A method and system for spatially inverse predicting key in-situ rock mechanics parameters of a reservoir

Abstract

The present invention provides a method and system for spatially inverse predicting key rock mechanical parameters in-situ in a reservoir, belonging to the field of shale oil and gas geology, and specifically including the following steps: conducting nano-indentation experiments to obtain an indentation experiment database, simulating rock indentation experiments under different conditions to obtain a macroscopic indentation numerical model; using the Latin hypercube sampling method to obtain a database of simulated parameter combinations, conducting batch numerical experiments to construct an inversion dataset; training a Bayesian-optimized deep neural network model based on the inversion dataset to obtain a single-point inversion prediction model; further establishing a spatial inversion prediction model according to the Kriging interpolation algorithm; inputting real-time detection data of deep rock formations into the spatial inversion prediction model to obtain spatial distribution data of real-time underground rock mechanical parameters of deep rock masses. It can relatively accurately inverse the key rock mechanical parameters in the deep reservoir space, providing a basis for well placement, well trajectory design, and optimization of fracturing schemes, etc.

Description

Technical Field

[0001] The present invention belongs to the field of shale oil and gas geology, and particularly relates to a method and system for spatially inverting and predicting key in-situ rock mechanical parameters of a reservoir. Background Art

[0002] In recent years, the focus of oil and gas exploration and development in China has gradually shifted to the deep part. The accurate inversion of key geological parameters such as the elastic modulus, hardness, and fracture toughness of deep rock masses is crucial for the evaluation of sweet spots, the design of wellbore trajectories, and the optimization of fracturing programs. Currently, macro rock mechanics experiments are generally used to measure rock mechanical parameters. However, this method requires relatively large rock samples, which are not easy to obtain, and the amount of data obtained is relatively limited, unable to obtain sufficient data support. In recent years, nano / micro-indentation experiments have been favored in the measurement of rock mechanical parameters. On the one hand, this method has lower requirements for sample specimens and sizes. On the other hand, continuous mechanical data can be obtained by combining well logging data, providing a sufficient data basis for the effective development of reservoirs. However, current relevant research results mainly focus on two aspects: one is the establishment of mechanical parameter models for different lithology rock samples. For example, the patent application document with the title of "A method for obtaining macro rock mechanical parameters based on nano-indentation technology" and the publication number of CN 119023472 A, which was published on November 6, 2024, records a method for obtaining macro rock mechanical parameters based on nano-indentation technology. This method is based on the macro hardness of rock obtained by nano-indentation, and successively establishes relationship models for compressive strength, Young's modulus, and Poisson's ratio to obtain key macro rock mechanical parameters. The other is the upgrade of mechanical parameters from the nano scale to the macro scale. For example, the paper "Study on the mechanical properties of laminated shale using nano-indentation experiments" published by Shi Xian et al. in 2019 mainly measured the basic rock mechanical parameters of shale using nano-indentation technology, proposed a three-component rock micro-mechanical model based on mineral strength classification, and used the Mori-Tanaka method to achieve the upgrade of mechanical parameters from the nano scale to the centimeter scale.

[0003] Therefore, there is less research on the applicability of nano-indentation experiments under extreme conditions of deep formations and mechanical parameter models at present, and it is impossible to effectively obtain the mechanical parameters of deep reservoirs, which restricts the effective development of deep reservoir energy. Summary of the Invention

[0004] In order to solve the problem of spatially inverting the key in-situ rock mechanical parameters of a reservoir under extreme conditions, the present invention provides a method and system for spatially inverting and predicting key in-situ rock mechanical parameters of a reservoir.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for spatially inverting and predicting key in-situ rock mechanical parameters of a reservoir includes the following steps:

[0007] Obtain the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters to construct a data set, and construct a macroscopic indentation numerical model according to the data set;

[0008] Sample the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method to form parameter combinations; input the parameter combinations into the macroscopic indentation numerical model for simulation analysis to obtain load-displacement curves under different conditions; extract multiple characteristic values from the load-displacement curves under different conditions as the training set; wherein, the characteristic values characterize the rock mechanics response parameters;

[0009] Optimize the hyperparameters of the deep neural network based on the Bayesian optimization algorithm, input the training set data into the optimized deep neural network to train the model, and obtain a single-point inversion prediction model; based on the Kriging interpolation algorithm, convert the single-point inversion prediction model into a spatial inversion prediction model;

[0010] Input the load-displacement data detected in real time of the rock stratum into the spatial inversion prediction model to obtain the spatial distribution data of the rock mechanics parameters in the deep rock stratum.

