Rock mass instability grading early warning method and system based on micro-seismic feature perception and classification enhancement algorithm

Abstract

The invention discloses a rock mass instability grading early warning method based on micro-seismic feature perception and a classification enhancement algorithm. The method comprises the following steps: acquiring rock mass instability case data information of different geotechnical engineering projects; selecting a plurality of microseismic parameters as rock mass instability risk candidate evaluation indexes, and performing correlation analysis to obtain rock mass instability grading early warning evaluation indexes; constructing a training data set; constructing a rock mass instability grading early warning primary model based on a classification enhancement algorithm and a genetic algorithm, and training to obtain a rock mass instability grading early warning model; and the obtained rock mass instability grading early warning model is adopted to complete rock mass instability grading early warning based on micro-seismic feature perception and a classification enhancement algorithm. The invention also discloses a system for realizing the rock mass instability grading early warning method based on the micro-seismic feature perception and classification enhancement algorithm. According to the method, rock mass instability grading early warning based on micro-seismic feature perception and a classification enhancement algorithm can be realized, the reliability is higher, and the accuracy is better.

Description

Technical Field

[0001] The present invention belongs to the field of civil engineering, and in particular relates to a rock mass instability classification early warning method and system based on microseismic feature perception and classification enhancement algorithm. Background Art

[0002] Rock instability is a common disaster phenomenon in deep mining and geological engineering. It is mainly caused by the sudden release of stored energy in rocks under high ground stress. Rock instability is characterized by great destructiveness, sudden occurrence and strong unpredictability. With the development of deep mineral resources, the frequency of rock instability disasters continues to rise, posing severe challenges to the safety of miners and engineering operations. Therefore, the prediction of rock instability is of great significance.

[0003] Microseismic monitoring technology is widely used in rock instability monitoring due to its high sensitivity to small-scale seismic activities. Through real-time monitoring of microseismic signals in rock formations, researchers can obtain the dynamic response of rocks under stress and then analyze the fracture characteristics and energy release of rocks. In practical applications, the multi-parameter characteristics of microseismic signals (such as magnitude, frequency, waveform characteristics, etc.) can provide important information support for the mechanism of rock instability.

[0004] At present, the traditional rock instability early warning methods based on microseismic monitoring mainly include methods based on numerical monitoring indicators, applied mathematical monitoring, probability methods, empirical theories, etc. However, these rock instability early warning schemes all have the problems of low reliability and poor accuracy. Machine learning is a rapidly developing technical field in recent years, so some researchers have proposed rock instability early warning schemes based on machine learning schemes. However, this type of rock instability early warning scheme based on machine learning still has problems such as insufficient model generalization ability and easy to fall into local optimal solutions in application, which makes the reliability and accuracy of the final prediction results poor. Summary of the invention

[0005] One of the purposes of the present invention is to provide a rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm with high reliability and good accuracy.

[0006] A second object of the present invention is to provide a system for realizing the rock mass instability graded early warning method based on microseismic feature perception and classification enhancement algorithm.

[0007] The rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm provided by the present invention comprises the following steps:

[0008] S1. Obtain rock mass instability case data information for different geotechnical engineering projects;

[0009] S2. Based on the data information obtained in step S1, select several microseismic parameters as candidate assessment indicators for rock mass instability risk;

[0010] S3. Perform correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability level to obtain the rock mass instability classification warning evaluation index;

[0011] S4. Constructing a training data set based on the rock mass instability classification warning evaluation index obtained in step S3 and the data information obtained in step S1;

[0012] S5. Based on the classification enhancement algorithm and genetic algorithm, a primary model for rock mass instability classification warning is constructed;

[0013] S6. Using the training data set obtained in step S4, the primary model for rock mass instability classification warning constructed in step S5 is trained to obtain a rock mass instability classification warning model;

[0014] S7. Using the rock mass instability classification warning model obtained in step S6, complete the rock mass instability classification warning based on microseismic feature perception and classification enhancement algorithm.

[0015] Step S2, according to the data information obtained in step S1, selects several microseismic parameters as candidate evaluation indicators for rock mass instability, which specifically includes the following steps:

[0016] Determine the rock mass instability level including no instability, slight instability, moderate instability and severe instability;

[0017] The selected candidate assessment indicators for rock mass instability include the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate, apparent volume rate, apparent stress, moment magnitude, focal radius, lithology, and initial geostress.

