Rock creep model construction method and system based on cyclic water flooding-stress coupling

Abstract

The application provides a rock creep model construction method and system based on cyclic water flooding-stress coupling, and relates to the technical field of rock mechanics and engineering.The initial physical and mechanical properties of a rock sample are tested through a permeation experiment and a stress permeability experiment to obtain water permeability and stress permeability, a covariance matrix of characteristic parameters is obtained by using principal component analysis, the principal components related to water permeability and stress permeability are screened out, rock fracture sensitivity coefficients and stress sensitivity coefficients are calculated, a water migration equation is constructed based on Darcy's law, and is coupled with the sensitivity coefficients to form a rock creep relationship and construct a rock creep prediction model, clustering analysis is performed by using historical data to obtain clustering intervals of different risk levels, and the risk level of the rock creep prediction value is determined.The application realizes accurate prediction of the creep behavior of rocks in a complex environment by coupling water permeability and stress permeability.

Description

Technical Field

[0001] This invention relates to the field of rock mechanics and engineering technology, specifically to a method and system for constructing a rock creep model based on cyclic water flooding-stress coupling. Background Technology

[0002] In the field of rock engineering, the stability and strength of rocks are affected by a variety of factors, among which the circulation of water and the stress state of rocks are two key factors. The presence of water not only changes the physical properties of rocks but also affects their mechanical behavior. For example, water infiltration can lead to changes in the pore pressure of rocks, thereby affecting the stress distribution and deformation characteristics of rocks. At the same time, the way water flows can cause rocks to creep or decrease in strength under different environmental conditions, which is particularly important in engineering projects such as tunnel construction, mining, and infrastructure construction. However, how to accurately construct a creep model of rocks under circulating water and stress to predict the long-term stability of rocks remains a major challenge, and how to assess the interaction between water circulation and stress state still needs further investigation.

[0003] In existing technologies, the construction of rock creep models often treats water infiltration and rock stress as independent factors, failing to effectively couple their mutual influence. This results in the inability to accurately predict the creep behavior of rocks in complex environments in practical engineering, thus affecting the accuracy of the model in predicting rock creep. Furthermore, traditional rock creep models rely heavily on simple experimental data and lack systematic data analysis, such as feature analysis and sensitivity assessment of experimental data, leading to low model accuracy. Existing technologies also lack effective clustering and classification methods for handling multivariate factors, failing to provide targeted risk assessments for different types of rocks, thereby limiting their application in practical engineering.

[0004] Therefore, it is necessary to provide a method and system for constructing a rock creep model based on cyclic water flooding-stress coupling to solve the aforementioned problem.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for constructing a rock creep model based on cyclic water flooding-stress coupling, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for constructing a rock creep model based on cyclic water flooding-stress coupling includes the following steps:

[0009] Step 1: Test the initial physical and mechanical properties of the rock sample to obtain relevant characteristic parameters. Perform permeability test and stress permeability test on the rock sample to determine the water permeability and stress permeability of the rock sample. Preprocess the relevant characteristic parameters of the rock sample, including normalization of the relevant characteristic parameters.

[0010] Step 2: Obtain the covariance matrix between relevant characteristic parameters based on principal component analysis, solve the eigenvalues ​​of the covariance matrix to screen out the principal component matrix related to water permeability and stress permeability, and perform sensitivity analysis on the principal component matrix to determine the key characteristic parameters related to water permeability and stress permeability. Calculate the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample based on the relevant key characteristic parameters.

[0011] Step 3: Construct a water migration equation based on Darcy's law, and couple the water migration equation with the calculated rock crack sensitivity coefficient and stress sensitivity coefficient to establish a rock creep relationship, so as to obtain the creep of the rock under stress and water cycle over time.

[0012] Step 4: Construct and train a rock creep prediction model with rock crack sensitivity coefficient and stress sensitivity coefficient as model input and rock creep variable as model output, and optimize the model based on the established loss function;

[0013] Step 5: Based on historical data, cluster analysis is performed on the creep variables measured in rock experiments to obtain cluster intervals for different risk levels of rock creep. The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the predicted value of rock creep variables. The distance between the predicted value of rock creep variables and the center value of different risk level intervals is calculated to determine the final risk level of rock creep.

[0014] Furthermore, the water permeability and stress permeability of the rock samples were determined, and the relevant characteristic parameters of the rock samples were preprocessed using the following method:

[0015] Based on the constant head method, a rock sample is placed in a permeameter, maintaining a constant head, and the water flow rate through the rock sample per unit time is measured. The thickness and cross-sectional area of ​​the rock sample are obtained, and the water permeability of the rock sample is calculated by combining the water flow time and head difference. The formula used is as follows:

[0016] ;

[0017] in, This indicates the water permeability of the rock sample. The flow rate of water passing through the rock sample per unit time is denoted as . The thickness of the rock sample. Let be the cross-sectional area of ​​the rock sample. For the time it takes for the water to flow, Due to head difference;

[0018] The stress permeability experiment was conducted in a triaxial testing apparatus. The stress permeability was determined by combining stress loading and seepage measurement. The sample was placed in the triaxial testing apparatus and confining pressure was applied. The confining pressure was gradually increased, and the water flow rate was measured through the permeation holes simultaneously. The water head before and after stress application was obtained. The formula used to calculate the stress permeability of the rock sample is as follows:

[0019] ;

[0020] in, Indicates the stress permeability of the rock sample. The head of water before stress is applied. The head of water after stress is applied;

[0021] The relevant characteristic parameters of the obtained rock samples are normalized to ensure that they are within the same dimension. These relevant characteristic parameters include, but are not limited to, the rock's density, porosity, water content, compressive strength, elastic modulus, and Poisson's ratio. The normalization formula is as follows:

[0022] ;

[0023] in, This represents the normalized parameter value. Indicates the original parameter value. This is the minimum value of the parameter across all samples. This is the maximum value of the parameter across all samples.

[0024] Furthermore, the covariance matrix among the relevant characteristic parameters is obtained. By solving the eigenvalues ​​of the covariance matrix, principal component matrices related to water permeability and stress permeability are selected from the relevant characteristic parameters of the rock sample. The method used is as follows:

[0025] After normalization, the relevant feature parameters are all numerical values. Let's assume... One sample and The dataset is standardized using several feature parameters, and the covariance matrix of the dataset is calculated.

