Prediction method of deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data

Abstract

The present invention mainly relates to the field of geological prediction technology. A method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data is provided, including predicting the stability of underground cavern surrounding rock and constructing a stability classification model of underground cavern surrounding rock; collecting multi-source monitoring data of underground cavern surrounding rock to establish a multi-source feature data set, from which the principal component factors are screened out; establishing an influencing factor set for deformation prediction of underground cavern surrounding rock based on the screened principal component factors and the stability of the surrounding rock; constructing an underground cavern surrounding rock deformation prediction model and training it; constructing an influencing factor set for deformation prediction of underground cavern surrounding rock of target surrounding rock, and the trained underground cavern surrounding rock deformation prediction model predicts the deformation of the target surrounding rock based on the influencing factor set for deformation prediction of underground cavern surrounding rock of target surrounding rock, which can quickly and accurately analyze the deformation evolution results of surrounding rock of deep caverns under extremely high stress, and provide a decision-making basis for on-site construction.

Description

Technical Field

[0001] The present invention mainly relates to the field of geological prediction technology, and in particular to a method for predicting deformation evolution of surrounding rocks of extremely high stress deep cave groups based on multi-source data. Background Art

[0002] In water conservancy and hydropower projects, the safe excavation of underground caverns is one of the most important issues. Some underground caverns face engineering geological problems such as large spans, high ground stress and steep inclination. The decision-making of each link, such as surrounding rock deformation prediction and construction support scheme selection, is related to the quality of the entire hydropower project. In actual engineering, it is very necessary to predict the surrounding rock deformation and evolution characteristics of high ground stress caverns. Limited by the influence of complex prerequisite geological conditions, construction site conditions and other controlling factors, most of the research and application of deep cavern surrounding rock deformation evolution prediction technology only focuses on the consideration of a single data source.

[0003] Especially in extremely high stress environments, cracks and microcracks in the rock mass of deep caverns are more likely to expand, which reduces the integrity of the surrounding rock, aggravates the degree of rock fragmentation, and is more likely to cause rock bursts and rock layer movement, posing a threat to engineering safety. In addition, there are many factors that affect the deformation of the surrounding rock of caverns under extremely high stress, and there are complex nonlinear relationships between them. It is difficult for a single data source to accurately predict the deformation evolution of the surrounding rock of caverns. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for predicting the deformation evolution of surrounding rocks of extremely high stress deep cave groups based on multi-source data, so as to quickly and accurately analyze the deformation evolution results of surrounding rocks of underground caverns under extremely high stress conditions.

[0005] The technical solution adopted by the present invention to solve the above technical problems is

[0006] The prediction method of deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data includes:

[0007] Step S1: constructing a classification model for the stability of the surrounding rock of underground caverns to predict the degree of stability of the surrounding rock of underground caverns;

[0008] Step S2: Collect multi-source monitoring data of the surrounding rock of the underground cavern to establish a multi-source characteristic data set, analyze the internal correlation of each factor in the multi-source characteristic data set and the degree of contribution to the surrounding rock deformation, screen out the principal component factors, and use the screened principal component factors and the stability of the surrounding rock of the underground cavern as the influencing factor set for the prediction of the deformation of the surrounding rock of the underground cavern;

[0009] Step S3: collecting deformation data of surrounding rock of excavated underground caverns and parameters of a set of influencing factors for predicting deformation of surrounding rock of underground caverns;

[0010] Step S4: constructing an underground cavern surrounding rock deformation prediction model, and training the constructed underground cavern surrounding rock deformation prediction model based on the collected excavated underground cavern surrounding rock deformation data and underground cavern surrounding rock deformation prediction influencing factor set parameters;

[0011] Step S5: construct a set of influencing factors for deformation prediction of the target underground cavern surrounding rock, and perform deformation prediction on the target underground cavern surrounding rock based on the trained underground cavern surrounding rock deformation prediction model.

[0012] Furthermore, step S1 of constructing a classification model for the stability of the surrounding rock of an underground cavern includes:

[0013] Formulate classification and quantitative standards for the stability of surrounding rock of underground caverns;

[0014] Based on the grey correlation method, the key geological characteristic parameters affecting the stability of the surrounding rock of underground caverns were screened out;

[0015] The key geological characteristic parameter values ​​that affect the stability of the surrounding rock of the underground cavern are used as the input of the DCNN-SVC model, and the classification and quantitative standard values ​​of the stability degree of the surrounding rock of the underground cavern are used as the output to train the DCNN-SVC model.