[0018] Step S3 performs correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability grade to obtain the rock mass instability graded early warning assessment index, which specifically includes the following steps:

[0019] Conduct correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability grade, and select the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate and apparent volume rate as rock mass instability graded early warning assessment indicators;

[0020] Among them, the calculation formula of microseismic energy E is: Where ρ is rock density, c is wave velocity, R is the propagation distance of microseismic signal, V(f) is the velocity spectrum of source signal, and f is the frequency of microseismic signal;

[0021] The calculation formula for the apparent volume V is: Where M0 is the seismic moment and μ is the rock mass shear modulus;

[0022] The calculation formula of event rate NR is: Where N is the number of microseismic events occurring within a specific time, A is the area or volume of the monitoring area, and T is the length of a specific time;

[0023] The calculation formula of energy rate ER is: Where EE is the total amount of energy released within a specific time;

[0024] The calculation formula of apparent volume rate VR is: Where V is the apparent volume of the monitoring area.

[0025] The step S5 described in which a primary model for rock mass instability classification warning is constructed based on the classification enhancement algorithm and the genetic algorithm includes the following steps:

[0026] Genetic algorithm is used to optimize the hyperparameters of the classification enhancement algorithm;

[0027] The optimized hyperparameters are used as the hyperparameters of the classification enhancement algorithm;

[0028] According to the obtained classification enhancement algorithm, a primary model for rock mass instability classification warning is constructed.

[0029] The step S5 specifically includes the following steps:

[0030] Create an initial population: The initial population is a set of chromosomes, each of which contains a set of hyperparameters of the classification enhancement algorithm;

[0031] Calculate fitness: calculate the fitness of the current population;

[0032] Selection, crossover and mutation: According to the fitness of the current population, select the top individuals and perform selection, crossover and mutation;

[0033] Judgment termination: If the number of iterations reaches the set value, or the fitness difference between two consecutive calculations is less than the set threshold, the iteration is stopped; the hyperparameters corresponding to the individual with the best fitness are used as the hyperparameters of the classification enhancement algorithm;

[0034] For the data set of microseismic events, (N, E, V, NR, ER, VR) is the feature vector, y i is the target vector;

[0035] The initial classification enhancement algorithm model with optimal hyperparameters obtained by genetic algorithm is expressed as

[0036] F0(x)=argmin c ∑ i L(y i ,c)

[0037] Where L(yi ,c) is the loss function of the classification enhancement algorithm; y i is the target vector, indicating the instability level of the rock mass; c is the parameter of the initialization model;

[0038] The classification enhancement algorithm model is iterated, and in the tth iteration, the residual of each earthquake parameter is calculated:

[0039]

[0040] Where r it is the residual of each earthquake parameter in round t; F(x i ) is the model input x i The predicted value of is the partial derivative of the loss function with respect to the model prediction value;

[0041] Train a weak learner to fit the residual, expressed as

[0042] h t (x) = argmin h ∑ i (r it -h(x i )) 2

[0043] Where h t (x) is the weak learner in the tth round of training; h(x i ) is a weak learner for input x i The predicted value of; h is the parameter of the weak learner;

[0044] Thus, the classification enhancement algorithm model is updated and expressed as

[0045] F i (x) = F i-1 (x)+αh t (x)

[0046] Where F i (x) is the classification enhancement algorithm model after the tth round of update; α is the learning rate;

[0047] Finally, a classification enhancement algorithm model with the selected rock instability classification early warning evaluation index as input and the rock instability classification prediction result as output is obtained, which serves as the primary model of rock instability classification early warning.

