Random fracture parameter setting method in DFN model

By using random fracture parameter setting method and intelligent algorithm optimization in the DFN model, the problem that fixed parameters in the traditional model cannot reflect fracture variability is solved, the prediction accuracy and parameter tuning efficiency are improved, and the effectiveness of fracture dynamic processes is achieved Simulation and prediction.

CN119026015APending Publication Date: 2024-11-26GUANGDONG PROVINCE DABAOSHAN MINING CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410974314.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The fixed fracture parameters in the traditional DFN model cannot fully reflect the spatial variability and uncertainty of the fracture parameters, affect the prediction accuracy and reliability of the model, and lack the simulation and prediction capabilities of the dynamic development process of the fracture.

Method used

A random crack parameter setting method in the DFN model is adopted, including data collection, parameter definition, probability model establishment, random parameter generation, parameter verification and adjustment, parameter application, combined with Monte Carlo simulation and intelligent algorithm optimization, using machine learning, Deep learning and genetic algorithms automatically optimize fissure parameters to ensure the scientificity and rationality of the parameters, and simulate the dynamic process of fissure development.

Benefits of technology

It significantly improves the prediction accuracy and parameter tuning efficiency of the DFN model, can more realistically reflect the natural characteristics of the fissure system, simulate and predict the dynamic process of fissure development, and improves the research ability of rock rupture and groundwater flow.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026015A_ABST
    Figure CN119026015A_ABST
Patent Text Reader

Abstract

The invention discloses a random fracture parameter setting method in a DFN model, and relates to the technical field of random fracture parameter setting methods. Comprising the following steps: data collection: collecting geological data of a target area; parameter definition: defining the range and distribution characteristics of fracture parameters according to the collected data; probability model establishment: establishing a probability distribution model of each fracture parameter based on the principle of statistics; random parameter generation: using Monte Carlo simulation or other random sampling methods to generate a group of random fracture parameters according to the probability distribution model; and parameter verification and adjustment: carrying out comparison verification on the generated random parameters and actual fracture data. According to the method, the natural characteristics of the fracture system can be reflected more truly, so that the prediction precision of the DFN model is improved; intelligent algorithms such as machine learning, deep learning and a genetic algorithm are introduced, fracture parameters are automatically optimized, and the parameter tuning efficiency and the performance of the model are remarkably improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of random crack parameter setting methods, and in particular to a random crack parameter setting method in a DFN model. Background Art

[0002] In the fields of geoengineering and hydrogeology, discrete fracture network models are an important tool for simulating and analyzing rock fracture systems and their effects on underground fluid flow, stress transfer, and rock failure processes. Traditional DFN models usually rely on fixed fracture parameters such as fracture length, width, direction, and density, which are often based on empirical estimates or limited field measurement data. However, this approach may not fully reflect the spatial variability and uncertainty of fracture parameters, thus affecting the prediction accuracy and reliability of the model.

[0003] After searching, the Chinese patent application number 202110498351.X discloses a random crack network generation method based on strength spatial variability, which can fully consider the spatial variability of material strength, reflect the real crack morphology, and has high computational efficiency. The method includes: step 1. determining the length and width of the simulation area; step 2. determining the crack width, and dividing the entire simulation area by regular triangle mesh; step 3. generating a random field of material tensile strength in the simulation area to quantitatively characterize the spatial variability of tensile strength; step 4. determining the tensile stress acting on the entire simulation area according to the actual site conditions; step 5. determining the maximum number of iteration steps and the number of cracks according to the actual conditions of the simulation object; step 6. generating an initial fracture point in the simulation area; step 7. performing iterative calculations based on the initial fracture point to determine the direction and probability of crack development, and realize the influence of material strength on the direction of crack development; step 8. generating a crack model combined with a random field. The random crack network generation method in the above patent has the following shortcomings: although it can meet certain application requirements, the ability to simulate and predict the dynamic development process of cracks is insufficient and needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method for setting random crack parameters in a DFN model.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for setting random crack parameters in a DFN model comprises the following steps:

[0007] S1: Data collection: Collect geological data of the target area;

[0008] S2: Parameter definition: Based on the collected data, define the range and distribution characteristics of the fracture parameters;

[0009] S3: Probability model establishment: Based on statistical principles, the probability distribution model of each fracture parameter is established;

[0010] S4: Random parameter generation: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods;

[0011] S5: Parameter verification and adjustment: Compare and verify the generated random parameters with the actual fracture data, and adjust the probability model or generation algorithm if necessary to ensure the authenticity and rationality of the parameters;

[0012] S6: Parameter application: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process.