[0016] During the training process, the DCNN network structure is used to extract the key geological characteristic parameters. The key geological characteristic parameters after feature extraction are used as input into the SVC model. The optimal hyperparameter combination of the SVC model is searched based on the ISABO algorithm. The optimal hyperparameter combination is assigned to the SVC model to obtain the trained underground cavern surrounding rock stability classification model.

[0017] Furthermore, the key geological characteristic parameters that affect the stability of the surrounding rock of the underground cavern are preprocessed using the minimum-maximum normalization method and the one-hot encoding method and then used as the input of the DCNN-SVC model.

[0018] Furthermore, step S2 also includes: identifying and removing abnormal parameters of the multi-source feature data set based on the Laida criterion, performing data cleaning on the multi-source feature data set after the abnormal parameters are identified and removed, and performing quantitative numerical representation on the multi-source feature data set after the data cleaning.

[0019] Furthermore, in step S2, the principal component factors are screened out from the multi-source feature data set by using the KPCA algorithm.

[0020] Furthermore, the multi-source feature data set in step S2 includes monitoring data of underground surrounding rock multi-point displacement meters, monitoring data of stress meters, the location of monitoring instrument deployment, the spatiotemporal state of surrounding rock before instrument measurement, temperature data of the construction environment, and the spatiotemporal state of surrounding rock during instrument measurement.

[0021] Furthermore, step S4 of constructing the underground cavern surrounding rock deformation prediction model specifically includes:

[0022] Dividing the set of influencing factors for prediction of deformation of surrounding rock of the excavated underground cavern and the corresponding deformation data of surrounding rock of the underground cavern into a training set and a test set in proportion;

[0023] The parameters of the underground cavern surrounding rock deformation prediction influencing factor set in the training set are used as the input parameters of the SVR model, and the corresponding underground cavern surrounding rock deformation data are used as the real output label of the SVR model to train the SVR model; during the model training process, the SVR model parameters are optimized based on the NRBO optimization algorithm to determine the kernel function parameters and penalty factors of the SVR model;

[0024] The optimized kernel function parameters and penalty factors are assigned to the SVR model to obtain a trained underground cavern surrounding rock deformation prediction model.

[0025] Furthermore, the mean square error and mean absolute error were used to evaluate the prediction effect of the underground cavern surrounding rock deformation prediction model.

[0026] Furthermore, the method further includes step S6: setting multiple microseismic monitoring systems based on the deformation prediction results of the target underground cavern surrounding rock according to the underground cavern surrounding rock deformation prediction model, and inverting the spatiotemporal distribution and energy size information of microseismic events generated by rock mass rupture during the excavation disturbance process of the underground cavern surrounding rock based on the monitoring results of the microseismic monitoring systems;

[0027] Based on the spatiotemporal distribution and energy information of microseismic events and the deformation prediction model of underground cavern surrounding rock, the deformation evolution of the surrounding rock inside the underground cavern is analyzed by the deformation prediction results of the target underground cavern surrounding rock.

[0028] The beneficial effects of the present invention are:

[0029] (1) Starting from the rock mass characteristics of the surrounding rock of the extremely high-stress deep cave group, the present invention quantifies the stability degree of the surrounding rock of the extremely high-stress deep cave group and combines it with the main component factors screened from the multi-source monitoring data of the surrounding rock of the underground cavern. The prediction parameters for the deformation evolution prediction of the surrounding rock of the extremely high-stress deep cave group are used to train the machine learning model. Based on the trained machine learning model, the deformation evolution results of the surrounding rock of the deep cave group under extremely high stress can be quickly and accurately analyzed, providing a decision-making basis for on-site construction;

[0030] (2) In the present invention, by constructing an underground cavern surrounding rock stability classification model to predict the underground cavern surrounding rock stability, the consistency of the underground cavern surrounding rock stability evaluation results can be effectively improved, which is beneficial to improving the underground cavern surrounding rock deformation prediction accuracy of the underground cavern surrounding rock deformation prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The schematic diagram of the construction method of the underground cavern surrounding rock stability classification model;