[0026] ;

[0027] in, This represents the standardized data matrix. For the sample size, Let the covariance matrix be , and its dimension be . , This represents the new data matrix after standardization. Represents the transpose of the new data matrix;

[0028] For covariance matrix The method used to perform eigenvalue decomposition and solve for the eigenvalues ​​is as follows:

[0029] ;

[0030] in, Represents the eigenvector. Indicates the corresponding eigenvalue;

[0031] Sort the obtained eigenvalues ​​in descending order to obtain the sorted eigenvalues. and the corresponding feature vector , the original standardized data Multiplying the result by the sorted eigenvectors yields the principal component matrix. Principal component matrix Each column represents a principal component.

[0032] Furthermore, sensitivity analysis was performed on the principal component matrix to determine the key characteristic parameters related to water permeability and stress permeability, based on the following formula:

[0033] Water permeability and stress permeability are respectively and The Pearson correlation coefficients between the principal component matrix and water permeability and stress permeability were calculated based on the characteristic correlation method formula. The formula used is as follows:

[0034] ;

[0035] in, Indicates the first Pearson correlation coefficient, Indicates the index of the Pearson correlation coefficient. When, it represents the Pearson correlation coefficient between water permeability and the principal component matrix. When, represents the Pearson correlation coefficient between stress permeability and the principal component matrix. Represents the th principal component matrix The values ​​of the principal components, This represents the mean of all principal components in the principal component matrix. hour, Indicates the first The value of water permeability. Indicates the first The average value of water permeability. hour, Indicates the first The value of stress permeability, Indicates the first The average value of stress permeability;

[0036] Based on the calculated Pearson correlation coefficient, the principal components with the highest correlation to water permeability and stress permeability are identified. For the selected principal components, the corresponding eigenvectors are examined. The larger the weight of a feature, the greater its contribution to the principal component. Based on the weight of the eigenvectors, features closely related to water permeability and stress permeability are selected as key feature parameters.

[0037] Furthermore, the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample were calculated based on the characteristic correlation method. The method used was as follows:

[0038] The rock fracture sensitivity coefficient of a rock sample is calculated using key characteristic parameters related to water permeability combined with a feature combination algorithm. These key characteristic parameters include the rock's porosity, water content, and density. Crack characteristic values ​​are calculated based on the crack growth model in the feature combination algorithm. The ratio of these crack characteristic values ​​to the change in water permeability is the rock fracture sensitivity coefficient, based on the following formula:

[0039] ;

[0040] ;

[0041] in, Indicates the characteristic value of the crack. Indicates the rock fracture sensitivity coefficient. The maximum crack characteristic value, The water content of the rock. The density of the rock, The crack growth rate, Porosity of the rock The critical porosity of the rock. This indicates the change in water permeability;

[0042] Similarly, the stress sensitivity coefficient is calculated. Key characteristic parameters related to stress permeability include the rock's compressive strength, elastic modulus, and Poisson's ratio. The stress characteristic value is calculated based on the stress growth model in the feature combination algorithm. The ratio of this stress characteristic value to the stress permeability is the stress sensitivity coefficient, based on the following formula:

[0043] ;

[0044] ;

[0045] in, Represents the stress eigenvalue. Indicates the stress sensitivity coefficient. The maximum stress characteristic value, Poisson's ratio of the rock The elastic modulus of the rock. The rate of increase of stress, The compressive strength of the rock. This represents the critical stress level of the rock. This represents the change in stress permeability.

[0046] Furthermore, a rock creep relationship was established to obtain the rock creep variation, based on the following method:

[0047] First, a water migration equation is constructed based on Darcy's law. Then, the rock crack sensitivity coefficient and stress sensitivity coefficient obtained by the feature combination algorithm are coupled with the water migration equation to establish the rock creep relationship. Finally, Darcy's law is combined with the water migration characteristics of rocks to obtain the water migration equation:

[0048] ;

[0049] in, Indicates that the rock is in Moisture content at time Where is the diffusion coefficient. Let be the Laplace operator, representing the second derivative of water content in space. Indicates in The flow rate of water passing through the cross-sectional area of ​​the rock within a given time period. Indicates the volume of the rock;

[0050] The creep behavior of rocks is represented by a time-dependent relationship, and the coupling relationship between creep strain and water migration is derived by combining the rock crack sensitivity coefficient and the stress sensitivity coefficient:

[0051] ;

[0052] ;

[0053] in, Indicates that the rock is in The total creep variable at time t, This represents the initial creep variation of the rock. This represents the creep variable that the rock undergoes as it changes over time.

[0054] Furthermore, a rock creep prediction model was constructed, and a loss function was established to train and optimize the model. The method used was as follows:

[0055] A rock creep prediction model is constructed based on a neural network structure. The rock creep prediction model includes an input layer, a hidden layer, and an output layer. The input layer is responsible for receiving the rock crack sensitivity coefficient and stress sensitivity coefficient as input to the model, where each coefficient represents an input node. The hidden layer is an intermediate layer in the neural network between the input layer and the output layer. It is responsible for feature extraction and information processing of the input data. Nonlinear activation functions are used in the hidden layer to capture complex patterns. The output layer generates the final prediction result of the model. The output layer has one node to output the predicted rock creep value.

[0056] The mean squared error is chosen as the loss function, based on the following formula:

[0057] ;

[0058] in, The mean squared error (MSE) value represents the model's error and is used to measure the difference between the predicted and actual values. For the index of the rock sample, and , Indicates the first The true value of creep variables in a rock sample Indicates the first Predicted values ​​of creep variables for a rock sample;

[0059] The constructed rock creep prediction model is trained by inputting rock crack sensitivity coefficient and stress sensitivity coefficient. The model is calculated through each layer of the neural network to generate the predicted value of the output creep variable. The difference between the predicted value and the true value is calculated using the set MSE loss function. The gradient of the loss function with respect to the model parameters is calculated through the backpropagation algorithm, and the gradient of the loss is passed from the output layer back to the input layer. The model parameters are updated using the gradient descent algorithm to minimize the loss function. The process of forward propagation, loss calculation, backpropagation and parameter update is repeated until the loss function reaches the preset convergence threshold.

