Rock fracture physical modeling method and system and storage medium

Abstract

The invention is suitable for the technical field of rock mechanics, and provides a rock fracture physical modeling method and system and a storage medium, and the method comprises the steps: carrying out on-site detection and drilling, obtaining geometric parameters and spatial distribution data of a fracture set, and generating a fracture network probability distribution model; obtaining interaction parameters among crack groups according to the model, and determining statistical distribution characteristics of collaborative sliding; judging whether the statistical distribution characteristics are subjected to data enhancement or not to obtain a crack dynamic behavior data set; constructing spatio-temporal evolution simulation of crack group collaborative sliding by using the data set, and judging the occurrence probability of a synergistic effect; acquiring local stress field change data based on the occurrence probability, and determining a rock mass instability critical point predicted value; integrating seepage characteristics and mechanical response parameters according to the predicted value, determining long-term reliability evaluation indexes of the underground system, and adjusting parameters of the probability distribution model according to the indexes to obtain a prediction model; according to the invention, the accuracy and reliability of underground system instability risk prediction can be improved.

Description

Technical Field

[0001] This application relates to the field of rock mechanics technology, specifically a method, system, and storage medium for physical modeling of rock fractures. Background Technology

[0002] Underground engineering plays a crucial role in ensuring energy extraction, tunnel construction, and geological disaster prevention, with its safety directly impacting the long-term stability and socio-economic benefits of the project. However, the complexity of fracture networks in underground rock masses makes instability risk prediction a critical challenge. Fracture networks not only affect the mechanical properties of the rock mass but are also closely related to groundwater seepage and stress distribution; accurate prediction of their behavior is essential for engineering safety. Existing methods typically rely on static fracture distribution data or single mechanical parameter analysis, making it difficult to capture the collaborative behavior of fracture communities in dynamic environments. This results in inaccurate predictions under complex geological conditions, particularly in long-term reliability assessments, where a comprehensive consideration of the dynamic evolution of the fracture network is lacking. Current solutions have significant limitations in handling the collaborative sliding characteristics of fracture networks. Many methods focus only on the geometric features of individual fractures or local stress changes, neglecting the dynamic characteristics of interactions between fracture communities. This limitation prevents prediction models from accurately reflecting the overall behavior of the fracture network in complex geological environments. Especially during long-term operation, collaborative sliding of fractures can trigger abrupt changes in the local stress field, leading to rock mass instability. For example, in deep mining, the sliding of fracture communities can be exacerbated by groundwater seepage or changes in external loads. However, existing methods struggle to quantify how these factors collectively influence the dynamic evolution of the fracture network, making it impossible to accurately determine the critical conditions for instability. The core technical challenge lies in comprehensively capturing the dynamic characteristics of the collaborative sliding of fracture communities. This characteristic involves not only the geometry and spatial distribution of fractures but also is closely related to the stress field, seepage characteristics, and temporal evolution. Because the interaction of fracture communities in dynamic environments is highly nonlinear, a single mechanical or seepage analysis cannot fully describe its complex behavior. For instance, in tunnel construction, the collaborative sliding of fracture communities may trigger a chain reaction due to localized stress concentration. However, existing technologies struggle to infer the dynamic response of the entire fracture network from limited detection data, thus failing to accurately predict instability risks. This lack of effective analysis of dynamic characteristics often causes predictive models to fail when faced with complex geological conditions.

[0003] How to comprehensively characterize the dynamic behavior of coordinated sliding of fracture groups by integrating multi-source data, and quantify the critical conditions for rock mass instability based on this, has become a key issue in predicting the long-term reliability of underground systems. Therefore, in view of the above situation, there is an urgent need to provide a physical modeling method, system, and storage medium for rock fractures to overcome the shortcomings in current practical applications. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, and storage medium for physical modeling of rock fractures, which can improve the accuracy and reliability of predicting the instability risk of underground systems.

[0005] In a first aspect, this application provides a method for physical modeling of rock fractures, the method comprising:

[0006] Step S1: Obtain the geometric parameters and spatial distribution data of the rock fracture set, and generate a probability distribution model of the fracture network;

[0007] Step S2: Obtain the interaction parameters between the crack group based on the probability distribution model, and determine the statistical distribution characteristics of the cooperative sliding of the crack group;

[0008] Step S3: Perform data augmentation on the statistical distribution characteristics and generate a crack dynamic behavior dataset. Based on the crack dynamic behavior dataset, construct a spatiotemporal evolution simulation of crack group cooperative sliding and calculate the probability of the cooperative effect occurring.