[0013] Preferably, in the random crack parameter setting method, the probability density function of the crack length L is f(L):

[0014]

[0015] Where μ is the mean of the natural logarithm and σ is the standard deviation of the logarithm;

[0016] The probability density function of the crack direction θ is:

[0017]

[0018] where κ is the concentration parameter, μ is the mean direction, and I0 is the modified Bessel function of the first kind, zeroth order.

[0019] Preferably: in S1, data collection includes:

[0020] S11: Collect geological data of the target area, including rock types, historical fracture records, and ground stress distribution;

[0021] S12: Conduct field exploration to obtain deep geological information;

[0022] S13: Analyze the collected data and screen out key parameters related to crack characteristics;

[0023] In S2, the parameter definition includes:

[0024] S21: Based on the collected data, define the range and distribution characteristics of the fracture parameters;

[0025] S22: Statistical analysis of historical fracture records to determine typical values ​​and coefficients of variation of each parameter;

[0026] S23: Set the physical limits and logical relationships of parameters to ensure the scientificity and rationality of parameter settings.

[0027] Preferably: in S3, establishing the probability model includes:

[0028] S31: Based on statistical principles, a probability distribution model of each crack parameter is established;

[0029] S32: performing correlation analysis on different parameters to determine the mutual influence and dependency between the parameters;

[0030] S33: Use probability plots and goodness-of-fit tests to verify the suitability of the selected distribution model;

[0031] In S4, random parameter generation includes:

[0032] S41: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods;

[0033] S42: Evaluate the impact of different parameters on the output of the DFN model through sensitivity analysis;

[0034] S43: Use multi-scenario simulation to generate fracture parameter combinations under different probability distributions.

[0035] Preferably: in S5, parameter verification and adjustment include:

[0036] S51: Compare and verify the generated random parameters with the actual fracture data to evaluate the rationality of the parameters;

[0037] S52: Adjust the probability model or generation algorithm when necessary to ensure the authenticity and rationality of the parameters;

[0038] S53: Further verify the effectiveness of parameter settings by comparing with field test results;

[0039] In S6, the parameter application includes:

[0040] S61: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process;

[0041] S62: Analyze the consistency between simulation results and actual observation data and evaluate the prediction ability of the model;

[0042] S63: Based on the simulation results, optimize the fracture parameter settings to improve the accuracy and applicability of the model.

[0043] Preferred: Also includes: Intelligent algorithm optimization:

[0044] Apply machine learning algorithms to train and optimize the distribution model of fracture parameters;

[0045] Genetic algorithms are used to automatically tune the fracture parameters and find the optimal parameter combination;

[0046] Through deep learning algorithms, the complex relationship between fracture parameters and geological conditions is automatically identified and learned.

[0047] Preferably, in the intelligent algorithm optimization, the specific method of using the support vector machine algorithm to train and optimize the distribution model of the crack parameters is:

[0048] S71: Feature extraction using existing fracture datasets;

[0049] S72: input the extracted features into the machine learning model for training;

[0050] S73: Use cross-validation and performance evaluation metrics to optimize model parameters;

[0051] Its objective function is expressed as:

[0052]

[0053] sty (i) (ω T x (i) +b)≥1-ξ i ,ξ i ≥0,i=1,…,n

[0054] Where ω and b are the parameters of the hyperplane, ξ i is the slack variable and C is the penalty parameter.

[0055] Preferably, in the intelligent algorithm optimization, the specific method of automatically tuning the crack parameters using the genetic algorithm and finding the optimal parameter combination is:

[0056] S81: Initialize a set of random crack parameters;

[0057] S82: define a fitness function that measures how well the parameter combination fits the training data;

[0058] S83: iteratively updating the parameter combination through selection, crossover and mutation operations;

[0059] S84: Repeat the above process until the termination condition is met;

[0060] The fitness function of individual x is expressed as:

[0061]

[0062] Among them, y i is the observed value, y'i is the model prediction value.

[0063] Preferred: Also includes: Intelligent decision support:

[0064] Use intelligent optimization algorithms to provide decision support for the selection of fracture parameters;

[0065] Combine geomechanical models and intelligent algorithms to simulate and predict the dynamic process of fracture development;

[0066] Through visualization tools and user interfaces, engineers can enjoy intuitive parameter adjustment and result display.