[0060] Furthermore, the method used to determine the final risk level of rock creep is as follows:

[0061] K-means clustering was performed on the creep variables measured in the rock samples from historical data. K initial cluster centers were randomly selected, and each sample was assigned to the nearest cluster center. The cluster center was updated to be the mean of all samples in each cluster. The above steps were repeated until the cluster centers no longer changed. The center value of each cluster interval and the cluster category to which each sample belonged were recorded. The cluster categories included high creep risk, medium creep risk, and low creep risk.

[0062] The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the creep prediction value. Calculate its distance from each cluster center:

[0063] ;

[0064] in, Indicates the predicted value of the creep variable and the first The distance between the center values ​​of each cluster interval Indicates the first The center value of each cluster interval The index is the number of cluster intervals, and ;

[0065] Find the cluster centers with the smallest distance Determine the corresponding creep risk level.

[0066] The present invention also provides a rock creep model construction system based on cyclic water flooding-stress coupling, the rock creep model construction system being used to execute the above-described rock creep model construction method based on cyclic water flooding-stress coupling, comprising:

[0067] The initial feature acquisition module is used to test the initial physical and mechanical properties of rock samples to obtain relevant feature parameters. It performs permeability tests and stress permeability tests on the rock samples to determine the water permeability and stress permeability of the rock samples, and preprocesses the relevant feature parameters of the rock samples, including normalization of the relevant feature parameters.

[0068] The principal component analysis module obtains the covariance matrix between relevant characteristic parameters based on the principal component analysis method, solves the eigenvalues ​​of the covariance matrix to screen out the principal component matrices related to water permeability and stress permeability, performs sensitivity analysis on the principal component matrices to determine the key characteristic parameters related to water permeability and stress permeability, and calculates the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample based on the relevant key characteristic parameters.

[0069] The water migration coupling construction module constructs a water migration equation based on Darcy's law, and couples the water migration equation with the calculated rock crack sensitivity coefficient and stress sensitivity coefficient to establish a rock creep relationship, so as to obtain the creep of rock under stress and water cycle over time.

[0070] The rock creep prediction model building module is used to build and train a rock creep prediction model with rock crack sensitivity coefficient and stress sensitivity coefficient as model input and rock creep variable as model output, and optimize the model based on the established loss function.

[0071] The creep risk clustering analysis module performs clustering analysis on creep variables measured in rock experiments based on historical data to obtain creep risk clustering intervals for rocks with different risks. The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the predicted value of rock creep variables. The distance between the predicted value of rock creep variables and the center value of different risk level intervals is calculated to determine the final risk level of rock creep.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] Based on the coupling effect of circulating water flooding and stress, this invention constructs a comprehensive rock creep model that can more accurately simulate the behavior of rocks in complex environments. By combining principal component analysis and sensitivity analysis, key characteristic parameters are identified, making the prediction of rock creep under different stress and moisture conditions more accurate.

[0074] This invention uses cluster analysis to classify rock creep risks in historical data, effectively dividing cluster intervals into different risk levels. This not only enhances the applicability and flexibility of the model, but also provides targeted risk assessments for practical engineering projects. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the overall method flow of the present invention.

[0076] Figure 2 This is a schematic diagram of the system module flow of the present invention. Detailed Implementation

[0077] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0078] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0079] Example:

[0080] Please see Figure 1 A method for constructing a rock creep model based on cyclic water flooding-stress coupling includes the following steps:

[0081] Step 1: Test the initial physical and mechanical properties of the rock sample to obtain relevant characteristic parameters. Perform permeability test and stress permeability test on the rock sample to determine the water permeability and stress permeability of the rock sample. Preprocess the relevant characteristic parameters of the rock sample, including normalization of the relevant characteristic parameters.

[0082] Step 2: Obtain the covariance matrix between relevant characteristic parameters based on principal component analysis, solve the eigenvalues ​​of the covariance matrix to screen out the principal component matrix related to water permeability and stress permeability, and perform sensitivity analysis on the principal component matrix to determine the key characteristic parameters related to water permeability and stress permeability. Calculate the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample based on the relevant key characteristic parameters.

[0083] Step 3: Construct a water migration equation based on Darcy's law, and couple the water migration equation with the calculated rock crack sensitivity coefficient and stress sensitivity coefficient to establish a rock creep relationship, so as to obtain the creep of the rock under stress and water cycle over time.

[0084] Step 4: Construct and train a rock creep prediction model with rock crack sensitivity coefficient and stress sensitivity coefficient as model input and rock creep variable as model output, and optimize the model based on the established loss function;

[0085] Step 5: Based on historical data, cluster analysis is performed on the creep variables measured in rock experiments to obtain cluster intervals for different risk levels of rock creep. The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the predicted value of rock creep variables. The distance between the predicted value of rock creep variables and the center value of different risk level intervals is calculated to determine the final risk level of rock creep.

[0086] It should be noted that the permeability test and stress permeability test are set up to accurately evaluate the water migration characteristics and mechanical response of rocks under different environmental conditions. Water permeability is a key parameter for evaluating the permeability performance of rocks under the action of water flow. By using the constant head method, the water flow rate can be accurately measured under controlled head conditions, and then the water permeability of the rock sample can be calculated. Stress permeability reflects the seepage capacity of rocks under different stress conditions, and can better simulate the environmental conditions in actual engineering.