[0009] Step S4: Based on the occurrence probability, obtain the variation data of the local stress field and calculate the critical point prediction value of rock mass instability. Based on the critical point prediction value, combined with seepage characteristics and mechanical response parameters, determine the long-term reliability assessment index of the underground system.

[0010] Step S5: Adjust the parameters of the probability distribution model according to the long-term reliability assessment index to obtain an optimized crack network behavior prediction model.

[0011] Secondly, this application provides a physical modeling system for rock fractures, the system comprising:

[0012] The acquisition module is used to acquire the geometric parameters and spatial distribution data of the rock fracture set and generate a probability distribution model of the fracture network.

[0013] The determination module is used to obtain the interaction parameters between the crack group based on the probability distribution model, and to determine the statistical distribution characteristics of the cooperative sliding of the crack group;

[0014] The calculation module is used to perform data augmentation on the statistical distribution characteristics and generate a crack dynamic behavior dataset. Based on the crack dynamic behavior dataset, a spatiotemporal evolution simulation of crack group cooperative sliding is constructed, and the probability of the occurrence of cooperative effect is calculated.

[0015] The evaluation module is used to obtain local stress field variation data based on the occurrence probability, calculate the critical point prediction value of rock mass instability, and determine the long-term reliability evaluation index of the underground system based on the critical point prediction value, combined with seepage characteristics and mechanical response parameters.

[0016] An optimization module is used to adjust the parameters of the probability distribution model according to the long-term reliability assessment index to obtain an optimized crack network behavior prediction model.

[0017] Thirdly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a physical modeling method for rock fractures as described above.

[0018] Compared with the prior art, the beneficial effects of this application are as follows:

[0019] This application provides a physical modeling method for rock fractures. By establishing a systematic approach from field data acquisition to probability distribution model generation, discrete fracture data is transformed into a statistically significant network model. Monte Carlo simulation is employed to fully consider geological uncertainties, significantly improving the accuracy and reliability of fracture network modeling. Mechanical interaction parameters and a random forest algorithm are introduced, combined with data augmentation techniques to construct a dynamic fracture behavior dataset, enabling intelligent identification and quantitative prediction of collaborative sliding patterns in fracture communities. By integrating seepage characteristics and mechanical response parameters, the constructed mechanical-seepage coupling model more realistically reflects the actual working state of underground systems. A feedback optimization mechanism based on long-term reliability assessment indicators enables continuous improvement of the fracture network model, significantly enhancing the accuracy of long-term predictions. This method provides reliable technical support for risk assessment and disaster early warning in underground engineering. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of an embodiment of a physical modeling method for rock fractures in this application.

[0022] Figure 2 This is a flowchart illustrating data augmentation of statistical distribution characteristics in an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of one embodiment of a physical modeling system for rock fractures in this application. Detailed Implementation

[0024] This application provides a method and system for physical modeling of rock fractures. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] Example 1:

[0026] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of a physical modeling method for rock fractures in this application includes:

[0027] Step S1: Obtain the geometric parameters and spatial distribution data of the rock fracture set, and generate a probability distribution model of the fracture network.

[0028] The probability distribution model for generating the crack network in step S1 includes:

[0029] Geometric parameters of the fracture set, including length, width, and dip angle, were obtained through on-site detection. Spatial distribution data of the fracture set, including spatial distribution coordinates and density data, were obtained through drilling. Based on the geometric parameters and spatial distribution data, multiple random scenarios were generated using Monte Carlo simulation. For multiple random scenarios, the spatial connectivity and geometric distribution probability of the fracture network were calculated. Based on the spatial connectivity and geometric distribution probability, a probability distribution model of the fracture network was constructed, whereby the probability distribution model characterizes the spatial distribution pattern and geometric characteristics of the fracture set.

[0030] Specifically, raw data is collected through a combination of on-site exploration and drilling. On-site exploration mainly targets rock masses exposed on the surface or those revealed by excavation trenches. Geological compasses, rangefinders, and scanning imaging technologies are used to accurately obtain the geometric parameters of the fracture set. These geometric parameters include the fracture trace length, fracture surface width, and the angle between the fracture surface and the horizontal plane, i.e., the dip angle, thus forming a basic dataset characterizing the fracture morphology itself. At the same time, engineering drilling operations penetrate the target rock mass. Core logging and borehole optical television imaging technologies are used to record the spatial coordinates of each drilled fracture, and the number of fractures developed per unit length of borehole is counted to calculate the fracture density. This yields spatial distribution data defining the fracture distribution in three-dimensional space.