[0048] The present invention also provides a system for implementing the rock instability graded warning method based on microseismic feature perception and classification enhancement algorithm, comprising a data acquisition module, an indicator selection module, an indicator evaluation module, a training set construction module, a model construction module, a model training module and a risk prediction module; the data acquisition module, the indicator selection module, the indicator evaluation module, the training set construction module, the model construction module, the model training module and the risk prediction module are connected in series in sequence; the data acquisition module is used to obtain rock instability case data information of different geotechnical engineering projects, and upload the data information to the indicator selection module; the indicator selection module is used to select several microseismic parameters as candidate assessment indicators for rock instability risk based on the received data information, and upload the data information to the indicator evaluation module; the indicator evaluation module is used to perform correlation analysis on the selected several microseismic parameters and the rock instability grade based on the received data information , obtain rock instability graded warning evaluation indicators, and upload the data information to the training set construction module; the training set construction module is used to construct a training data set based on the received data information, the obtained rock instability graded warning evaluation indicators, and the acquired data information, and upload the data information to the model construction module; the model construction module is used to construct a rock instability graded warning primary model based on the received data information, based on the classification enhancement algorithm and the genetic algorithm, and upload the data information to the model training module; the model training module is used to train the constructed rock instability graded warning primary model based on the received data information and the obtained training data set, obtain the rock instability graded warning model, and upload the data information to the risk prediction module; the risk prediction module is used to complete the rock instability graded warning based on the microseismic feature perception and classification enhancement algorithm using the obtained rock instability graded warning model according to the received data information.

[0049] The rock instability graded warning method and system based on microseismic feature perception and classification enhancement algorithm provided by the present invention selects evaluation indicators through rock instability case data, and trains and uses the rock instability graded warning model based on the classification enhancement algorithm optimized by the genetic algorithm. Therefore, the present invention can not only realize the rock instability graded warning based on microseismic feature perception and classification enhancement algorithm, but also has higher reliability and better accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The figure is a schematic diagram of the method flow of the present invention.

[0051] Figure 2 It is a schematic diagram of performance indicators of a comparative example of the method of the present invention.

[0052] Figure 3 Schematic diagram of the functional modules of the system of the present invention. DETAILED DESCRIPTION

[0053] like Figure 1 The method flow diagram of the method of the present invention is shown as follows: The rock mass instability classification early warning based on microseismic feature perception and classification enhancement algorithm provided by the present invention comprises the following steps:

[0054] S1. Obtain rock mass instability case data information for different geotechnical engineering projects. Some sample data are shown in Table 1.

[0055] Table 1 Schematic diagram of some sample data

[0056]

[0057] S2. According to the data information obtained in step S1, several microseismic parameters are selected as candidate assessment indicators for rock mass instability risk; specifically, the steps include:

[0058] Determine the rock mass instability level including no instability, slight instability, moderate instability and severe instability;

[0059] The selected candidate evaluation indicators for rock mass instability include the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate, apparent volume rate, apparent stress, moment magnitude, focal radius, lithology, initial geostress, etc.;

[0060] S3. Perform correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability grade to obtain rock mass instability graded warning evaluation indicators; specifically comprising the following steps:

[0061] The correlation analysis between the microseismic parameters selected in step S2 and the rock mass instability level is performed:

[0062] In addition to the strong positive correlation between microseismic energy and energy rate, apparent volume and apparent volume rate, and the number of microseismic events and event rate, other parameters also have certain correlations, and each microseismic parameter has an obvious positive correlation with the rock mass instability grade. The larger the Spearman coefficient value, the greater the impact of this feature on the rock mass instability grade. Among them, the larger the correlation coefficient between the number of microseismic events and the rock mass instability grade, the strongest positive correlation.

[0063] Therefore, the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate and apparent volume rate were selected as the rock mass instability classification early warning assessment indicators;

[0064] Among them, the calculation formula of microseismic energy E is: Where ρ is rock density, c is wave velocity, R is the propagation distance of microseismic signal, V(f) is the velocity spectrum of source signal, and f is the frequency of microseismic signal;

[0065] The calculation formula for the apparent volume V is: Where M0 is the seismic moment and μ is the rock mass shear modulus;

[0066] The calculation formula of event rate NR is: Where N is the number of microseismic events occurring within a specific time, A is the area or volume of the monitoring area, and T is the length of a specific time;

[0067] The calculation formula of energy rate ER is: Where EE is the total amount of energy released within a specific time;

[0068] The calculation formula of apparent volume rate VR is: Where V is the apparent volume of the monitoring area;

[0069] These evaluation indicators are closely related to the micro-crack activities in the process of rock mass instability disaster formation, and comprehensively reflect the total fracture time, strength and deformation accumulation of the rock mass. They can better characterize the development trend of rock mass instability, reflect the stress state and fracture of the rock mass, and thus evaluate the development trend and severity of rock mass instability in real time.