[0087] Therefore, it is necessary to define the water permeability and stress permeability of the rock samples and preprocess the relevant characteristic parameters of the rock samples. The method used is as follows:

[0088] Based on the constant head method, a rock sample is placed in a permeameter, maintaining a constant head, and the water flow rate through the rock sample per unit time is measured. The thickness and cross-sectional area of ​​the rock sample are obtained, and the water permeability of the rock sample is calculated by combining the water flow time and head difference. The formula used is as follows:

[0089] ;

[0090] in, This indicates the water permeability of the rock sample. The flow rate of water passing through the rock sample per unit time is denoted as . The thickness of the rock sample. Let be the cross-sectional area of ​​the rock sample. For the time it takes for the water to flow, Due to head difference;

[0091] The stress permeability experiment was conducted in a triaxial testing apparatus. The stress permeability was determined by combining stress loading and seepage measurement. The sample was placed in the triaxial testing apparatus and confining pressure was applied. The confining pressure was gradually increased, and the water flow rate was measured through the permeation holes simultaneously. The water head before and after stress application was obtained. The formula used to calculate the stress permeability of the rock sample is as follows:

[0092] ;

[0093] in, Indicates the stress permeability of the rock sample. The head of water before stress is applied. The head of water after stress is applied;

[0094] The relevant characteristic parameters of the obtained rock samples are normalized to ensure that they are within the same dimension. These relevant characteristic parameters include, but are not limited to, the rock's density, porosity, water content, compressive strength, elastic modulus, and Poisson's ratio. The normalization formula is as follows:

[0095] ;

[0096] in, This represents the normalized parameter value. Indicates the original parameter value. This is the minimum value of the parameter across all samples. This is the maximum value of the parameter across all samples.

[0097] It should be noted that the calculation of the covariance matrix and eigenvalue decomposition can help us extract the most important features from the high-dimensional feature space, thereby reducing the dimensionality of the data. This is crucial for subsequent model building and analysis because it reduces the interference of redundant features, making the model more concise and efficient. Furthermore, by sorting the eigenvalues, we can identify key feature parameters related to water permeability and stress permeability. These key features have a significant impact on the permeability behavior of rocks, which helps improve the prediction accuracy and reliability of the model.

[0098] Since the weights of feature parameters are not identical after normalization, standardization ensures that all features have the same weights, thus avoiding excessive influence of certain features on the covariance matrix and subsequent calculations. Calculating the covariance matrix quantifies the correlation between various feature parameters, helping to reveal their intrinsic relationships and mutual influences, providing a data foundation for selecting key features. Furthermore, eigenvalue decomposition reveals the structural characteristics of the covariance matrix; the magnitude of the eigenvalues ​​reflects the explanatory power of each principal component for data variation. By multiplying the original standardized data with the sorted feature vectors, the resulting principal component matrix effectively integrates information from multiple features, forming new comprehensive features. These principal components can be more effectively used in subsequent creep prediction models, making the models more robust.

[0099] Therefore, it is necessary to obtain the covariance matrix among the relevant characteristic parameters. By solving the eigenvalues ​​of the covariance matrix, the principal component matrix related to water permeability and stress permeability can be screened from the relevant characteristic parameters of the rock sample. The method used is as follows:

[0100] After normalization, the relevant feature parameters are all numerical values. Let's assume... One sample and The dataset is standardized using several feature parameters, and the covariance matrix of the dataset is calculated.

[0101] ;

[0102] in, This represents the standardized data matrix. For the sample size, Let the covariance matrix be , and its dimension be . , This represents the new data matrix after standardization. Represents the transpose of the new data matrix;

[0103] For covariance matrix The method used to perform eigenvalue decomposition and solve for the eigenvalues ​​is as follows:

[0104] ;

[0105] in, Represents the eigenvector. Indicates the corresponding eigenvalue;

[0106] Sort the obtained eigenvalues ​​in descending order to obtain the sorted eigenvalues. and the corresponding feature vector , the original standardized data Multiplying the result by the sorted eigenvectors yields the principal component matrix. Principal component matrix Each column represents a principal component.

[0107] It is important to note that sensitivity analysis of the principal component matrix is ​​crucial for identifying key characteristic parameters related to water permeability and stress permeability. This process, through calculating the Pearson correlation coefficient, clearly identifies which principal components have a significant impact on explaining the target variable. This analysis not only optimizes feature selection, improving the efficiency and accuracy of the model, but also enhances its interpretability. It allows researchers to clearly understand the contribution of each characteristic factor to water and stress permeability, thus providing effective decision support for practical applications. This approach can better advance the research and engineering practice of rock permeability.

[0108] Therefore, sensitivity analysis of the principal component matrix is ​​required to determine the key characteristic parameters related to water permeability and stress permeability. The formula used is as follows:

[0109] Water permeability and stress permeability are respectively and The Pearson correlation coefficients between the principal component matrix and water permeability and stress permeability were calculated based on the characteristic correlation method formula. The formula used is as follows:

[0110] ;

[0111] in, Indicates the first Pearson correlation coefficient, Indicates the index of the Pearson correlation coefficient. When, it represents the Pearson correlation coefficient between water permeability and the principal component matrix. When, represents the Pearson correlation coefficient between stress permeability and the principal component matrix. Represents the th principal component matrix The values ​​of the principal components, This represents the mean of all principal components in the principal component matrix. hour, Indicates the first The value of water permeability. Indicates the first The average value of water permeability. hour, Indicates the first The value of stress permeability, Indicates the first The average value of stress permeability;

[0112] Based on the calculated Pearson correlation coefficient, the principal components with the highest correlation to water permeability and stress permeability are identified. For the selected principal components, the corresponding eigenvectors are examined. The larger the weight of a feature, the greater its contribution to the principal component. Based on the weight of the eigenvectors, features closely related to water permeability and stress permeability are selected as key feature parameters.

[0113] It is worth noting that calculating the crack sensitivity coefficient and stress sensitivity coefficient of rock samples based on the characteristic correlation method is of great significance. This is because by integrating key characteristic parameters related to water permeability and stress permeability, this method can reveal in depth the crack development and stress response behavior of rocks under different external conditions. This analysis provides a quantitative basis for predicting the performance of rock materials in engineering applications, enabling engineers to better assess the stability and safety of rock structures, thereby optimizing design and construction plans, effectively preventing potential engineering risks, and ensuring the smooth progress of projects.