[0031] After acquiring the aforementioned raw data, a random crack network was constructed using the Monte Carlo simulation method. The specific implementation process was as follows: First, probability distribution fitting was performed on the collected geometric parameters and spatial distribution data. Statistical methods such as maximum likelihood estimation were used to determine statistical laws such as crack length following a negative exponential distribution and dip angle following a normal distribution, and a functional relationship between crack density and depth was established. Then, based on these fitted probability distribution functions, a large number of random crack network scenarios conforming to actual statistical laws were repeatedly generated in the computer. Each scenario contained a set of statistically equivalent crack combinations. Next, for each generated random crack network scenario, graph theory algorithms were used to identify interconnected cracks. The crack clustering algorithm calculates the spatial connectivity index between crack groups. The calculation process is as follows: Each crack in three-dimensional space is abstracted as a "node" in graph theory; the connection rules between cracks are defined: when the shortest distance between any two cracks in three-dimensional space is less than a preset critical value or the two cracks directly intersect in space, the two cracks are considered connected, and an edge is established between their corresponding nodes; a depth-first search or disjoint-set data structure algorithm is applied to traverse the entire graph network to identify all interconnected crack clusters; the connectivity index is calculated, specifically including the total number of cracks contained in the largest connected cluster, the total surface area and total volume of the largest connected cluster, and the total number of connected clusters.

[0032] Based on the statistical analysis results of spatial connectivity indices and geometric distribution probabilities of a large number of random scenarios, a probability distribution model of crack networks is constructed that can fully characterize the spatial distribution pattern, geometric characteristics, and potential water-conducting or mechanical conduction path capabilities of crack assemblies. Mathematically, this model is expressed as a set of probability density functions and their covariance relationships that define crack generation rules, and it incorporates system-level statistical features based on connectivity analysis, providing a solid probabilistic and physical foundation for subsequent analysis of crack group interactions and collaborative behavior.

[0033] Step S2: Obtain the interaction parameters between the crack groups based on the probability distribution model, and determine the statistical distribution characteristics of the cooperative sliding of the crack groups.

[0034] In step S2, determining the statistical distribution characteristics of the coordinated sliding of the crack community includes:

[0035] Spatial proximity and geometric correlation parameters among crack groups are extracted from the probability distribution model; mechanical interaction parameters among crack groups are calculated based on the spatial proximity and geometric correlation parameters; the mechanical interaction parameters are classified using the random forest algorithm to determine the cooperative sliding mode of crack groups; for the cooperative sliding mode of crack groups, the distribution characteristics of sliding rate and sliding direction are statistically analyzed and defined as the statistical distribution characteristics of cooperative sliding of crack groups.

[0036] Specifically, based on the generated crack network probability distribution model, spatial clustering methods are used to identify spatially close crack groups. The average distance between each crack group is identified as a spatial proximity parameter, and the angle between the crack surface normals is calculated as a geometric correlation parameter. A computational model is established based on the Coulomb friction criterion in rock mechanics, incorporating both the average distance representing spatial proximity and the crack surface angle representing geometric correlation into the computational framework. Spatial proximity directly influences the mechanical interaction intensity through a distance attenuation function, while geometric correlation adjusts the effective friction coefficient through a direction effect function. Together, they constitute a complete mechanical interaction evaluation system. The equivalent friction coefficient is calculated using the following formula: Quantification is performed, among which, The basic friction coefficient of rock was obtained through laboratory direct shear tests. Equivalent coefficient of friction It is the angle between the normal vectors of the crack surface. When calculating the mechanical interaction parameters, the formula is first used: Calculate the potential shear stress, where It is the cohesion of rocks. To determine the effective normal stress, we then use the formula: Calculate the mechanical interaction parameters, where, It is the crack spacing (spatial proximity parameter). This is the attenuation coefficient, typically ranging from 0.5 to 1.0. By incorporating spatial proximity and geometric correlation parameters into rock mechanics criteria, a reliable mapping from fracture geometry to mechanical coupling strength is established, enabling the calculation of interaction parameters to possess both physical explicitness and quantitative precision.