[0070] S4. Constructing a training data set based on the rock mass instability classification warning evaluation index obtained in step S3 and the data information obtained in step S1;

[0071] In the specific implementation, random sampling is used to divide the 103 groups of sample data into training sets and test sets, of which 8 sets are training sets and 2 sets are test sets. That is, 82 groups of sample data are randomly selected from the 103 groups as training set samples in the process of building the evaluation model to determine the optimization parameters and model structure, and the remaining 21 groups are used as validation set samples to evaluate the model performance. The two groups of data are independent of each other, have no intersection and are both representative;

[0072] S5. Based on the classification enhancement algorithm and genetic algorithm, a primary model for rock mass instability classification warning is constructed; the steps include:

[0073] Genetic algorithm is used to optimize the hyperparameters of the classification enhancement algorithm;

[0074] The optimized hyperparameters are used as the hyperparameters of the classification enhancement algorithm;

[0075] According to the obtained classification enhancement algorithm, a primary model for rock mass instability classification warning is constructed;

[0076] The specific implementation includes the following steps:

[0077] Create an initial population: The initial population is a set of chromosomes, each of which contains a set of hyperparameters of the classification enhancement algorithm;

[0078] Calculate fitness: calculate the fitness of the current population;

[0079] Selection, crossover and mutation: According to the fitness of the current population, select the top individuals and perform selection, crossover and mutation;

[0080] Judgment termination: If the number of iterations reaches the set value, or the fitness difference between two consecutive calculations is less than the set threshold, the iteration is stopped; the hyperparameters corresponding to the individual with the best fitness are used as the hyperparameters of the classification enhancement algorithm;

[0081] For the data set of microseismic events, (N, E, V, NR, ER, VR) is the feature vector, y i is the target vector (i.e. the predicted rock mass instability level);

[0082] The initial classification enhancement algorithm model with optimal hyperparameters obtained by genetic algorithm is expressed as

[0083] F0(x)=argmin c ∑ i L(y i ,c)

[0084] Where L(y i ,c) is the loss function of the classification enhancement algorithm; y i is the target vector, indicating the instability level of the rock mass; c is the parameter of the initialization model;

[0085] The classification enhancement algorithm model is iterated, and in the tth iteration, the residual of each earthquake parameter is calculated:

[0086]

[0087] Where r it is the residual of each earthquake parameter in round t; F(x i ) is the model input x i The predicted value of is the partial derivative of the loss function with respect to the model prediction value;

[0088] Train a weak learner to fit the residual, expressed as

[0089] h t (x) = argmin h ∑ i (r it -h(x i )) 2

[0090] Where h t (x) is the weak learner in the tth round of training; h(x i ) is a weak learner for input x i The predicted value of; h is the parameter of the weak learner;

[0091] Thus, the classification enhancement algorithm model is updated and expressed as

[0092] F i (x) = F i-1 (x)+αh t (x)

[0093] Where F i (x) is the classification enhancement algorithm model after the tth round of update; α is the learning rate;

[0094] Finally, a classification enhancement algorithm model with the selected rock mass instability classification early warning evaluation index as input and the rock mass instability classification prediction result as output is obtained as the primary model of rock mass instability classification early warning;

[0095] S6. Using the training data set obtained in step S4, the primary model for rock mass instability classification warning constructed in step S5 is trained to obtain a rock mass instability classification warning model;

[0096] S7. Using the rock mass instability classification warning model obtained in step S6, complete the rock mass instability classification warning based on microseismic feature perception and classification enhancement algorithm.

[0097] The method of the present invention is tested on a test set with existing solutions (XGBoost (gradient boost decision tree), RF (random forest), Logistics Regression (logistic regression algorithm), BP neural network and other classification models), and the comprehensive performance is compared using several indicators such as accuracy, precision, recall, F1 value, AUC value, etc. The comparison data are as follows: Figure 2 As shown in Table 2:

[0098] Table 2 Performance comparison diagram

[0099]

[0100] pass Figure 2 As can be seen from Table 2, the accuracy of the method of the present invention reaches 90%, which is 9% higher than that of the CatBoost model (classification enhancement model), 5% higher precision, 1% higher recall, and 4% higher F1 value; therefore, the method of the present invention has better accuracy and evaluation effect;

[0101] The method of the present invention is compared with machine learning algorithms such as CatBoost, random forest (RF), XGBoost, logistic regression (LogisticsRegression), BP neural network, etc., and the accuracy rate is improved by 9%, 4%, 9%, 19%, and 14% respectively. Its precision rate, recall rate, and F1 value are all higher than those of other models, and its performance is excellent.