[0114] Therefore, the method used to calculate the rock fracture sensitivity coefficient and stress sensitivity coefficient of rock samples based on the characteristic correlation method is as follows:

[0115] The rock fracture sensitivity coefficient of a rock sample is calculated using key characteristic parameters related to water permeability combined with a feature combination algorithm. These key characteristic parameters include the rock's porosity, water content, and density. Crack characteristic values ​​are calculated based on the crack growth model in the feature combination algorithm. The ratio of these crack characteristic values ​​to the change in water permeability is the rock fracture sensitivity coefficient, based on the following formula:

[0116] ;

[0117] ;

[0118] in, Indicates the characteristic value of the crack. Indicates the rock fracture sensitivity coefficient. The maximum crack characteristic value, The water content of the rock. The density of the rock, The crack growth rate, Porosity of the rock The critical porosity of the rock. This represents the change in moisture permeability; in the formula for calculating the characteristic value of cracks above, moisture content... This refers to the proportion of water in a rock, which directly affects the rock's mechanical properties and crack propagation behavior. Higher water content significantly reduces the rock's strength, making cracks more likely to propagate. Increase, crack characteristic value It will increase; the density of the rock This is a fundamental physical property of rocks; the higher the density, the stronger the rock. This indicates that higher rock density inhibits crack propagation, making the rock more stable. Increase, crack characteristic value This will reduce the rate of crack growth. It reflects the crack propagation rate under changing external conditions. It represents the crack propagation speed and intensity under given porosity conditions, as... As the exponent increases, the negative value of the exponent term increases, which reduces the overall exponent value, thus making the larger exponent appear smaller. This leads to a smaller exponential term, which in turn reduces the crack characteristic value. This indicates that a higher crack growth rate means the crack is more sensitive to changes in porosity; porosity It reflects the density of the pore structure inside the rock, which affects the rock's permeability and mechanical properties. An increase means higher porosity, leading to a larger difference between it and the critical porosity, which in turn increases the negative value of the exponential term, thus reducing the overall porosity. This indicates that higher porosity usually represents a more fragile rock structure, making it more prone to cracking. Therefore, in the formula, this is reflected in a reduction of the crack characteristic value. In summary... The higher the value, the more active the crack propagation in the rock, and the greater the potential risk.

[0119] In the formula for calculating the rock fracture sensitivity coefficient above, The larger, The larger the value, the more sensitive the rock is to changes in external conditions. This means that under the same environmental or load variations, the crack propagation rate is greater, demonstrating the rock's fragile nature. The rock crack sensitivity coefficient... The larger the value, the stronger the rock's response to crack propagation when external conditions change, indicating increased rock fragility and potential risk.

[0120] Similarly, the stress sensitivity coefficient is calculated. Key characteristic parameters related to stress permeability include the rock's compressive strength, elastic modulus, and Poisson's ratio. The stress characteristic value is calculated based on the stress growth model in the feature combination algorithm. The ratio of this stress characteristic value to the stress permeability is the stress sensitivity coefficient, based on the following formula:

[0121] ;

[0122] ;

[0123] in, Represents the stress eigenvalue. Indicates the stress sensitivity coefficient. The maximum stress characteristic value, Poisson's ratio of the rock The elastic modulus of the rock. The rate of increase of stress, The compressive strength of the rock. This represents the critical stress level of the rock. This represents the change in stress permeability; in the formula for calculating the stress characteristic value above, Poisson's ratio... It reflects the deformation characteristics of rocks under axial stress. The larger the Poisson's ratio, the stronger the rock's ability to deform in the lateral direction, indicating that the rock's ductility will be enhanced. Increase; while elastic modulus This reflects the stiffness of the rock. The greater the stiffness, the stronger its resistance to deformation. Therefore, under the same stress, the rock will have less strain, resulting in a higher stress characteristic value. Reduce; Stress growth rate An increase indicates a more significant change in stress experienced by the rock over a short period, leading to a higher stress characteristic value. Increase; compressive strength of rock Increase or even exceed the critical stress level hour, An increase reflects the enhanced ability of the rock to withstand greater stress, while the rock's compressive strength... Reduced to below the critical stress level time, A decrease in density means that the rock becomes more brittle and more prone to creep failure.

[0124] The above calculation of stress sensitivity coefficient In the formula, The larger the value, the higher the stress sensitivity coefficient. The larger the value, the stronger the rock's response to stress changes when subjected to external stress, indicating a higher stress characteristic value. As the size increases, changes in microstructure or breakage become more likely, while An increase means that the rock is more sensitive to changes in external stress, that is, under the same stress change, the possibility of crack or defect propagation increases.

[0125] It is important to note that establishing a rock creep equation to obtain the creep variation of rocks is crucial because this model can effectively describe the mechanical behavior of rocks under long-term loads and environmental changes. By coupling the moisture migration equation with the rock crack sensitivity coefficient and stress sensitivity coefficient, we can gain a deeper understanding of the impact of moisture on rock creep characteristics, thus providing more accurate predictions and analyses. This coupling relationship helps assess the stability of rocks under different moisture conditions, thereby providing a scientific basis for engineering design, construction, and maintenance, ensuring the safety and long-term reliability of underground engineering projects. It has significant practical value, especially in applications such as mining, tunnel construction, and groundwater management.

[0126] Therefore, it is necessary to establish a rock creep relationship to obtain the creep variation of rocks. The method used is as follows:

[0127] First, a water migration equation is constructed based on Darcy's law. Then, the rock crack sensitivity coefficient and stress sensitivity coefficient obtained by the feature combination algorithm are coupled with the water migration equation to establish the rock creep relationship. Finally, Darcy's law is combined with the water migration characteristics of rocks to obtain the water migration equation:

[0128] ;

[0129] in, Indicates that the rock is in Moisture content at time Where is the diffusion coefficient. Let be the Laplace operator, representing the second derivative of water content in space. Indicates in The flow rate of water passing through the cross-sectional area of ​​the rock within a given time period. Indicates the volume of the rock;

[0130] The creep behavior of rocks is represented by a time-dependent relationship, and the coupling relationship between creep strain and water migration is derived by combining the rock crack sensitivity coefficient and the stress sensitivity coefficient:

[0131] ;

[0132] ;

[0133] in, Indicates that the rock is in The total creep variable at time t, This represents the initial creep variation of the rock. This represents the creep variable of rocks over time; in the above calculations, the rocks in... In the formula for the total creep variable at time t, Stress sensitivity coefficient The rock crack sensitivity coefficient reflects the rock's sensitivity to external stress; the higher the stress sensitivity, the greater the creep deformation produced by the rock under stress changes. This indicates the rock's sensitivity to crack propagation. The larger the crack, the more prone the rock is to crack propagation. Crack propagation leads to a decrease in rock strength, resulting in greater creep deformation. An increase in water content over time will lead to a decrease in the strength of the rock. Total creep variable at time step Increased water content typically leads to higher pore water pressure in rocks, thereby reducing their effective stress. This often results in decreased rock strength because the presence of water reduces friction and cohesion between rock particles, making the rock more susceptible to deformation or failure under external loads. Furthermore, as the water content increases, water seeps into the microcracks and pores, further expanding existing cracks and promoting the transition from brittle fracture to plastic deformation. As cracks expand, the rock's load-bearing capacity decreases, leading to a decline in strength and an increase in creep.