[0037] The obtained mechanical interaction parameters, along with supplementary features such as crack density and population size, form a feature vector, which serves as input to a random forest classification model. This pre-trained model can classify crack populations into different cooperative sliding patterns based on the input features, including typical patterns such as parallel sliding, cross-locking, and chain reactions. Combined with the random forest classification model, it can adaptively distinguish cooperative sliding patterns under complex geological conditions, significantly improving the ability to identify and predict the dynamic behavior of crack populations. In practical applications, the number of decision trees in the random forest can be adjusted according to the complexity of the crack network. 100 trees are used in complex scenarios to ensure classification accuracy, while 50 trees are used in simple scenarios to improve computational efficiency. For each cooperative sliding pattern identified in the classification, its corresponding sliding behavior data, including sliding rate observations, are statistically analyzed. The sliding rate sample observations of the crack population under different cooperative sliding patterns are combined to generate statistical distribution features. This distribution feature generation method based on sliding pattern statistics provides key input for subsequent data augmentation and long-term stability assessment, effectively supporting engineering risk early warning and prevention decisions.

[0038] Step S3: Perform data augmentation on the statistical distribution characteristics and generate a crack dynamic behavior dataset. Based on the crack dynamic behavior dataset, construct a spatiotemporal evolution simulation of crack group cooperative sliding and calculate the probability of the occurrence of cooperative effects.

[0039] Step S3, which involves data augmentation of statistical distribution characteristics and generation of a crack dynamic behavior dataset, includes:

[0040] The sliding rate and sliding direction are extracted from the statistical distribution characteristics, and the mean and variance of the sliding rate are calculated. If the mean or variance of the sliding rate exceeds a preset threshold, the statistical distribution characteristics are extended using a finite data interpolation method. Based on the extended statistical distribution characteristics, an enhanced sliding rate dataset is generated. For the enhanced sliding rate dataset, a crack dynamic behavior dataset is constructed by combining the geometric distribution probability in the probability distribution model.

[0041] Specifically, such as Figure 2The flowchart illustrates the data augmentation process for statistical distribution characteristics. The mean and variance parameters of the sliding rate are extracted from the statistical distribution characteristics. Assuming a mean threshold of 5 mm / year and a variance threshold of 2.0, if the mean or variance of the sliding rate exceeds the preset threshold, it indicates insufficient representativeness or excessive uncertainty in the existing data, necessitating the data augmentation process. The data augmentation process uses the Kriging spatial interpolation method to expand the original statistical distribution characteristics. This method first constructs a variogram model, analyzes the spatial correlation structure between existing data points, and determines the optimal semivariogram parameters. Then, based on the principle of spatial autocorrelation, new interpolation points are generated by minimizing the estimated variance while maintaining the original statistical characteristics. After data expansion, an enhanced sliding rate dataset is generated using Monte Carlo random sampling. The sampling process strictly follows the expanded statistical distribution characteristics to ensure that the newly generated data maintains the statistical regularity of the original data and effectively fills data gaps. If the mean or variance of the sliding rate does not exceed the preset threshold, it indicates that the current statistical distribution characteristics are sufficiently representative, meaning no data augmentation is needed, and the enhanced sliding rate dataset is generated directly based on the current statistical distribution characteristics. Subsequently, the enhanced slip rate dataset was integrated with the geometric probability distributions in the probability distribution model. These geometric probability distributions include key parameters such as crack length distribution, orientation distribution, and spacing distribution. During the integration process, a feature matching algorithm was used to ensure that each slip rate data point was correctly associated with its corresponding geometric feature. This resulted in the construction of a crack dynamic behavior dataset containing multidimensional correlations between rate and geometric features. By using a data augmentation mechanism combining quantization thresholds and Kriging interpolation, the statistical feature uncertainty caused by the sparsity of the original data was effectively addressed. The multidimensional dynamic behavior dataset constructed using the feature matching algorithm ensured the spatiotemporal consistency between mechanical parameters and geometric features, providing a reliable data foundation for subsequent accurate simulations.

[0042] Step S3, which calculates the probability of the synergistic effect occurring, includes:

[0043] Time series data on sliding rate and sliding direction are extracted from the crack dynamic behavior dataset; a dynamic evolution model of crack group cooperative sliding is constructed based on the time series data; the sliding behavior of crack group in time and space is simulated through the dynamic evolution model; and the probability of the occurrence of cooperative effect among crack groups is calculated for the sliding behavior.