[0032] Figure 2 Schematic diagram of the construction method of the surrounding rock deformation prediction model of underground caverns. DETAILED DESCRIPTION

[0033] Step S1: Construct an underground cavern surrounding rock stability classification model to predict the underground cavern surrounding rock stability. First, formulate a classification and quantification standard for the underground cavern surrounding rock stability, use the DCNN-SVC model as the basic model, select key geological characteristic parameters from the multivariate geological characteristic parameters that affect the underground cavern surrounding rock stability as the input parameters of the DCNN-SVC model, and use the classification and quantification standard values ​​of the underground cavern surrounding rock stability as the output of the DCNN-SVC model to train the DCNN-SVC model.

[0034] like Figure 1 As shown, the construction of the underground cavern surrounding rock stability classification model in this embodiment specifically includes:

[0035] From the rock quality index (RQD), rock uniaxial saturated compressive strength (R W ), integrity factor (K V ), structural surface strength coefficient (K F ), rock mass compactness (wave velocity and degree of interlocking, etc.), structural surface characteristics (number of structural surface groups, spacing, opening and filling materials, etc.), groundwater seepage (W), rock mass structural characteristics and other parameters, and the key geological characteristic parameters with a high correlation with the stability of the surrounding rock of the underground cavern are selected.

[0036] In this embodiment, the key geological characteristic parameters with high correlation with the stability of the surrounding rock of the underground cavern are screened based on the grey correlation method. The specific process is as follows:

[0037] (1) Determine the analysis series. Determine the reference series that reflects the system behavior characteristics and the comparison series that affects the system behavior. The rock mass stability is the reference series and the geological characteristic parameters are the comparison series.

[0038] (2) Dimensionless data processing. The geological characteristic parameters are normalized and preprocessed, and the formula is as follows: ; In the formula, Indicates geological characteristic parameters; For the The number of data in each geological characteristic parameter, ; and They are The maximum and minimum values ​​of the data within a geological characteristic parameter.

[0039] (3) Calculate the correlation coefficient. Calculate the correlation coefficient between the comparison sequence and the reference sequence using the following formula: ; In the formula, is the correlation coefficient; is the resolution coefficient, and its value is generally between 0 and 1; Take the absolute value of the difference between the comparison sequence and the reference sequence.

[0040] (4) Calculate the grey relational degree. The formula is as follows: , is the grey correlation value. The geological characteristic parameters with correlation values ​​less than 0.7 are eliminated. The function expression of the optimal hyperplane in the classification feature space is: ,in is the high-dimensional feature vector of the original dimension linearly inseparable features, is the normal vector corresponding to the optimal hyperplane, is the intercept.

[0041] In order to prevent the situation where big data overwhelms small data and ensure the consistency of the selected key geological characteristic parameters in the DCNN-SVC model training, in this embodiment, the key geological characteristic parameters selected based on the grey correlation degree are normalized before training the DCNN-SVC model: ,in, and are the original value and normalized value of the mth sample of each key geological characteristic parameter, and are the minimum and maximum values ​​of each key geological characteristic parameter respectively; the natural breakpoint method is used to reclassify the numerical parameter values ​​under each key geological characteristic parameter, and the geological characteristic parameters are converted into a two-dimensional matrix data form by combining the one-hot encoding method, where the size of the two-dimensional matrix depends on the number of key geological characteristic parameters and the number of interval reclassifications, and the missing data are filled in by the zero-filling method. The DCNN network structure in the model is combined with the residual network module to prevent the risk of overfitting caused by too deep a network structure layer. The DCNN network structure is shown in Table 1, including an input layer, a convolutional layer, a pooling layer, and multiple residual modules and fully connected layers. The specific and precise configuration of the DCNN network structure should be adjusted according to the actual project, and the result of the feature extraction of the key geological characteristic parameters by the DCNN network structure is used as the input of the SVC for training and prediction.

[0042] Table 1 DCNN network structure

[0043]

[0044] Based on the key geological characteristic parameters that affect the stability of the surrounding rock and the classification and quantification standards of the stability of the surrounding rock, a sample data set for training the DCNN-SVC model was produced.