[0134] It is important to note that constructing a rock creep prediction model and training and optimizing it using a loss function is crucial because this process enables accurate prediction of the long-term mechanical behavior of rocks. By leveraging the powerful characteristics of neural networks, the model can effectively handle and analyze complex nonlinear relationships such as rock crack sensitivity coefficients and stress sensitivity coefficients, thereby capturing the creep characteristics of rocks under different environmental conditions. This accurate predictive capability is essential for the design, monitoring, and maintenance of rock engineering projects, effectively preventing potential engineering risks and safety hazards. Furthermore, by optimizing the loss function, the predictive accuracy of the model can be continuously improved, ensuring its reliability in practical applications, thus providing a scientific basis for mining, civil engineering, and geological engineering, and promoting sustainable development.

[0135] Therefore, it is necessary to construct a rock creep prediction model and establish a loss function to train and optimize the model. The method used is as follows:

[0136] A rock creep prediction model is constructed based on a neural network structure. The rock creep prediction model includes an input layer, a hidden layer, and an output layer. The input layer is responsible for receiving the rock crack sensitivity coefficient and stress sensitivity coefficient as input to the model, where each coefficient represents an input node. The hidden layer is an intermediate layer in the neural network between the input layer and the output layer. It is responsible for feature extraction and information processing of the input data. Nonlinear activation functions are used in the hidden layer to capture complex patterns. The output layer generates the final prediction result of the model. The output layer has one node to output the predicted rock creep value.

[0137] The mean squared error is chosen as the loss function, based on the following formula:

[0138] ;

[0139] in, The mean squared error (MSE) value represents the model's error and is used to measure the difference between the predicted and actual values. For the index of the rock sample, and , Indicates the first The true value of creep variables in a rock sample Indicates the first Predicted values ​​of creep variables for a rock sample;

[0140] The constructed rock creep prediction model is trained by inputting rock crack sensitivity coefficient and stress sensitivity coefficient. The model is calculated through each layer of the neural network to generate the predicted value of the output creep variable. The difference between the predicted value and the true value is calculated using the set MSE loss function. The gradient of the loss function with respect to the model parameters is calculated through the backpropagation algorithm, and the gradient of the loss is passed from the output layer back to the input layer. The model parameters are updated using the gradient descent algorithm to minimize the loss function. The process of forward propagation, loss calculation, backpropagation and parameter update is repeated until the loss function reaches the preset convergence threshold.

[0141] It's important to note that by utilizing K-means clustering to classify the creep variables of rock samples from historical data, different creep risk levels can be effectively identified. This allows engineers to quickly assess the potential risks of new samples. The key to this method is that it not only relies on the accumulation of historical data but also incorporates the current rock fracture sensitivity coefficient and stress sensitivity coefficient, making risk assessment more accurate and real-time. By calculating the distance between the predicted creep variable value of the rock to be predicted and each cluster center, the most representative risk level can be found, providing a reliable basis for engineering decisions. Effective risk assessment helps engineers take preventative measures during the design phase to reduce potential geological hazard risks and enables real-time monitoring and adjustments during construction and operation.

[0142] Therefore, the final risk level of rock creep needs to be determined, and the method used is as follows:

[0143] K-means clustering was performed on the creep variables measured in the rock samples from historical data. K initial cluster centers were randomly selected, and each sample was assigned to the nearest cluster center. The cluster center was updated to be the mean of all samples in each cluster. The above steps were repeated until the cluster centers no longer changed. The center value of each cluster interval and the cluster category to which each sample belonged were recorded. The cluster categories included high creep risk, medium creep risk, and low creep risk.

[0144] The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the creep prediction value. Calculate its distance from each cluster center:

[0145] ;

[0146] in, Indicates the predicted value of the creep variable and the first The distance between the center values ​​of each cluster interval Indicates the first The center value of each cluster interval The index is the number of cluster intervals, and ;

[0147] Find the cluster centers with the smallest distance Determine the corresponding creep risk level.

[0148] The present invention also provides a rock creep model construction system based on cyclic water flooding-stress coupling, the rock creep model construction system being used to execute the above-described rock creep model construction method based on cyclic water flooding-stress coupling, comprising:

[0149] The initial feature acquisition module is used to test the initial physical and mechanical properties of rock samples to obtain relevant feature parameters. It performs permeability tests and stress permeability tests on the rock samples to determine the water permeability and stress permeability of the rock samples, and preprocesses the relevant feature parameters of the rock samples, including normalization of the relevant feature parameters.

[0150] The principal component analysis module obtains the covariance matrix between relevant characteristic parameters based on the principal component analysis method, solves the eigenvalues ​​of the covariance matrix to screen out the principal component matrices related to water permeability and stress permeability, performs sensitivity analysis on the principal component matrices to determine the key characteristic parameters related to water permeability and stress permeability, and calculates the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample based on the relevant key characteristic parameters.

[0151] The water migration coupling construction module constructs a water migration equation based on Darcy's law, and couples the water migration equation with the calculated rock crack sensitivity coefficient and stress sensitivity coefficient to establish a rock creep relationship, so as to obtain the creep of rock under stress and water cycle over time.

[0152] The rock creep prediction model building module is used to build and train a rock creep prediction model with rock crack sensitivity coefficient and stress sensitivity coefficient as model input and rock creep variable as model output, and optimize the model based on the established loss function.

[0153] The creep risk clustering analysis module performs clustering analysis on creep variables measured in rock experiments based on historical data to obtain creep risk clustering intervals for rocks with different risks. The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain the predicted value of rock creep variables. The distance between the predicted value of rock creep variables and the center value of different risk level intervals is calculated to determine the final risk level of rock creep.