[0044] Specifically, firstly, time-series data on slip rate and slip direction are systematically extracted from the crack dynamic behavior dataset. The dataset is then sorted and reorganized according to the time dimension to establish a continuous and complete time-series dataset. During data extraction, special attention is paid to maintaining the spatiotemporal correspondence between the rate and direction parameters, ensuring that the data at each time point contains complete kinematic features. Based on the extracted time-series data, a dynamic evolution model of crack swarm cooperative slip is constructed. This model uses Markov chain theory to describe the state transition process among crack swarms, quantifying the evolutionary laws between different slip modes by calculating the state transition probability matrix. In practice, the discrete state space of the crack swarm is first defined, including stable states, micro-slip states, and cooperative slip states. Then, based on historical time-series data, the transition frequency between each state is statistically analyzed, and a transition probability matrix is ​​constructed. To improve the model's adaptability to complex geological conditions, a random forest algorithm is introduced for intelligent identification and classification of cooperative slip modes. The random forest model takes features such as slip rate, direction change rate, and crack spacing as input and outputs the classification results of the slip modes. The parameter settings of the Markov model are dynamically adjusted based on these results. By constructing a comprehensive dynamic evolution model, the sliding behavior of the crack community in time and space is simulated. The simulation process adopts a discrete time stepping method. Within each time step, the motion state of the crack community is updated according to the current state and transition probability. The simulation output includes the spatial position, sliding rate and motion direction of each crack at different time points, forming a complete spatiotemporal evolution path.

[0045] Finally, based on the simulated sliding behavior samples, the Monte Carlo method is used to calculate the probability of synergistic effects. Specifically, firstly, criteria for determining synergistic effects are defined, including indicators such as the sliding direction consistency threshold and the rate synergistic coefficient. Then, simulation scenarios are generated through extensive random sampling, and the proportion of crack groups satisfying the synergistic conditions is statistically analyzed in each scenario. After a sufficient number of sampling calculations, the frequency of synergistic events is used as an estimate of the probability of synergistic effects. In scenarios where seepage coupling effects need to be considered, seepage characteristic parameters such as pore pressure also need to be incorporated into the sampling rules of the Monte Carlo simulation to ensure the engineering applicability of the probability calculation results. The final output probability value and corresponding spatiotemporal distribution characteristics provide key input parameters for subsequent local stress field analysis and rock mass stability assessment.

[0046] Step S4: Obtain the variation data of the local stress field based on the probability of occurrence, and calculate the predicted value of the critical point of rock mass instability. Based on the predicted value of the critical point, combined with the seepage characteristics and mechanical response parameters, determine the long-term reliability assessment index of the underground system.

[0047] Step S4, which calculates the predicted critical point for rock mass instability, includes:

[0048] High-probability sliding regions are extracted from the probability of occurrence of synergistic effects among crack clusters. High-probability sliding regions are those with an occurrence probability greater than a preset threshold. For high-probability sliding regions, the stress distribution change of the local stress field is calculated. Based on the stress distribution change, the influence parameters of crack cluster sliding on rock mass stability are determined. Through the influence parameters, a prediction model for rock mass instability is constructed. Based on the prediction model, the predicted value of the critical point for rock mass instability is output.

[0049] Specifically, high-probability slip regions are extracted from the probability of synergistic effects. Assuming a probability threshold of 0.7, regions exceeding this threshold are marked as high-probability slip regions, and a corresponding list of spatial coordinates is generated. For these marked high-probability slip regions, a finite element method is used for mesh generation. Based on the interaction parameters of the crack community, the principal stress and shear stress changes of each mesh node are calculated. The principal stress is calculated using the eigenvalues ​​of the stress tensor, and the shear stress is solved based on the Mohr's circle principle, ultimately forming a contour map of stress distribution changes. Based on the stress distribution change results, the influence parameters of crack community slip on rock mass stability are determined. This includes extracting peak stress points from the contour map to calculate the slip influence coefficient, assessing the stability decay rate by combining mechanical response parameters such as the rock mass elastic modulus, and statistically analyzing the distribution characteristics of the influence parameters through Monte Carlo simulation. Based on the obtained influence parameters, a critical point prediction model for rock mass instability is constructed. This model uses regression analysis to establish a linear relationship between the predicted critical point value and the influence parameters, in the form that the predicted value equals the initial stability threshold minus the decay rate multiplied by the stress change. Simultaneously, the random forest algorithm is used to validate the model and optimize the parameters, ensuring that the model accuracy exceeds 85%. Finally, the current stress data is input into the trained prediction model, which outputs the predicted critical point of rock mass instability for subsequent long-term reliability assessment.

[0050] In step S4, the long-term reliability assessment indicators for the underground system are determined, including:

[0051] The stress threshold for rock mass instability is extracted from the critical point prediction values; based on the stress threshold, the seepage characteristic parameters of the underground system are obtained, including permeability and pore pressure; combining the seepage characteristic parameters and mechanical response parameters, a mechanical-seepage coupling model of the underground system is constructed; the long-term stability index of the underground system is calculated through the mechanical-seepage coupling model; based on the long-term stability index, a long-term reliability assessment index of the underground system is generated.