[0067] Preferably: in the intelligent decision support, the basic steps of simulating and predicting the dynamic process of fracture development by combining the geomechanical model and the intelligent algorithm include:

[0068] S91: Define the objective function: the fit between the model prediction results and the actual observed data;

[0069] S92: Initialize parameter space: define possible value ranges for unknown parameters in the geomechanical model;

[0070] S93: Apply intelligent optimization algorithm: Search for the best parameter combination in the parameter space to minimize or maximize the objective function;

[0071] S94: Model simulation: Use optimized parameters to perform dynamic simulation of fracture development in a geomechanical model.

[0072] The beneficial effects of the present invention are:

[0073] 1. The method of the present invention can more realistically reflect the natural characteristics of the fracture system, thereby improving the prediction accuracy of the DFN model; it introduces intelligent algorithms such as machine learning, deep learning and genetic algorithms to automatically optimize fracture parameters, significantly improving the efficiency of parameter tuning and the performance of the model.

[0074] 2. The present invention makes full use of geological data, historical fracture records and field exploration information to ensure the scientificity and rationality of parameter settings; combined with geomechanical models and intelligent algorithms, it can simulate and predict the dynamic process of fracture development, so as to better study rock fracture and groundwater flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 The present invention provides a flow chart of a method for setting random crack parameters in a DFN model. DETAILED DESCRIPTION

[0076] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0077] Embodiment 1:

[0078] A method for setting random crack parameters in a DFN model comprises the following steps:

[0079] S1: Data collection: Collect geological data of the target area, including rock types, historical fracture records, ground stress distribution, etc.;

[0080] S2: Parameter definition: Based on the collected data, define the range and distribution characteristics of crack parameters, such as crack length, width, spacing, direction and density;

[0081] S3: Probability model establishment: Based on statistical principles, a probability distribution model of each fracture parameter is established, such as normal distribution, lognormal distribution or exponential distribution;

[0082] S4: Random parameter generation: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods;

[0083] S5: Parameter verification and adjustment: Compare and verify the generated random parameters with the actual fracture data, and adjust the probability model or generation algorithm if necessary to ensure the authenticity and rationality of the parameters;

[0084] S6: Parameter application: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process.

[0085] Among them, in the random crack parameter setting method, the probability density function of the crack length L is:

[0086]

[0087] Where μ is the mean of the natural logarithm and σ is the standard deviation of the logarithm;

[0088] The probability density function of the crack direction θ is:

[0089]

[0090] where κ is the concentration parameter, μ is the mean direction, and I0 is the modified Bessel function of the first kind, zeroth order.

[0091] Wherein, in said S1, data collection includes:

[0092] S11: Collect geological data of the target area, including rock types, historical fracture records, ground stress distribution, etc.;

[0093] S12: Conduct field exploration and obtain deep geological information by means of drilling, seismic exploration, etc.;

[0094] S13: Analyze the collected data and screen out key parameters related to crack characteristics.

[0095] Among them, in S2, the parameter definition includes:

[0096] S21: Based on the collected data, define the range and distribution characteristics of crack parameters, such as crack length, width, spacing, direction and density;

[0097] S22: Statistical analysis of historical fracture records to determine typical values ​​and coefficients of variation of each parameter;

[0098] S23: Set the physical limits and logical relationships of parameters to ensure the scientificity and rationality of parameter settings.

[0099] Wherein, in said S3, the probability model establishment includes:

[0100] S31: Based on statistical principles, a probability distribution model of each fracture parameter is established, such as normal distribution, lognormal distribution or exponential distribution;

[0101] S32: performing correlation analysis on different parameters to determine the mutual influence and dependency between the parameters;

[0102] S33: Use probability plots and goodness-of-fit tests to verify the suitability of the selected distribution model.

[0103] Wherein, in said S4, random parameter generation includes:

[0104] S41: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods;

[0105] S42: Evaluate the impact of different parameters on the output of the DFN model through sensitivity analysis;

[0106] S43: Use multi-scenario simulation to generate fracture parameter combinations under different probability distributions.