[0011] Preferably, the sampling of the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method is specifically through the following formula:

[0012] ;

[0013] Wherein, represents the estimated value; represents the expectation of; represents the number of samplings participating in the computer operation; represents the linear coefficient; represents the standard deviation of; represents the standard deviation of; represents the sampling result value, and the sampling result is any one of the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample; represents the index subscript.

[0014] Preferably, the optimization of the hyperparameters of the deep neural network based on the Bayesian optimization algorithm is specifically:

[0015] Based on the Bayesian optimization algorithm, with the mean absolute error as the objective function, optimize the hyperparameters of the deep neural network model to screen out the optimal deep neural network structure parameter combination.

[0016] Preferably, obtaining the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters specifically involves conducting a mineral component experiment, a scanning electron microscope experiment, and an indentation experiment on the target rock sample to obtain the mineral composition of the target rock sample and the microscopic information of the sample rock after the indentation experiment; meanwhile, conducting a triaxial stress experiment to obtain the DP criterion parameters.

[0017] Preferably, the multiple characteristic values specifically include: temperature, pressure, maximum applied load, indentation depth corresponding to the maximum applied load, and indentation depth after complete unloading.

[0018] The present invention also provides a system for in-situ key rock mechanics parameter spatial inversion prediction of a reservoir, specifically including:

[0019] A data processing module, configured to obtain the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters to construct a data set, and construct a macroscopic indentation numerical model according to the data set.

[0020] A parameter calculation module, configured to sample the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method to form parameter combinations; input the parameter combinations into the macroscopic indentation numerical model for simulation analysis to obtain load-displacement curves under different conditions; extract multiple characteristic values from the load-displacement curves under different conditions as a training set; wherein, the characteristic values characterize the rock mechanics response parameters.

[0021] A training module, configured to optimize the hyperparameters of the deep neural network based on the Bayesian optimization algorithm, input the training set data into the optimized deep neural network to train the model, and obtain a single-point inversion prediction model; based on the Kriging interpolation algorithm, convert the single-point inversion prediction model into a spatial inversion prediction model.

[0022] A parameter inversion module, configured to input the load-displacement data detected in real time in the rock formation into the spatial inversion prediction model to obtain the spatial distribution data of the rock mechanics parameters in the deep rock formation.

[0023] The present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps in the method for in-situ key rock mechanics parameter spatial inversion prediction of a reservoir.

[0024] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is loaded by a processor, it can execute the steps in the method for in-situ key rock mechanics parameter spatial inversion prediction of a reservoir.

[0025] The method and system for spatially inverse predicting key rock mechanical parameters in-situ in a reservoir provided by the present invention have the following beneficial effects:

[0026] The present invention establishes a macroscopic indentation numerical model through the experimental parameter information of a target rock sample, simulates and analyzes the combined parameters of the target rock sample through the macroscopic indentation numerical model, constructs a training set, and obtains the mechanical parameters of the target rock sample that are relatively accurate under extreme conditions of deep formations, providing a data basis for subsequent model training. The deep neural network model is trained with the training set data and combined with the Kriging interpolation algorithm to obtain a spatially inverse prediction model. The real-time detection data of deep rock formations are input into the spatially inverse prediction model to obtain the spatial distribution data of rock mechanical parameters in deep rock formations. It can relatively accurately inverse the key rock mechanical parameters in the deep reservoir space, providing a basis for well placement, well trajectory design, and fracturing scheme optimization of deep reservoir energy, etc. Description of the Drawings

[0027] To more clearly illustrate the embodiments of the present invention and their design schemes, the accompanying drawings required for this embodiment will be briefly introduced below. The accompanying drawings in the following description are only partial embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0028] Figure 1 It is a flowchart of a method for spatially inverse predicting key rock mechanical parameters in-situ in a reservoir according to an embodiment of the present invention.