[0052] Specifically, the stress threshold for rock mass instability is extracted from the predicted critical point values. By analyzing the predicted value data structure, the maximum stress point is located as the key threshold parameter. Based on this stress threshold, corresponding seepage characteristic parameters are obtained from a pre-stored underground system database, including permeability parameters calculated based on Darcy's law and pore pressure parameters obtained based on the hydrostatic pressure distribution. Subsequently, these seepage characteristic parameters are combined with mechanical response parameters (such as elastic modulus, Poisson's ratio, etc.) to construct a mechanical-seepage coupled model of the underground system. This model integrates mechanical and seepage equations through Biot's theory, establishes the coupling relationship between effective stress and pore pressure, uses the finite element method for numerical solution, and sets appropriate boundary conditions to ensure model stability. Based on the constructed mechanical-seepage coupled model, a simulation framework in the time dimension is first established, with the simulation duration set to the service life required by the engineering design (e.g., 50 years), and a time step of 1 month. Within each time step, the model updates the distribution of the rock mass stress field and seepage field by iteratively solving the coupled equations. The specific solution process employs the Newton-Raphson method to ensure convergence of mechanical and seepage equilibrium at each time step. The stability index is quantified based on the Mohr-Coulomb criterion by calculating the safety factor for each grid cell. The safety factor is defined as follows: ,in, For rock mass cohesion, It is the internal friction angle. For effective normal stress, To address shear stress, the model performs spatial statistics on the safety factors of all mesh elements at each time step, recording the minimum safety factor, average safety factor, and the proportion of critical elements with a safety factor less than 1. The long-term stability index consists of three core parameters: first, stability persistence, defined as the proportion of time during which the safety factor remains greater than 1.2 throughout the simulation period; second, stability decay rate, obtained through linear regression analysis of the slope of the safety factor over time; and finally, critical state probability, statistically analyzing the cumulative frequency of safety factors falling below 1.0. These three parameters characterize the long-term stability features of the system from different dimensions. Based on these parameters, a long-term reliability assessment index is finally generated using a weighted synthesis method. ,in, For stability and durability, For stability decay rate, This represents the critical state probability. , , The weighting coefficient is determined based on the importance of the project, and is typically set to 0.4, 0.3, or 0.3. The value ranges from 0 to 1, with a value closer to 1 indicating higher system reliability. A multi-parameter weighted comprehensive evaluation method fully considers all dimensions of the system's long-term stability; dynamic time-series analysis accurately captures the stability evolution trend; and the established quantitative indicator system provides an intuitive and reliable basis for project decision-making.

[0053] Step S5: Adjust the parameters of the probability distribution model according to the long-term reliability assessment index to obtain an optimized crack network behavior prediction model.

[0054] In step S5, the optimized crack network behavior prediction model is obtained, including:

[0055] The long-term reliability assessment index is compared with the preset reliability benchmark to calculate the stability deviation value. It is then determined whether the stability deviation value is less than the preset threshold. If not, the key parameters in the Monte Carlo simulation are adjusted, including the random seed, crack density, length distribution, and tilt angle distribution. Based on the adjusted key parameters, the probability distribution model of the crack network is regenerated, and steps S2-S4 are repeated to obtain a new long-term reliability assessment index and calculate a new stability deviation value until the stability deviation value is less than the preset threshold. The optimized probability distribution model is then defined as the optimized crack network behavior prediction model.