[0102] like Figure 3 The functional module schematic diagram of the system of the present invention is shown as follows: the system disclosed in the present invention for realizing the rock mass instability graded warning method based on microseismic feature perception and classification enhancement algorithm comprises a data acquisition module, an indicator selection module, an indicator evaluation module, a training set construction module, a model construction module, a model training module and a risk prediction module; the data acquisition module, the indicator selection module, the indicator evaluation module, the training set construction module, the model construction module, the model training module and the risk prediction module are connected in series in sequence; the data acquisition module is used to obtain rock mass instability case data information of different geotechnical engineering projects, and upload the data information to the indicator selection module; the indicator selection module is used to select a number of microseismic parameters as candidate assessment indicators for rock mass instability risk according to the received data information, and upload the data information to the indicator evaluation module; the indicator evaluation module is used to compare the selected several microseismic parameters with the rock mass instability risk according to the received data information. The correlation analysis is performed on the level to obtain the rock instability graded early warning evaluation index, and the data information is uploaded to the training set construction module; the training set construction module is used to construct a training data set based on the received data information, the obtained rock instability graded early warning evaluation index, and the acquired data information, and upload the data information to the model construction module; the model construction module is used to construct a primary model for rock instability graded early warning based on the received data information, the classification enhancement algorithm and the genetic algorithm, and upload the data information to the model training module; the model training module is used to train the constructed primary model for rock instability graded early warning based on the received data information and the obtained training data set, to obtain the rock instability graded early warning model, and upload the data information to the risk prediction module; the risk prediction module is used to complete the rock instability graded early warning based on microseismic feature perception and classification enhancement algorithm by using the obtained rock instability graded early warning model according to the received data information.

Claims

1. A rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm, comprising the following steps: S1. Obtain rock mass instability case data information for different geotechnical engineering projects; S2. Based on the data information obtained in step S1, select several microseismic parameters as candidate assessment indicators for rock mass instability risk; S3. Perform correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability level to obtain the rock mass instability classification warning evaluation index; S4. Constructing a training data set based on the rock mass instability classification warning evaluation index obtained in step S3 and the data information obtained in step S1; S5. Based on the classification enhancement algorithm and genetic algorithm, a primary model for rock mass instability classification warning is constructed; S6. Using the training data set obtained in step S4, the primary model for rock mass instability classification warning constructed in step S5 is trained to obtain a rock mass instability classification warning model; S7. Using the rock mass instability classification warning model obtained in step S6, complete the rock mass instability classification warning based on microseismic feature perception and classification enhancement algorithm.

2. The rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm according to claim 1 is characterized in that Step S2, according to the data information obtained in step S1, selects several microseismic parameters as candidate evaluation indicators for rock mass instability, which specifically includes the following steps: Determine the rock mass instability level including no instability, slight instability, moderate instability and severe instability; The selected candidate evaluation indicators for rock mass instability include the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate, apparent volume rate, apparent stress, moment magnitude, focal radius, lithology, and initial geostress.

3. The rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm according to claim 2 is characterized in that Step S3 performs correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability grade to obtain the rock mass instability graded early warning assessment index, which specifically includes the following steps: Conduct correlation analysis on the several microseismic parameters selected in step S2 and the rock mass instability grade, and select the number of microseismic events, microseismic energy, apparent volume, event rate, energy rate and apparent volume rate as rock mass instability graded early warning assessment indicators; Among them, the calculation formula of microseismic energy E is: Where ρ is rock density, c is wave velocity, R is the propagation distance of microseismic signal, V(f) is the velocity spectrum of source signal, and f is the frequency of microseismic signal; The calculation formula for the apparent volume V is: Where M0 is the seismic moment, μ is the rock mass shear modulus; The calculation formula of event rate NR is: Where N is the number of microseismic events occurring within a specific time, A is the area or volume of the monitoring area, and T is the length of a specific time; The calculation formula of energy rate ER is: Where EE is the total amount of energy released within a specific time; The calculation formula of apparent volume rate VR is: Where V is the apparent volume of the monitoring area.