[0029] Figure 2 It is a schematic diagram of the DP criterion and the principle of rock damage discrimination in an embodiment of the present invention, where Figure 2 (a) is the p-t relationship curve, Figure 2 (b) is the D-P yield surface, Figure 2 and (c) is the stress-strain relationship curve during the rock damage process.

[0030] Figure 3 It is the nano-indentation numerical model in an embodiment of the present invention.

[0031] Figure 4 It is the macroscopic indentation numerical model in an embodiment of the present invention.

[0032] Figure 5 It is a comparison diagram of the predicted elastic modulus and the true elastic modulus in an embodiment of the present invention.

[0033] Figure 6 It is the spatial distribution characteristics of the elastic modulus in an embodiment of the present invention.

[0034] Figure 7 It is a flowchart of a method for spatially inverse predicting key rock mechanical parameters in-situ in a reservoir according to the present invention. Detailed Implementation Modes

[0035] To enable those skilled in the art to better understand the technical solutions of the present invention and be able to implement them, the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0036] Embodiment

[0037] In Area A, the generally buried depth of the favorable productive formation is greater than 3500 m, and the high-temperature and high-pressure geological environment has a huge impact on the mechanical properties of reservoir rocks. It is crucial to clarify the distribution characteristics of mechanical parameters in the in-situ environment of the reservoir for the efficient development of the reservoir.

[0038] The present invention provides a method for spatially inverse predicting key rock mechanical parameters in the reservoir in-situ. Taking Area A as an example, as Figure 1 shown, it specifically includes the following steps:

[0039] Step 1: Select typical deep wells in Area A, continuously collect cuttings samples of the target layer in the horizontal section at intervals of 10 m (at least 3 samples are collected at each sampling point), prepare them into square specimens of 15 mm×15 mm×8 mm, then polish the test surface with sandpapers of different mesh numbers in turn to reduce the surface roughness, and finally use argon ion polishing to further reduce the surface roughness to complete the preparation of the specimens. According to the actual formation temperature conditions of the target layer, set high-temperature conditions of 100 - 200 °C, carry out nanoindentation experiments under high temperature, obtain the load-displacement relationship curves at different temperatures and depths, and establish an indentation experiment database.

[0040] Step 2: For the samples in Step 1, carry out mineral composition experiments and scanning electron microscope experiments, obtain the mineral composition of the target rock samples and conduct microtopography analysis of the cross-section after the indentation experiment, and carry out triaxial stress experiments to obtain the key parameters required by the DP criterion. Based on the microscopic information of rock minerals and the relevant parameters of the DP criterion, use Abaqus finite element numerical simulation software to establish a nanoindentation numerical model, as Figure 3 shown. The nanoindentation numerical model includes two parts: the indenter and the specimen. Among them, referring to the indoor experimental process, during the nanoindentation test, the specimen is fixed on the stage and the indenter only moves in the loading direction. Therefore, a positive displacement constraint in the Y direction is applied at the bottom of the numerical model, and an axisymmetric constraint is applied on the axis of symmetry. At the same time, the horizontal movement and rotation of the indenter in the model are constrained so that the indenter can only move in the vertical direction. The numerical specimen considers the basic material parameters and thermodynamic parameters of the rock, and the specific parameters are shown in Table 1.