[0154] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0155] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0156] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0157] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for constructing a rock creep model based on cyclic water flooding-stress coupling, characterized in that, The specific steps include: Step 1: Test the initial physical and mechanical properties of the rock sample to obtain relevant characteristic parameters, conduct permeability experiments and stress permeability experiments on the rock sample to determine the moisture permeability and stress permeability of the rock sample, and preprocess the relevant characteristic parameters of the rock sample, including normalizing the relevant characteristic parameters; Step 2: Obtain the covariance matrix between the relevant characteristic parameters based on principal component analysis, solve the eigenvalues of the covariance matrix to screen out the principal component matrix related to the moisture permeability and stress permeability, and perform sensitivity analysis on the principal component matrix to determine the key characteristic parameters related to the moisture permeability and stress permeability, and calculate the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock sample based on the relevant key characteristic parameters; Step 3: Construct a moisture migration equation based on Darcy's law, and couple the moisture migration equation with the rock fracture sensitivity coefficient and stress sensitivity coefficient obtained in the calculation to establish a rock creep relationship to obtain the creep variable of the rock under the action of stress and water circulation over time; Step 4: Construct and train a rock creep prediction model with the rock fracture sensitivity coefficient and stress sensitivity coefficient as the model input and the rock creep variable as the model output, and optimize the model based on the established loss function; Step 5: Cluster the measured creep variables of the rock in the past historical data to obtain different risk rock creep risk clustering intervals, input the rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted into the optimized model to obtain the rock creep variable prediction value, and calculate the distance between the rock creep variable prediction value and the center value of different risk level intervals to determine the final risk level of the rock creep.

2. The method according to claim 1, wherein, The method for determining the moisture permeability and stress permeability of the rock sample and preprocessing the relevant characteristic parameters of the rock sample is as follows: Place the rock sample in the permeameter based on the constant head method, maintain a constant water head, and measure the water flow through the rock sample per unit time, obtain the thickness and cross-sectional area of the rock sample, and calculate the moisture permeability of the rock sample based on the water flow time and water head difference, the formula is: ； wherein, represents the moisture permeability of the rock sample, is the water flow rate through the rock sample per unit time, is the thickness of the rock sample, is the cross-sectional area of the rock sample, is the time of water flow, is the water head difference; The stress permeability experiment is conducted in a triaxial test device, the stress loading and seepage measurement are combined to determine the stress permeability, the sample is placed in the triaxial test device to apply confining pressure, the confining pressure is gradually increased, and the water flow is measured through the permeation hole, the water head before and after stress application is obtained respectively, and the formula for calculating the stress permeability of the rock sample is: ； wherein, represents the stress permeability of the rock sample, is the water head before the stress is applied, is the water head after the stress is applied; The obtained relevant characteristic parameters of the rock sample are normalized to make the relevant characteristic parameters of the rock sample in the same dimension, the relevant characteristic parameters include but are not limited to the density, porosity, water content, compressive strength, elastic modulus and Poisson's ratio of the rock, and the formula for normalization is: ； wherein, denotes the normalized parameter value, denotes the original parameter value, is the minimum value of the parameter over all samples, is the maximum value of the parameter over all samples.

3. The method according to claim 2, wherein, The covariance matrix between the relevant characteristic parameters is obtained, and the principal component matrix related to the moisture permeability and stress permeability is screened out from the relevant characteristic parameters of the rock sample by solving the eigenvalues of the covariance matrix, and the method is: The data of the normalized correlation characteristic parameters, the value of each parameter is a numerical value, provided with samples and characteristic parameters, the data is standardized, and a covariance matrix of the data set is calculated: ； wherein, denotes the standardized data matrix, is the number of samples, denotes the covariance matrix, which has dimensions , denotes the new standardized data matrix, denotes the transpose of the new data matrix; Eigenvalue decomposition of the covariance matrix is performed, the method for solving the eigenvalues being: ； wherein denotes the eigenvector, denotes the corresponding eigenvalue; The characteristic values obtained by solving are sorted from large to small to obtain sorted characteristic values and corresponding eigenvectors The original normalized data is multiplied by the sorted eigenvectors to obtain a principal component matrix Each column of the principal component matrix represents a principal component.

4. The method according to claim 3, wherein, The sensitivity analysis is performed on the principal component matrix to determine the key characteristic parameters related to the water permeability and stress permeability, and the formula is: The water permeability and stress permeability are respectively and The Pearson correlation coefficients between the principal component matrix and the water permeability and the stress permeability are calculated based on the characteristic correlation method formula, and the formula is: ； in, Indicates the first Pearson correlation coefficient, Indicates the index of the Pearson correlation coefficient. When, it represents the Pearson correlation coefficient between water permeability and the principal component matrix. When, represents the Pearson correlation coefficient between stress permeability and the principal component matrix. Represents the th principal component matrix The values ​​of the principal components, This represents the mean of all principal components in the principal component matrix. hour, Indicates the first The value of water permeability. Indicates the first The average value of water permeability. hour, Indicates the first The value of stress permeability, Indicates the first The average value of stress permeability; Based on the calculated Pearson correlation coefficient, the principal component with the highest correlation with the water permeability and stress permeability is found, and for the selected principal component, the corresponding eigenvector is viewed, and the larger the weight of the feature, the greater the contribution to the principal component. Based on the weight of the eigenvector, the features closely related to the water permeability and stress permeability are selected as the key characteristic parameters.

5. The method according to claim 4, wherein, The rock cracking sensitivity coefficient and stress sensitivity coefficient of the rock sample are calculated based on the feature correlation method, and the method is: The rock cracking sensitivity coefficient of the rock sample is calculated using the key characteristic parameters related to the water permeability and the feature combination algorithm, wherein the key characteristic parameters related to the water permeability include the porosity, water content and density of the rock. The crack characteristic value is calculated based on the crack growth model in the feature combination algorithm, and the ratio of the crack characteristic value to the water permeability change is the rock cracking sensitivity coefficient. The formula is: ； ； wherein, represents a crack characteristic value, represents a rock crack sensitivity coefficient, is a maximum crack characteristic value, is a water content of the rock, is a density of the rock, is a crack growth rate, is a porosity of the rock, is a critical porosity of the rock, represents a water permeability change amount; Similarly, the stress sensitivity coefficient is calculated, wherein the key characteristic parameters related to the stress permeability include the compressive strength, elastic modulus and Poisson's ratio of the rock. Similarly, the stress characteristic value is calculated based on the stress growth model in the feature combination algorithm, and the ratio of the stress characteristic value to the stress permeability is the stress sensitivity coefficient. The formula is: ； ； wherein, denotes a stress eigenvalue, denotes a stress sensitivity coefficient, is a maximum stress eigenvalue, is a Poisson's ratio of the rock, is an elastic modulus of the rock, is a stress growth rate, is a compressive strength of the rock, is a critical stress level of the rock, denotes a stress permeability change amount.