[0056] Specifically, the system first compares long-term reliability assessment indicators with a preset reliability benchmark. The stability deviation value is obtained by calculating the root mean square of the difference between the two. This benchmark value, derived from historical engineering data, represents the system's reliability level under ideal conditions. During deviation calculation, the system considers both the instantaneous values ​​and trends of the indicators to ensure a comprehensive assessment. When the stability deviation value is not less than a preset threshold (e.g., 0.05), the system initiates a parameter optimization program. First, it adjusts the random seed in the Monte Carlo simulation, exploring new parameter spaces by changing the random number generation sequence. Simultaneously, based on the magnitude and direction of the deviation value, it dynamically adjusts the ranges of key parameters such as crack density, length distribution, and dip angle distribution. For example, if the deviation value indicates that the model prediction is too conservative, the system will appropriately expand the upper limit of the crack length distribution, extending the original 5-10 meter range to 4-12 meters; if the prediction is insufficient, the range of the density parameter will be adjusted accordingly. Based on the adjusted parameters, the system re-executes the Monte Carlo simulation to generate a new crack network probability distribution model. This model is then input into the complete process from steps S2 to S4 to recalculate the long-term reliability assessment index and update the stability deviation value. This iterative process employs an adaptive algorithm, dynamically optimizing the parameter adjustment range based on the previous improvement effect to ensure convergence efficiency. The iteration continues until the stability deviation value is less than a preset threshold. At this point, the system performs multi-dimensional verification of the final probability distribution model, including goodness-of-fit tests with historical data and assessments of predictive ability for unknown data. After passing all verifications, the probability distribution model is formally defined as the optimized crack network behavior prediction model, and the relevant parameter configurations are locked for use in predictive analysis of the long-term behavior of underground systems. By establishing a closed-loop optimization mechanism based on quantitative indicators, continuous improvement of the crack network model is achieved; the adaptive parameter adjustment strategy ensures the efficiency and stability of the optimization process; and the final optimized model significantly improves the accuracy of long-term predictions, providing a reliable basis for risk assessment in underground engineering.

[0057] Example 2:

[0058] The above describes a method for physical modeling of rock fractures in an embodiment of this application. The following describes a system for physical modeling of rock fractures in an embodiment of this application. Please refer to [link to relevant documentation]. Figure 3 One embodiment of a physical modeling system for rock fractures in this application includes:

[0059] The acquisition module is used to acquire the geometric parameters and spatial distribution data of the rock fracture set and generate a probability distribution model of the fracture network.

[0060] The determination module is used to obtain the interaction parameters between crack groups based on the probability distribution model, and to determine the statistical distribution characteristics of the cooperative sliding of the crack group;

[0061] The computation module is used to augment the statistical distribution characteristics and generate a crack dynamic behavior dataset. Based on the crack dynamic behavior dataset, a spatiotemporal evolution simulation of crack group cooperative sliding is constructed, and the probability of the occurrence of cooperative effects is calculated.

[0062] The evaluation module is used to obtain local stress field variation data based on the probability of occurrence and calculate the critical point prediction value of rock mass instability. Based on the critical point prediction value, combined with seepage characteristics and mechanical response parameters, the long-term reliability evaluation index of the underground system is determined.

[0063] The optimization module is used to adjust the parameters of the probability distribution model based on long-term reliability assessment indicators to obtain an optimized crack network behavior prediction model.

[0064] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described method for physical modeling of rock fractures.

[0065] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0066] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0067] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method of physical modeling of rock fractures, characterized by, The method comprises the following steps: Step S1, obtaining the geometric parameters and spatial distribution data of the rock fracture set, and generating a probability distribution model of the fracture network; Step S2, obtaining the interaction parameters between fracture groups according to the probability distribution model, and determining the statistical distribution characteristics of the cooperative sliding of the fracture groups; Step S3, data augmentation is performed on the statistical distribution characteristics, and a fracture dynamic behavior data set is generated, a space-time evolution simulation of the cooperative sliding of the fracture groups is constructed based on the fracture dynamic behavior data set, and the occurrence probability of the cooperative effect is calculated; Step S4, based on the occurrence probability, the variation data of the local stress field is obtained, and the critical point prediction value of the rock mass instability is calculated, and according to the critical point prediction value, the long-term reliability evaluation index of the underground system is determined in combination with the seepage characteristics and the mechanical response parameters; Step S5, the parameters of the probability distribution model are adjusted according to the long-term reliability evaluation index, and an optimized fracture network behavior prediction model is obtained.

2. The method of claim 1, wherein, In step S1, the probability distribution model of the fracture network is generated, comprising: Obtain the geometric parameters of the fracture set through field detection, including length, width and inclination data, and obtain the spatial distribution data of the fracture set through drilling, including spatial distribution coordinates and density data; According to the geometric parameters and the spatial distribution data, a plurality of random field scenes are generated by Monte Carlo simulation; For the plurality of random field scenes, the spatial connectivity and geometric distribution probability of the fracture network are calculated; According to the spatial connectivity and geometric distribution probability, the probability distribution model of the fracture network is constructed, wherein the probability distribution model represents the distribution rule and geometric characteristics of the fracture set in space.