[0041] Table 1 Basic Material Parameters of Nanoindentation Numerical Model

[0042]

[0043] The nano-indentation numerical model uses the Drucker-Prager criterion (abbreviated as the DP criterion) to characterize the elastoplastic deformation process of the rock during the indenter penetration. The DP criterion is an optimized generalization based on the Mohr-Coulomb criterion and the von Mises criterion. The DP criterion and the rock damage discrimination principle are as Figure 2 shown Figure 2 The P-T curve in (a) of [] is the yield surface of the linear D-P model on the meridian plane, where is the friction angle, is the material cohesion. Figure 2 (b) of [] is the D-P yield surface, which is the typical yield / flow surface of the D-P linear model in the deviatoric plane. Here, K is the ratio of the triaxial tensile yield stress to the triaxial compressive yield stress. Figure 2 (c) of [] is the stress-strain relationship curve during the rock damage process. Here, σ0 is the elastic limit stress of the rock; D is the damage degree of the rock; σ y0 is the yield stress when the rock starts to be damaged, and at this time the damage factor D = 0; when the plastic strain increases to ε f , the damage factor D = 1, and the rock completely fails and detaches. Its expression is:

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] where is the internal friction angle of the geotechnical material; is the cohesion of the geotechnical material; and are both constants related to the cohesion of the geotechnical material and the internal friction angle ; is the first invariant of the stress tensor; is the second invariant of the stress deviator tensor; , , are the first principal stress, the second principal stress and the third principal stress respectively; D is the damage factor; , are the Young's moduli of the material before and after respectively; is strain; is the equivalent plastic strain when the material begins to be damaged; is stress; is the stress after the material is damaged.

[0051] Compare the load-displacement curves at different temperatures and different depths obtained from the nanoindentation numerical model with those from the nanoindentation experiment to verify the accuracy of the nanoindentation numerical model.

[0052] Step 3: Since in actual engineering, the scale is larger, generally on the centimeter scale. Therefore, based on the verified accurate nanoindentation numerical model parameters, further establish a macroscopic indentation numerical model, as Figure 4 shown. Combine the Latin hypercube sampling method to sample the reservoir temperature and pressure conditions and basic rock mechanics parameters that affect the experimental results. The reservoir temperature and pressure conditions and basic rock mechanics parameters include elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure. According to the actual formation conditions in Area A, the temperature setting range is 100 - 200 °C, the confining pressure setting range is 10 - 80 MPa, the elastic modulus setting range is 15 - 60 GPa, the Poisson's ratio setting range is 0.15 - 0.3, and the compressive strength setting range is 100 - 500 MPa. Randomly select 200 groups of parameters to form a parameter combination database as the conditions for carrying out numerical experiments.

[0053] Latin hypercube sampling method:

[0054] ;

[0055] Among them: represents the estimated value; represents the expectation of; represents the number of samples participating in the computer operation; represents the linear coefficient; represents the standard deviation of; represents the standard deviation of; represents the sampling result value, and the sampling result is any one of the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample; represents the index subscript.

[0056] Apply these combined parameters to the macroscopic indentation numerical model for simulation analysis to obtain the load-displacement curves under different conditions.

[0057] Step 4: Extract three parameters, namely the maximum applied load, the corresponding indentation depth after full loading, and the indentation depth after complete unloading, from the inverted elastic modulus target parameters based on the load-displacement curves under different conditions. Set these three parameters together with the two parameters of temperature and pressure as characteristic values, set the elastic modulus data as the target value, and store the data in the database as the inversion training dataset for the trained neural network.

[0058] Step 5: Based on the Bayesian optimization algorithm, use the mean absolute error as the objective function to optimize the hyperparameter problem of the deep neural network method, screen out the optimal deep neural network combination parameters (as shown in Table 2), and build a deep neural network model accordingly. Input the training set data into the optimized deep neural network to train the model and obtain a single-point inversion prediction model. Taking the elastic modulus output by the model as a reference, analyze the accuracy of the model training results. After 1000 training times, the final result is obtained, and the MAE value of the model is 32.143. The comparison results between the predicted elastic modulus and the true elastic modulus are as Figure 5 shown, indicating that the model predicts the elastic modulus with a very small loss, and thus a single-point inversion prediction model is obtained.

[0059] The calculation method of the inversion error is as follows:

[0060] ;

[0061] where MAE is the mean absolute error, and are the target value and the predicted value respectively, is the number of data.