[0107] Among them, in S5, parameter verification and adjustment include:

[0108] S51: Compare and verify the generated random parameters with the actual fracture data, and use statistical methods such as hypothesis testing to evaluate the rationality of the parameters;

[0109] S52: Adjust the probability model or generation algorithm when necessary to ensure the authenticity and rationality of the parameters;

[0110] S53: Further verify the effectiveness of parameter settings by comparing with field test results.

[0111] Among them, in said S6, the parameter application includes:

[0112] S61: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process;

[0113] S62: Analyze the consistency between simulation results and actual observation data and evaluate the prediction ability of the model;

[0114] S63: Based on the simulation results, optimize the fracture parameter settings to improve the accuracy and applicability of the model.

[0115] Embodiment 2:

[0116] A method for setting random crack parameters in a DFN model. Based on Example 1, this embodiment further includes: intelligent algorithm optimization:

[0117] Apply machine learning algorithms to train and optimize the distribution model of fracture parameters;

[0118] Genetic algorithms are used to automatically tune the fracture parameters and find the optimal parameter combination;

[0119] Through deep learning algorithms, the complex relationship between fracture parameters and geological conditions is automatically identified and learned.

[0120] Among them, in the intelligent algorithm optimization, the specific method of using the support vector machine algorithm to train and optimize the distribution model of the crack parameters is:

[0121] S71: Feature extraction using existing fracture datasets;

[0122] S72: input the extracted features into the machine learning model for training;

[0123] S73: Use cross-validation and performance evaluation indicators (such as coefficient of determination, mean square error, etc.) to optimize model parameters;

[0124] Its objective function is expressed as:

[0125]

[0126] sty (i) (ω T x (i) +b)≥1-ξ i ,ξ i ≥0,i=1,…,n

[0127] Where ω and b are the parameters of the hyperplane, ξ i is the slack variable and C is the penalty parameter.

[0128] Among them, in the intelligent algorithm optimization, the specific method of using genetic algorithm to automatically tune the crack parameters and find the optimal parameter combination is:

[0129] S81: Initialize a set of random crack parameters;

[0130] S82: define a fitness function that measures how well the parameter combination fits the training data;

[0131] S83: iteratively updating the parameter combination through selection, crossover and mutation operations;

[0132] S84: Repeat the above process until the termination condition is met (such as reaching the maximum number of iterations or the quality of the solution is no longer improved);

[0133] The fitness function of individual x is expressed as:

[0134]

[0135] Among them, y i is the observed value, y' i is the model prediction value.

[0136] Embodiment 3:

[0137] A method for setting random crack parameters in a DFN model. Based on Example 1, this embodiment further includes: intelligent decision support:

[0138] Use intelligent optimization algorithms to provide decision support for the selection of fracture parameters;

[0139] Combine geomechanical models and intelligent algorithms to simulate and predict the dynamic process of fracture development;

[0140] Provide engineers with intuitive parameter adjustment and result display through visualization tools and user interfaces;

[0141] Among them, in the intelligent decision support, the basic steps of simulating and predicting the dynamic process of fracture development by combining the geomechanical model and the intelligent algorithm include:

[0142] S91: Define the objective function: the fit between the model prediction results and the actual observed data;

[0143] S92: Initialize parameter space: define possible value ranges for unknown parameters in the geomechanical model (such as elastic modulus, Poisson's ratio, shear strength, etc.);

[0144] S93: Apply intelligent optimization algorithm: search for the best parameter combination in the parameter space to minimize (or maximize) the objective function;

[0145] S94: Model simulation: Use optimized parameters to perform dynamic simulation of fracture development in a geomechanical model.

[0146] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for setting random crack parameters in DFN model. It is characterized in that The steps include: S1: Data collection: Collect geological data of the target area; S2: Parameter definition: Based on the collected data, define the range and distribution characteristics of the fracture parameters; S3: Probability model establishment: Based on statistical principles, the probability distribution model of each fracture parameter is established; S4: Random parameter generation: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods; S5: Parameter verification and adjustment: Compare and verify the generated random parameters with the actual fracture data; S6: Parameter application: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process.

2. According to the method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that In the random crack parameter setting method, the probability density function of the crack length L, f(L), is: Where μ is the mean of the natural logarithm and σ is the standard deviation of the logarithm; The probability density function of the crack direction θ is: Where κ is the concentration parameter, μ is the average direction, and I 0 is the modified Bessel function of the first kind, zeroth order.