3. The method of claim 1, wherein, In step S2, the statistical distribution characteristics of the cooperative sliding of the fracture groups are determined, comprising: Extract the spatial proximity and geometric correlation parameters between the fracture groups from the probability distribution model; According to the spatial proximity and the geometric correlation parameters, the mechanical interaction parameters between the fracture groups are calculated; The random forest algorithm is used to classify the mechanical interaction parameters to determine the mode of cooperative sliding of the fracture groups; For the mode of cooperative sliding of the fracture groups, the distribution characteristics of the sliding rate and the sliding direction are calculated, which are defined as the statistical distribution characteristics of the cooperative sliding of the fracture groups.

4. The method of claim 1, wherein, In step S3, the statistical distribution characteristics are augmented, and a fracture dynamic behavior data set is generated, comprising: Extract the sliding rate and sliding direction from the statistical distribution characteristics, and calculate the mean and variance of the sliding rate; If the mean or variance of the sliding rate exceeds a preset threshold, the statistical distribution characteristics are expanded by a limited data interpolation method; According to the expanded statistical distribution characteristics, an augmented sliding rate data set is generated; For the augmented sliding rate data set, a fracture dynamic behavior data set is constructed in combination with the geometric distribution probability in the probability distribution model.

5. The method of claim 4, wherein, In step S3, the occurrence probability of the cooperative effect is calculated, comprising: Extract the time series data of the sliding rate and the sliding direction from the fracture dynamic behavior data set; According to the time series data, a dynamic evolution model of the cooperative sliding of the fracture groups is constructed; The dynamic evolution model is used to simulate sliding behaviors of the fracture groups in time and space. The occurrence probability of the synergistic effect between the fracture groups is calculated according to the sliding behaviors.

6. The method of claim 1, wherein, The critical point prediction value of the rock mass instability is calculated in step S4, including: A high-probability sliding area is extracted from the occurrence probability of the synergistic effect between the fracture groups, and the high-probability sliding area is an area with an occurrence probability greater than a preset threshold value; The stress distribution change of the local stress field is calculated for the high-probability sliding area; An influence parameter of the fracture group sliding on the stability of the rock mass is determined according to the stress distribution change; A prediction model of the rock mass instability is constructed through the influence parameter; The critical point prediction value of the rock mass instability is output according to the prediction model.

7. The method of claim 6, wherein, The long-term reliability evaluation index of the underground system is determined in step S4, including: A stress threshold value of the rock mass instability is extracted from the critical point prediction value; Seepage characteristic parameters of the underground system are obtained according to the stress threshold value, wherein the seepage characteristic parameters include permeability and pore pressure; A mechanical-seepage coupling model of the underground system is constructed in combination with the seepage characteristic parameters and the mechanical response parameters; The long-term stability index of the underground system is calculated through the mechanical-seepage coupling model; The long-term reliability evaluation index of the underground system is generated according to the long-term stability index.

8. The method of claim 1, wherein, The optimized fracture network behavior prediction model is obtained in step S5, including: The long-term reliability evaluation index is compared with a preset reliability benchmark to calculate a stability deviation value; It is judged whether the stability deviation value is less than a preset threshold value, and if not, the key parameters in the Monte Carlo simulation are adjusted, including a random seed, a fracture density, a length distribution, and an inclination distribution; According to the adjusted key parameters, a probability distribution model of the fracture network is regenerated, and steps S2-S4 are repeatedly executed to obtain a new long-term reliability evaluation index and calculate a new stability deviation value, until the stability deviation value is less than the preset threshold value, and the optimized probability distribution model is defined as the optimized fracture network behavior prediction model.

9. A rock fracture physical modeling system for implementing a rock fracture physical modeling method according to any one of claims 1-8, characterized by, The system includes: An acquisition module is configured to acquire geometric parameters and spatial distribution data of a rock fracture set, and generate a probability distribution model of a fracture network; A determination module is configured to acquire interaction parameters between fracture groups according to the probability distribution model, and determine statistical distribution characteristics of cooperative sliding of the fracture groups; A calculation module is configured to perform data enhancement on the statistical distribution characteristics, generate a fracture dynamic behavior data set, construct a time-space evolution simulation of cooperative sliding of the fracture groups based on the fracture dynamic behavior data set, and calculate an occurrence probability of a synergistic effect; An evaluation module is configured to acquire variation data of a local stress field based on the occurrence probability, calculate a critical point prediction value of rock mass instability, and determine a long-term reliability evaluation index of an underground system according to the critical point prediction value in combination with seepage characteristics and mechanical response parameters; An optimization module is configured to adjust parameters of the probability distribution model according to the long-term reliability evaluation index, and obtain an optimized fracture network behavior prediction model.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program, when executed by the processor, implements the method of any one of claims 1 to 8.

Images