4. The rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm according to claim 3 is characterized in that The step S5 described in which a primary model for rock mass instability classification warning is constructed based on the classification enhancement algorithm and the genetic algorithm includes the following steps: Genetic algorithm is used to optimize the hyperparameters of the classification enhancement algorithm; The optimized hyperparameters are used as the hyperparameters of the classification enhancement algorithm; According to the obtained classification enhancement algorithm, a primary model for rock mass instability classification warning is constructed.

5. The rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm according to claim 4 is characterized in that The step S5 specifically includes the following steps: Create an initial population: The initial population is a set of chromosomes, each of which contains a set of hyperparameters of the classification enhancement algorithm; Calculate fitness: calculate the fitness of the current population; Selection, crossover and mutation: According to the fitness of the current population, select the top individuals and perform selection, crossover and mutation; Judgment termination: If the number of iterations reaches the set value, or the fitness difference between two consecutive calculations is less than the set threshold, the iteration is stopped; the hyperparameters corresponding to the individual with the best fitness are used as the hyperparameters of the classification enhancement algorithm; For the data set of microseismic events, (N, E, V, NR, ER, VR) is the feature vector, y i is the target vector; The initial classification enhancement algorithm model with optimal hyperparameters obtained by genetic algorithm is expressed as F0(x)=argmin c ∑ i L(y i ,c) Where L(y i ,c) is the loss function of the classification enhancement algorithm; y i is the target vector, indicating the instability level of the rock mass; c is the parameter of the initialization model; The classification enhancement algorithm model is iterated, and in the tth iteration, the residual of each earthquake parameter is calculated: Where r it is the residual of each earthquake parameter in round t; F(x i ) is the model input x i The predicted value of is the partial derivative of the loss function with respect to the model prediction value; Train a weak learner to fit the residual, expressed as h t (x)=argmin h ∑ i (r it -h(x i )) 2 Where h t (x) is the weak learner in the tth round of training; h(x i ) is a weak learner for input x i The predicted value of; h is the parameter of the weak learner; Thus, the classification enhancement algorithm model is updated and expressed as F i (x)=F i-1 (x)+αh t (x) Where F i (x) is the classification enhancement algorithm model after the tth round of update; α is the learning rate; Finally, a classification enhancement algorithm model with the selected rock instability classification early warning evaluation index as input and the rock instability classification prediction result as output is obtained, which serves as the primary model of rock instability classification early warning.

6. A system for implementing the rock mass instability classification early warning method based on microseismic feature perception and classification enhancement algorithm as described in any one of claims 1 to 5, characterized in that It includes a data acquisition module, an indicator selection module, an indicator evaluation module, a training set construction module, a model construction module, a model training module and a risk prediction module; the data acquisition module, the indicator selection module, the indicator evaluation module, the training set construction module, the model construction module, the model training module and the risk prediction module are connected in series in sequence; the data acquisition module is used to obtain data information on rock mass instability cases of different geotechnical engineering projects, and upload the data information to the indicator selection module; The indicator selection module is used to select several microseismic parameters as candidate assessment indicators of rock mass instability risk based on the received data information and the acquired data information, and upload the data information to the indicator assessment module; The index evaluation module is used to perform correlation analysis on several selected microseismic parameters and rock mass instability levels based on the received data information, obtain rock mass instability classification warning evaluation indicators, and upload the data information to the training set construction module; The training set construction module is used to construct a training data set based on the received data information, the obtained rock mass instability classification warning evaluation indicators, and the acquired data information, and upload the data information to the model construction module; The model building module is used to build a primary model for rock mass instability classification warning based on the received data information, classification enhancement algorithm and genetic algorithm, and upload the data information to the model training module; The model training module is used to train the constructed rock mass instability graded early warning primary model based on the received data information and the obtained training data set, obtain the rock mass instability graded early warning model, and upload the data information to the risk prediction module; The risk prediction module is used to complete the rock mass instability classification warning based on microseismic feature perception and classification enhancement algorithm according to the received data information and the obtained rock mass instability classification warning model.

Images