[0045] When surrounding rock geological samples are limited, engineering data samples with similar geological conditions can be used for pre-training, and the prepared sample data set is divided into training samples and test samples. In this embodiment, the ratio of training samples to test samples is 3:1. The DCNN-SVC model is trained using training samples, and the accuracy of the DCNN-SVC model training results is obtained based on the test samples.

[0046] In the process of training the DCNN-SVC model, during the internal training of SVC, the improved mean subtraction algorithm (ISABO) is combined with the SVC model to predict the stability of the surrounding rock of the underground cavern. The optimal hyperparameter combination of the SVC model is searched based on the ISABO algorithm, and the optimal hyperparameter combination is assigned to the SVC model to obtain the trained underground cavern surrounding rock stability classification model.

[0047] The improved subtraction average algorithm (ISABO) adds chaos mapping and golden sine method, which can achieve the purpose of improving population generation method and fast optimization decision-making. The expression of individual dimension data generator is: ;in: represents the number of individuals, Represents dimension; For population size.

[0048] The expression of the golden sine method is:

[0049] ,

[0050] ,

[0051] ;

[0052] in: , is a random number, ranging from 0 to 2π; It means the number of iterations is The optimal position of the particle when is the golden ratio, , The initial value is -π.

[0053] Step S2: Collect multi-source monitoring data of underground cavern surrounding rock to establish a multi-source feature data set, analyze the internal correlation of each factor in the multi-source feature data set and the degree of contribution to the surrounding rock deformation, screen out the principal component factors, and use the screened principal component factors and the stability of the underground cavern surrounding rock as the influencing factor set for the prediction of underground cavern surrounding rock deformation.

[0054] The multi-source monitoring data includes but is not limited to the monitoring data of multi-point displacement meters, the monitoring data of stress meters, the location of monitoring instruments, the spatiotemporal state of surrounding rocks before instrument measurement, the temperature data of the construction environment, and the spatiotemporal state of surrounding rocks during instrument measurement. In order to use different forms of monitoring data as input data of the surrounding rock deformation prediction model, the multi-source monitoring data are represented by quantitative values. For example, the location of monitoring instruments, the spatiotemporal state of surrounding rocks before instrument measurement, and the spatiotemporal state of surrounding rocks during instrument measurement can be represented by a number of quantitative values, and the representation principle can be determined according to the on-site conditions.

[0055] The multi-source monitoring data after quantization is analyzed by kernel principal component analysis (KPCA algorithm) to screen out the principal component factors. The specific process is as follows:

[0056] (1) Calculate the kernel matrix. Use the kernel function to map the data points in the original space to a high-dimensional feature space. ;for Data points , , …, ; Calculate the kernel matrix , whose elements represent the inner products of data points in the kernel space.

[0057] (2) Centralize the kernel matrix. Subtract the corresponding mean from each row and column to get the centralized kernel matrix , ; Where: is the identity matrix; Represents a matrix whose elements are all 1.

[0058] (3) Eigenvalue decomposition. For the centralized kernel matrix Perform eigenvalue decomposition , and get the eigenvalue , , …, The corresponding eigenvector , , …, .

[0059] (4) Calculate the cumulative contribution rate of the eigenvalues. Set the threshold to 90%, retain the principal components whose cumulative contribution rate is greater than the threshold as the principal component factors, and use the eigenvectors corresponding to the principal component factors as the new feature space.

[0060] Since the deformation data and stress monitoring data of underground caverns such as hydropower stations have the characteristics of uncertain changes and short sequence samples, in this embodiment, before preparing the influencing factor set of underground cavern surrounding rock deformation prediction, the Laida criterion is used to identify and eliminate the abnormal data of underground caverns. The residual error satisfy It is believed that Bad values ​​should be removed.

[0061] Based on the stability degree of the surrounding rock and the screened main component factors, a set of influencing factors for predicting the deformation of the surrounding rock of underground caverns is established to prepare data for the subsequent training of the surrounding rock deformation prediction model of underground caverns.

[0062] Step S3: Collecting deformation data of surrounding rock of excavated underground caverns and parameters of a set of influencing factors for predicting deformation of surrounding rock of underground caverns.

