: For each logging depth point in the well logging interpretation data, calculate the priority parameters of various rock physics models in the set of candidate rock physics models; The reservoir type selection model is based on a preset priority calculation criterion. In the set of candidate rock physics models, the minerals at each logging depth point are analyzed for major components to obtain the priority parameters of each rock physics model. The P-wave and S-wave velocity parameters are calculated based on the rock physics model corresponding to the highest priority parameter. Input the reservoir type into the preset rock physics optimization model, and obtain the rock physics model based on the formation properties of the reservoir type; Using P-wave and S-wave velocity and density as objective functions, the calculation errors of the reservoir type optimization model and the rock physics optimization model relative to the well logging interpretation data are determined; If an error threshold is set, and the calculation error is greater than the error threshold, a new selection is made based on the reservoir type optimization model and the rock physics optimization model. If the calculation error is less than the error threshold, the adaptive rock physics modeling is completed.

3. The method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence according to claim 1, characterized in that, The angular velocity elastic impedance is obtained based on the P-wave and S-wave velocities and densities obtained from the rock physics modeling. A deep neural network model for predicting elastic parameters is trained using the angular elastic impedance, the P-wave and S-wave velocities, and the density. Elastic impedance data is obtained by inverting elastic impedance based on pre-stack seismic data. The trained network model is then applied to the inverted elastic impedance data to achieve elastic parameter inversion, including: Three incident angle gathers were extracted from the pre-stack seismic data, and elastic impedance data volumes at the three angles were inverted. Based on the P-wave and S-wave velocities and densities obtained from the rock physics modeling, elastic impedance data at three angles are calculated. The elastic impedance at the three angles and the P-wave and S-wave velocities and densities are then used to train a deep neural network model to predict elastic parameters. The trained deep neural network model is used in the elastic impedance data volume to invert the P-wave velocity ratio and density.

4. The method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence according to claim 1, characterized in that, The cross-plot of elastic parameters, longitudinal wave impedance, and measured gas content based on the actual well logging data is used to determine the basis for gas-bearing prediction, including: Based on the P-wave impedance characteristics and P-wave / S-wave velocity ratio characteristics of the reservoir favorable area, the reservoir favorable area is identified on the P-wave impedance-P-wave / S-wave velocity ratio-gas content cross plot. Based on the negative correlation between the density curve and the gas-bearing curve of the well logging interpretation data, two regions are divided with a P-wave impedance value of 4800 as the threshold. Density-gas-bearing curves are fitted to each region separately, serving as a quantitative prediction method for gas-bearing capacity. Where GC is the gas content, ρ is the density, and I ρ This is the longitudinal wave impedance.

5. The method for predicting the gas content of coal-bearing reservoirs based on artificial intelligence according to claim 1, characterized in that, Based on the gas content analysis, the gas content is predicted, and favorable gas content areas and quantitative calculation results of gas content are given, including: Using the gas content value corresponding to the favorable area of ​​the reservoir given in the cross plot of longitudinal wave impedance-longitudinal wave velocity ratio-gas content, the corresponding favorable gas content area in the inversion data is separated to obtain the corresponding favorable area of ​​the target reservoir in the entire work area. Within the favorable area of ​​the target reservoir, the gas content of the reservoir is calculated based on the inversion data using the density-gas content relationship curve.

6. An artificial intelligence-based system for predicting the gas content of coal-bearing reservoirs, characterized in that, include: The velocity and density calculation module is used for adaptive rock physics modeling based on well logging interpretation data to obtain the calculation results of P-wave and S-wave velocity and density; The inversion module is used to obtain the P-wave and S-wave velocities and densities obtained from the rock physics modeling to obtain the angular elastic impedance. It uses the angular elastic impedance, the P-wave and S-wave velocities, and the density to train a deep neural network model to predict elastic parameters. Based on pre-stack seismic data, it performs elastic impedance inversion to obtain an elastic impedance data volume. The trained network model is applied to the inverted elastic impedance data volume to realize the elastic parameter inversion. The elastic parameters include at least density and the P-wave and S-wave velocity ratio. The gas content prediction and analysis module is used to determine the basis for gas content prediction based on the elastic parameters, longitudinal wave impedance, and measured gas content cross-plot of the actual well logging data. The gas content prediction module is used to predict the gas content based on the gas content analysis, and to provide the favorable gas content area and the quantitative calculation results of the gas content.