[0062] Table 2 Optimal deep neural network combination parameter table

[0063]

[0064] Based on the Kriging interpolation algorithm, convert the single-point inversion prediction model into a spatial inversion prediction model. Specifically: obtain the elastic model parameters output by the single-point inversion prediction model; select a variogram model for fitting; construct and solve the Kriging equations according to the fitting results to obtain the weight coefficients of each known point; use the weight coefficients for spatial prediction, generate a spatial grid, and obtain the distribution law and variation characteristics of the elastic modulus in space, realizing the conversion of the single-point inversion prediction model into a spatial inversion prediction model.

[0065] Step 6: In-situ drilling technology can realize the downhole placement of indentation detection equipment while drilling, and collect the detection load-displacement data of wellbore rocks in real time. Through in-situ dot matrix method loading test, downhole rock mechanics parameters can be obtained in real time according to the spatial inversion prediction model, realizing the real-time establishment and update of the spatial distribution data of underground rock mechanics parameters of deep rock masses, and a spatial rock mechanics parameter field can be established. The results are as Figure 6 shown.

[0066] The present invention also provides a system for the spatial inversion prediction method of key in-situ reservoir rock mechanics parameters, specifically including:

[0067] A data processing module, which is used to obtain the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters to construct a data set, and construct a macroscopic indentation numerical model according to the data set.

[0068] A parameter calculation module, which is used to sample the elastic modulus, Poisson's ratio, compressive strength, temperature and pressure parameters of the target rock sample by the Latin hypercube sampling method to form a parameter combination; input the parameter combination into the macroscopic indentation numerical model for simulation analysis to obtain load-displacement curves under different conditions; extract multiple characteristic values from the load-displacement curves under different conditions as the training set; among them, the characteristic values characterize the rock mechanics response parameters.

[0069] A training module, which is used to optimize the hyperparameters of the deep neural network based on the Bayesian optimization algorithm, input the training set data into the optimized deep neural network to train the model, and obtain a single-point inversion prediction model; based on the Kriging interpolation algorithm, convert the single-point inversion prediction model into a spatial inversion prediction model.

[0070] A parameter inversion module, which is used to input the load-displacement data detected in real time of the rock formation into the spatial inversion prediction model to obtain the spatial distribution data of the rock mechanics parameters in the deep rock formation.

[0071] Each module in the above system for the spatial inversion prediction method of key in-situ reservoir rock mechanics parameters can be implemented in whole or in part by software, hardware and their combination. The above modules can be embedded in the processor of the computer device in the form of hardware or be independent of it, or be stored in the memory of the computer device in the form of software, so as to facilitate the processor to call and execute the operations corresponding to the above modules.

[0072] The present invention also provides a computer device, including a memory, a processor and a computer program stored on the memory. The processor executes the computer program to implement the steps in an embodiment of a method for the spatial inversion prediction of key in-situ reservoir rock mechanics parameters. The specific implementation method can refer to the method embodiment and will not be elaborated here.

[0073] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions, and the above instructions can be executed by a processor of a computer device to complete the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in an embodiment of a method for spatial inversion prediction of key in-situ reservoir rock mechanics parameters. The specific implementation method can be referred to the method embodiment and will not be elaborated here.

[0074] Those skilled in the art should understand that the embodiments of the present invention can provide a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0076] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide for realizing the functions in Figure 1 one flow or multiple flows and / or blocksFigure 1 Steps of functions specified in one or more boxes.

[0078] It should be noted that the specific embodiments described above can enable those skilled in the art to understand the invention more comprehensively, but do not limit the invention in any way. Therefore, although this specification and the embodiments have described the invention in detail, those skilled in the art should understand that the invention can still be modified or equivalently replaced; and all technical solutions and improvements that do not depart from the spirit and scope of the invention are covered by the protection scope of the invention patent. Any reference signs in the claims should not be regarded as limiting the claimed claims. Any simple changes or equivalent replacements of technical solutions that can be obviously obtained by any person skilled in the art within the technical scope disclosed by the invention all fall within the protection scope of the invention.