[0063] Step S4: constructing an underground cavern surrounding rock deformation prediction model, and training the constructed underground cavern surrounding rock deformation prediction model based on the collected excavated underground cavern surrounding rock deformation data and the underground cavern surrounding rock deformation prediction influencing factor set parameters.

[0064] The SVR model is improved by using the NRBO optimization algorithm. The NRBO algorithm is implemented by optimizing the Newton-Raphson algorithm (NRSR) by introducing the trap avoidance operator (TAO), so that it can improve the convergence efficiency and accuracy of the prediction. The NRSR solves the optimal weights, and the formula is as follows: ;

[0065] in Represents the new position; and Represents a random number, with a value range between 0 and 1; , and Represents the latest vector position found during the search process. The expression of TAO is: When it is less than 0.5, ;otherwise, , where , is a random parameter; is the adaptive coefficient; For the best position; is the current vector position, The position for the next iteration; , is a uniform random number, where , .

[0066] like Figure 2 As shown, the underground cavern surrounding rock influencing parameter factor set in step S3 and the corresponding underground cavern surrounding rock deformation data are divided into a test set and a training set in a ratio of 2:8. The parameters of the underground cavern surrounding rock deformation prediction influencing factor set in the training set are used as the input parameters of the SVR model, and the corresponding underground cavern surrounding rock deformation data are used as the real output labels of the SVR model to train the SVR model. During the model training process, the SVR model is optimized based on the NRBO optimization algorithm to determine the kernel function parameters and penalty factors of the SVR model; wherein the maximum number of iterations in NRBO is 100, and the population size is 20. The parameters after the search are assigned to the SVR model to obtain the trained underground cavern surrounding rock deformation prediction model. The test set is input into the trained underground cavern surrounding rock deformation prediction model for testing. In this embodiment, the mean square error (MSE) and mean absolute error (MAE) are used to evaluate the prediction effect of the surrounding rock deformation prediction model.

[0067] Step S5: construct a set of influencing factors for deformation prediction of the target underground cavern surrounding rock, and perform deformation prediction on the target underground cavern surrounding rock based on the trained underground cavern surrounding rock deformation prediction model.

[0068] Step S6: Combined with the microseismic monitoring system technology, combined with the deformation prediction results of the target underground cavern surrounding rock in step S5, focus on the monitoring points with obvious deformation phenomena, and further analyze the deformation of the underground cavern surrounding rock and reveal the spatiotemporal evolution mechanism. The sensor layout of the microseismic monitoring system should be adjusted according to the actual project. For example, the distance between sensors in adjacent sections should not be too close, otherwise it will affect the positioning error and monitoring range. At the same time, the horizontal distance between the sensor and the underground cavern surrounding rock face should not be less than 30m. By statistically analyzing the microseismic event data in a certain period of time, its distribution law in time and space can be obtained, and the relationship between the energy released by the rupture of the surrounding rock of the underground cavern and the events can be revealed. Timely guidance can be provided for the support and reinforcement of the surrounding rock of the underground cavern through typical characteristics. Finally, a new evaluation index, i.e., lgE / b, is formed by combining the microseismic energy value E and the earthquake probability coefficient b to evaluate the surrounding rock of the underground cavern. The b value is calculated gradually with a cycle of 7 days and a step of 1 day. The response process of lgE / b with the construction intensity can be analyzed through the change trend of lgE / b. According to the complete microseismic sequence data of the surrounding rock of the underground cavern, the time variation law of the cumulative released energy, energy index and cumulative apparent volume of microseismic events can be obtained, and the surrounding rock state stage can be analyzed more clearly.