3. According to the method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that In S1, data collection includes: S11: Collect geological data of the target area, including rock types, historical fracture records, and ground stress distribution; S12: Conduct field exploration to obtain deep geological information; S13: Analyze the collected data and screen out key parameters related to crack characteristics; In S2, the parameter definition includes: S21: Based on the collected data, define the range and distribution characteristics of the fracture parameters; S22: Statistical analysis of historical fracture records to determine typical values ​​and coefficients of variation of each parameter; S23: Set the physical limits and logical relationships of parameters to ensure the scientificity and rationality of parameter settings.

4. The method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that In S3, the probability model establishment includes: S31: Based on statistical principles, a probability distribution model of each crack parameter is established; S32: performing correlation analysis on different parameters to determine the mutual influence and dependency between the parameters; S33: Use probability plots and goodness-of-fit tests to verify the suitability of the selected distribution model; In S4, random parameter generation includes: S41: Generate a set of random fracture parameters according to the probability distribution model using Monte Carlo simulation or other random sampling methods; S42: Evaluate the impact of different parameters on the output of the DFN model through sensitivity analysis; S43: Use multi-scenario simulation to generate fracture parameter combinations under different probability distributions.

5. The method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that In S5, parameter verification and adjustment include: S51: Compare and verify the generated random parameters with the actual fracture data to evaluate the rationality of the parameters; S52: Adjust the probability model or generation algorithm when necessary to ensure the authenticity and rationality of the parameters; S53: Further verify the effectiveness of parameter settings by comparing with field test results; In S6, the parameter application includes: S61: Apply the verified random fracture parameters to the DFN model to simulate the rock fracture process; S62: Analyze the consistency between simulation results and actual observation data and evaluate the prediction ability of the model; S63: Based on the simulation results, optimize the fracture parameter settings to improve the accuracy and applicability of the model.

6. The method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that Also includes: Intelligent algorithm optimization: Apply machine learning algorithms to train and optimize the distribution model of fracture parameters; Genetic algorithms are used to automatically tune the fracture parameters and find the optimal parameter combination; Through deep learning algorithms, the complex relationship between fracture parameters and geological conditions is automatically identified and learned.

7. The method for setting random crack parameters in a DFN model according to claim 6, It is characterized in that In the intelligent algorithm optimization, the specific method of using the support vector machine algorithm to train and optimize the distribution model of the crack parameters is: S71: Feature extraction using existing fracture datasets; S72: input the extracted features into the machine learning model for training; S73: Use cross-validation and performance evaluation metrics to optimize model parameters; Its objective function is expressed as: style (i) (oh T x (i) +b)≥1-ξ i ,x i ≥0,i=1,…,n Where ω and b are the parameters of the hyperplane, ξ i is the slack variable and C is the penalty parameter.

8. The method for setting random crack parameters in a DFN model according to claim 6, It is characterized in that In the intelligent algorithm optimization, the genetic algorithm is used to automatically tune the crack parameters, and the specific method of finding the optimal parameter combination is as follows: S81: Initialize a set of random crack parameters; S82: define a fitness function that measures how well the parameter combination fits the training data; S83: iteratively updating the parameter combination through selection, crossover and mutation operations; S84: Repeat the above process until the termination condition is met; The fitness function of individual x is expressed as: Among them, y i is the observed value, y' i is the model prediction value.

9. The method for setting random crack parameters in a DFN model according to claim 1, It is characterized in that Also includes: Intelligent decision support: Use intelligent optimization algorithms to provide decision support for the selection of fracture parameters; Combine geomechanical models and intelligent algorithms to simulate and predict the dynamic process of fracture development; Through visualization tools and user interfaces, engineers can enjoy intuitive parameter adjustment and result display.

10. The method for setting random crack parameters in a DFN model according to claim 9, It is characterized in that In the intelligent decision support, the basic steps of simulating and predicting the dynamic process of fracture development by combining the geomechanical model and the intelligent algorithm include: S91: Define the objective function: the fit between the model prediction results and the actual observed data; S92: Initialize parameter space: define possible value ranges for unknown parameters in the geomechanical model; S93: Apply intelligent optimization algorithm: Search for the best parameter combination in the parameter space to minimize or maximize the objective function; S94: Model simulation: Use optimized parameters to perform dynamic simulation of fracture development in a geomechanical model.

Citation Information

Cited By

  • Rock mass stability analysis method and system based on DFN-VR

    CN121787126A

  • A rock mass stability analysis method and system based on DFN-VR

    CN121787126B