Claims

1. A method for spatially inverse predicting key in-situ rock mechanical parameters of a reservoir, characterized in that Including the following steps: Obtain the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters to construct a data set, and construct a macroscopic indentation numerical model according to the data set; Sample the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method to form parameter combinations; input the parameter combinations into the macroscopic indentation numerical model for simulation analysis to obtain load-displacement curves under different conditions; extract multiple characteristic values from the load-displacement curves under different conditions as the training set; wherein, the characteristic values characterize the rock mechanical response parameters; Optimize the hyperparameters of the deep neural network based on the Bayesian optimization algorithm, input the training set data into the optimized deep neural network to train the model, and obtain a single-point inversion prediction model; based on the Kriging interpolation algorithm, transform the single-point inversion prediction model into a spatial inversion prediction model; Input the load-displacement data detected in real time of the rock stratum into the spatial inversion prediction model to obtain the spatial distribution data of the rock mechanical parameters in the deep rock stratum.

2. The spatial inversion prediction method for in-situ key rock mechanical parameters of a reservoir according to claim 1, wherein The sampling of the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method is specifically through the following formula: ； Among them, represents the estimated value; represents the expectation of represents the number of samples participating in the computer operation; represents the linear coefficient; represents the standard deviation of represents the standard deviation of represents the sampling result value, and the sampling result is any one of the elastic modulus, Poisson's ratio, compressive strength, temperature and pressure parameters of the target rock sample; represents the index subscript.

3. The spatial inversion prediction method for in-situ key rock mechanical parameters of a reservoir according to claim 1, characterized in that, The optimization of the hyperparameters of the deep neural network based on the Bayesian optimization algorithm is specifically: Based on the Bayesian optimization algorithm, use the mean absolute error as the objective function to optimize the hyperparameters of the deep neural network model, and screen out the optimal combination of deep neural network structure parameters.

4. A method for spatially inverse predicting key in-situ reservoir rock mechanical parameters according to claim 1, characterized in that The obtaining of the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters is specifically to conduct mineral component experiments, scanning electron microscope experiments, and indentation experiments on the target rock sample to obtain the mineral composition of the target rock sample and the microscopic information of the sample rock after the indentation experiment; at the same time, conduct triaxial stress experiments to obtain the DP criterion parameters.

5. A method for predicting the spatial inversion of key in-situ rock mechanical parameters of a reservoir according to claim 1, characterized in that The multiple characteristic values specifically include: temperature, pressure, maximum applied load, indentation depth corresponding to the maximum applied load, and indentation depth after complete unloading.

6. A spatial inversion prediction system for in-situ key rock mechanics parameters of a reservoir, characterized in that, Including: A data processing module, configured to obtain the mineral composition of the target rock sample, the microscopic information of the sample rock after the indentation experiment, and the DP criterion parameters to construct a data set, and construct a macroscopic indentation numerical model according to the data set; A parameter calculation module, configured to sample the elastic modulus, Poisson's ratio, compressive strength, temperature, and pressure parameters of the target rock sample by the Latin hypercube sampling method to form parameter combinations; input the parameter combinations into the macroscopic indentation numerical model for simulation analysis to obtain load-displacement curves under different conditions; extract multiple characteristic values from the load-displacement curves under different conditions as the training set; wherein, the characteristic values characterize the rock mechanical response parameters; A training module, configured to optimize the hyperparameters of the deep neural network based on the Bayesian optimization algorithm, input the training set data into the optimized deep neural network to train the model, and obtain a single-point inversion prediction model; based on the Kriging interpolation algorithm, transform the single-point inversion prediction model into a spatial inversion prediction model; The parameter inversion module is used to input the load-displacement data detected in real time of the rock stratum into the spatial inversion prediction model to obtain the spatial distribution data of the rock mechanics parameters in the deep rock stratum.

7. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is loaded by the processor, it can execute the steps of the method according to any one of claims 1 to 5.

Images