6. The method according to claim 5, wherein, The rock creep relationship is established to obtain the creep of the rock, and the method is: Based on Darcy's law, the water migration equation is constructed, and the rock cracking sensitivity coefficient and stress sensitivity coefficient obtained based on the feature combination algorithm are coupled with the water migration equation to establish the rock creep relationship. Darcy's law is combined with the water migration characteristics of the rock to obtain the water migration equation: ； in, Indicates that the rock is in Moisture content at time The diffusion coefficient is... Let be the Laplace operator, representing the second derivative of water content in space. Indicates in The flow rate of water passing through the cross-sectional area of ​​the rock within a given time period. Indicates the volume of the rock; The creep behavior of the rock is represented by a time-dependent relationship, and the relationship between the creep strain and water migration is derived by combining the rock cracking sensitivity coefficient and the stress sensitivity coefficient: ； ； wherein, represents the total creep amount of the rock at represents the total creep amount of the rock at represents the initial creep amount of the rock, represents the creep amount of the rock over time.

7. The method according to claim 6, wherein, The rock creep prediction model is constructed, and the loss function is established to train and optimize the model, and the method is: The rock creep prediction model is constructed based on the neural network structure, which includes the input layer, hidden layer and output layer. The input layer is responsible for receiving the rock cracking sensitivity coefficient and stress sensitivity coefficient as the input of the model, wherein each coefficient represents an input node. The hidden layer is an intermediate layer between the input layer and the output layer in the neural network, which is responsible for feature extraction and information processing of the input data. Nonlinear activation functions are used in the hidden layer to capture complex patterns. The output layer generates the final prediction result of the model, and the output layer has one node for outputting the predicted rock creep. The mean square error is selected as the loss function, and the formula is: ； wherein, the mean square error value of the model, for measuring the difference between the predicted value and the true value, is an index of the rock sample, and , represents the true value of the creep variable of the i-th rock sample, represents the predicted value of the creep variable of the i-th rock sample, represents the predicted value of the creep variable of the i-th rock sample, represents the predicted value of the creep variable of the i-th rock sample. The constructed rock creep prediction model is trained, the rock crack sensitivity coefficient and the stress sensitivity coefficient are input, calculation is performed through each layer of the neural network, a predicted value of the output creep variable is generated, a preset MSE loss function is used to calculate the difference between the predicted value and the true value, the gradient of the loss function with respect to the model parameters is calculated through the back propagation algorithm, the gradient of the loss is transmitted back from the output layer to the input layer, the model parameters are updated using the gradient descent algorithm to minimize the loss function, and the process of forward propagation, loss calculation, back propagation and parameter updating is repeated until the loss function reaches the preset convergence threshold.

8. The method according to claim 7, wherein, The final risk level of rock creep is determined by the following method: The rock creep variables measured in the past historical data are clustered by K-means clustering, K initial cluster centers are randomly selected, each sample is assigned to the nearest cluster center, and the cluster center is updated to the mean of all samples in each cluster, and the above steps are repeated until the cluster center no longer changes, the center value of each cluster interval and the cluster category to which each sample belongs are recorded, and the cluster category includes high creep risk, medium creep risk and low creep risk; The rock fracture sensitivity coefficient and stress sensitivity coefficient of the rock to be predicted are input into the optimized model to obtain a creep variable prediction value , calculate the distance between it and each cluster center: ； in, Indicates the predicted value of the creep variable and the first The distance between the center values ​​of each cluster interval Indicates the first The center value of each cluster interval The index is the number of cluster intervals, and ; finding the cluster center with the smallest distance determining a corresponding creep risk level.

9. A system for constructing a rock creep model based on cyclic water flooding-stress coupling, characterized in that, The rock creep model construction system is used to perform the rock creep model construction method based on the cyclic water flooding-stress coupling effect according to any one of claims 1-8, comprising: An initial feature acquisition module is configured to test the initial physical and mechanical properties of the rock sample to obtain relevant feature parameters, perform permeation experiments and stress permeation rate experiments on the rock sample to determine the water permeation rate and stress permeation rate of the rock sample, and preprocess the relevant feature parameters of the rock sample, wherein the preprocessing includes normalizing the relevant feature parameters; A principal component analysis module is configured to obtain a covariance matrix between the relevant feature parameters based on principal component analysis, solve the eigenvalues of the covariance matrix to screen out a principal component matrix related to the water permeation rate and the stress permeation rate, and perform sensitivity analysis on the principal component matrix to determine key feature parameters related to the water permeation rate and the stress permeation rate, and calculate the rock crack sensitivity coefficient and the stress sensitivity coefficient of the rock sample based on the relevant key feature parameters; A water migration coupling construction module is configured to construct a water migration equation based on Darcy's law, and couple the water migration equation based on the calculated rock crack sensitivity coefficient and stress sensitivity coefficient to establish a rock creep relationship to obtain the creep variable of the rock under the stress and water cycle with time. A rock creep prediction model construction module is configured to construct and train a rock creep prediction model with the rock crack sensitivity coefficient and the stress sensitivity coefficient as the model input and the rock creep variable as the model output, and optimize the model based on the established loss function. A creep risk clustering analysis module is configured to perform clustering analysis on the rock creep variables measured in the past historical data to obtain different risk rock creep risk clustering intervals, input the rock crack sensitivity coefficient and the stress sensitivity coefficient of the rock to be predicted into the optimized model to obtain a rock creep variable prediction value, and calculate the distance between the rock creep variable prediction value and the center value of different risk level intervals to determine the final risk level of rock creep.

Images