Claims

1. A method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data, characterized in that: include: Step S1: constructing a classification model for the stability of the surrounding rock of underground caverns to predict the degree of stability of the surrounding rock of underground caverns; Step S2: Collect multi-source monitoring data of the surrounding rock of the underground cavern to establish a multi-source characteristic data set, analyze the internal correlation of each factor in the multi-source characteristic data set and the degree of contribution to the surrounding rock deformation, screen out the principal component factors, and use the screened principal component factors and the stability of the surrounding rock of the underground cavern as the influencing factor set for the prediction of the deformation of the surrounding rock of the underground cavern; Step S3: collecting deformation data of surrounding rock of excavated underground caverns and parameters of a set of influencing factors for predicting deformation of surrounding rock of underground caverns; Step S4: constructing an underground cavern surrounding rock deformation prediction model, and training the constructed underground cavern surrounding rock deformation prediction model based on the collected excavated underground cavern surrounding rock deformation data and underground cavern surrounding rock deformation prediction influencing factor set parameters; Step S5: constructing a set of influencing factors for deformation prediction of surrounding rock of target underground cavern, and performing deformation prediction on surrounding rock of target underground cavern based on the trained deformation prediction model of surrounding rock of underground cavern; Step S1 of constructing the underground cavern surrounding rock stability classification model includes: Formulate classification and quantitative standards for the stability of surrounding rock of underground caverns; Based on the grey correlation method, the key geological characteristic parameters affecting the stability of the surrounding rock of underground caverns were screened out; The key geological characteristic parameter values ​​that affect the stability of the surrounding rock of the underground cavern are used as the input of the DCNN-SVC model, and the classification and quantitative standard values ​​of the stability degree of the surrounding rock of the underground cavern are used as the output to train the DCNN-SVC model. During the training process, the DCNN network structure is used to extract the key geological characteristic parameters. The extracted key geological characteristic parameters are used as input into the SVC model. The optimal hyperparameter combination of the SVC model is searched based on the ISABO algorithm. The optimal hyperparameter combination is assigned to the SVC model to obtain the trained underground cavern surrounding rock stability classification model. Step S4 of constructing the underground cavern surrounding rock deformation prediction model specifically includes: Dividing the set of influencing factors for prediction of deformation of surrounding rock of the excavated underground cavern and the corresponding deformation data of surrounding rock of the underground cavern into a training set and a test set in proportion; The parameters of the influencing factor set of underground cavern surrounding rock deformation prediction in the training set are used as the input parameters of the SVR model, and the corresponding underground cavern surrounding rock deformation data are used as the real output label of the SVR model to train the SVR model; during the model training process, the parameters of the SVR model are optimized based on the NRBO optimization algorithm to determine the kernel function parameters and penalty factors of the SVR model; The optimized kernel function parameters and penalty factors are assigned to the SVR model to obtain the trained underground cavern surrounding rock deformation prediction model.

2. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to claim 1 is characterized in that: The key geological characteristic parameters that affect the stability of the surrounding rock of the underground cavern are preprocessed using the minimum-maximum normalization method and the one-hot encoding method and then used as the input of the DCNN-SVC model.

3. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to claim 1 is characterized in that: Step S2 also includes: identifying and removing abnormal parameters from the multi-source feature data set based on the Laida criterion, cleaning the multi-source feature data set after the abnormal parameters are identified and removed, and quantifying and numerically representing the cleaned multi-source feature data set.

4. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to claim 3 is characterized in that: In step S2, the principal component factors are screened out from the multi-source feature data set by using the KPCA algorithm.

5. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to claim 4 is characterized in that: The multi-source feature data set in step S2 includes monitoring data of underground surrounding rock multi-point displacement meters, monitoring data of stress meters, the location of monitoring instrument deployment, the spatiotemporal state of surrounding rock before instrument measurement, temperature data of the construction environment, and the spatiotemporal state of surrounding rock during instrument measurement.

6. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to claim 1 is characterized in that: The mean square error and mean absolute error are used to evaluate the prediction effect of the underground cavern surrounding rock deformation prediction model.

7. The method for predicting deformation evolution of surrounding rock of extremely high stress deep caverns based on multi-source data according to any one of claims 1 to 6, characterized in that: The step S6 is also included: according to the underground cavern surrounding rock deformation prediction model, multiple microseismic monitoring systems are set up for the deformation prediction results of the target underground cavern surrounding rock, and the spatiotemporal distribution and energy information of microseismic events generated by rock mass rupture during the excavation disturbance process of the underground cavern surrounding rock are inverted based on the monitoring results of the microseismic monitoring systems; Based on the spatiotemporal distribution and energy information of microseismic events and the deformation prediction model of underground cavern surrounding rock, the deformation evolution of the surrounding rock inside the underground cavern is analyzed by the deformation prediction results of the target underground cavern surrounding rock.

Images