7. The artificial intelligence-based coal-bearing reservoir gas-bearing prediction system according to claim 6, characterized in that, The velocity and density calculation module is specifically used for: The acquired well logging interpretation mineral data is processed into a regularization process, and the well logging mineral data at each depth point is expanded to include all minerals in the work area. The value of minerals that do not exist at the corresponding depth point is set to 0, and the data is processed into standard well logging mineral data that includes all mineral types in the work area. Establish a preferred rock physics model: For each logging depth point of the standard logging interpretation data, calculate the priority parameters of various rock physics models in the set of candidate rock physics models; The reservoir type selection model is based on a preset priority calculation criterion. In the set of candidate rock physics models, the minerals at each logging depth point are analyzed for major components to obtain the priority parameters of each rock physics model. The reservoir type is obtained based on the rock physics model corresponding to the highest priority parameter, and the P-wave and S-wave velocities are calculated. Input the reservoir type into the preset rock physics optimization model, and obtain the rock physics model based on the formation properties of the reservoir type; Using P-wave and S-wave velocity and density as objective functions, the calculation errors of the reservoir type optimization model and the rock physics optimization model relative to the well logging interpretation data are determined; Set an error threshold. If the calculation error is greater than the error threshold, then reselect based on the reservoir type optimization model and the rock physics optimization model. If the calculation error is less than the error threshold, then complete the adaptive rock physics modeling. The calculation results of P-wave and S-wave velocities and densities are obtained based on the adaptive rock physics model.

8. The artificial intelligence-based gas-bearing prediction system for coal-bearing reservoirs according to claim 6, characterized in that, The inversion module is specifically used for: Three incident angle gathers are extracted from pre-stack seismic data, and elastic impedance data volumes at three angles are inverted. Based on the P-wave and S-wave velocities and densities obtained from the rock physics modeling, elastic impedance data at three angles are calculated. The elastic impedance at the three angles and the P-wave and S-wave velocities and densities are then used to train a deep neural network model to predict elastic parameters. The trained deep neural network model is used in the elastic impedance data volume to invert the P-wave velocity ratio and density.

9. The artificial intelligence-based gas-bearing prediction system for coal-bearing reservoirs according to claim 6, characterized in that, The gas content prediction and analysis module is specifically used to: identify favorable reservoir areas on the cross-plot of P-wave impedance, P-wave velocity ratio, and gas content based on the P-wave impedance characteristics and P-wave velocity ratio characteristics of favorable reservoir areas. Based on the negative correlation between density and gas content curves from actual well logging, two regions are divided with a P-wave impedance value of 4800 as the threshold. Density-gas content relationship curves are fitted to each region separately, serving as a quantitative prediction method for gas content. Where GC is the gas content, ρ is the density, and I ρ For longitudinal wave impedance; The gas content prediction module is specifically used for: Using the gas content value corresponding to the favorable area of ​​the reservoir given in the cross plot of longitudinal wave impedance-longitudinal wave velocity ratio-gas content, the corresponding favorable gas content area in the inversion data is separated to obtain the corresponding favorable area of ​​the target reservoir in the entire work area. Within the favorable area of ​​the target reservoir, the gas content of the reservoir is calculated based on the inversion data using the density-gas content relationship curve.

10. A computer-readable storage medium, characterized in that: The storage medium stores instructions that, when executed, perform an artificial intelligence-based method for predicting the gas content of coal-bearing reservoirs as described in any one of claims 1-5.

Images