Coal rock dynamic failure multi-parameter early warning method based on SSA-SVM algorithm

Abstract

A coal rock dynamic destruction multi-parameter early warning method based on an SSA-SVM algorithm comprises the following steps that (1) a coal rock sample needing early warning is selected, and acoustic emission signals in the coal rock sample dynamic destruction process are collected; (2) extracting time-frequency domain characteristic information of the acoustic emission signals to obtain 10 acoustic emission indexes and evolution laws; (3) carrying out normalization processing; (4) repeating the steps (1)-(3), carrying out a test on not less than 50 coal rocks, and selecting acoustic emission indexes of which the success rate is greater than a preset percentage threshold value; (5) constructing a feature matrix, and dividing a training subset and a verification subset; (6) constructing an SVM regression model architecture; (7) optimizing a penalty factor c and a parameter g of a radial basis kernel function by using a sparrow search algorithm, and determining an optimal hyper-parameter combination through cross validation; and (8) predicting the energy change at the future moment by using the prediction model so as to realize the real-time early warning of the coal rock dynamic disaster. The early warning accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of coal and rock dynamic damage monitoring and early warning technology, specifically to a multi-parameter early warning method for coal and rock dynamic damage based on the SSA-SVM algorithm. Background Technology

[0002] Early warning of coal mine dynamic disasters is a key technological support for ensuring the safe and efficient mining of deep resources. As my country's shallow coal resources are gradually depleted, mining depths are extending to depths of over 1,000 meters annually at a rate of 10-25 meters. The "three highs" environment—high ground stress, high gas pressure, and strong mining disturbance—leads to frequent and clustered coal and rock dynamic disasters (such as rockbursts and coal and gas outbursts). These disasters are characterized by their suddenness, intense energy release, and significant chain-reaction damage. In the past decade, coal mine dynamic disasters nationwide have caused an average of over 100 casualties and hundreds of millions of yuan in direct economic losses annually, seriously threatening miners' lives and the stable operation of enterprises.

[0003] Currently, many scholars have used microseismic signals for rockburst early warning and achieved certain results, but there are still many shortcomings: (1) Although multiple early warning indicators are used, the correlation analysis between different indicators is insufficient, and some indicators contradict each other; (2) When performing data fusion calculation for multiple indicators, the weighted average method is used, which ignores the nonlinear characteristics of coal and rock dynamic disasters, resulting in a low early warning success rate; (3) The machine learning algorithm used is prone to getting trapped in local optima, resulting in a high false alarm rate; (4) The real-time performance of dynamic early warning is insufficient, and the static model used does not take into account changes in geological conditions; (5) The noise interference suppression capability is weak. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention proposes a multi-parameter early warning method for dynamic damage of coal and rock based on the SSA-SVM algorithm. This method can collect acoustic emission data of coal and rock samples in real time during the dynamic damage process and achieve real-time and accurate early warning through the SSA-SVM algorithm, providing a reference for early warning of dynamic disasters in coal and rock mass (such as rockburst).

[0005] The technical solution adopted by this invention to solve its technical problem is:

[0006] A multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm includes the following steps:

[0007] (1) Select coal and rock samples that require early warning, conduct indoor uniaxial or triaxial compression tests, and use an acoustic emission monitoring system to collect acoustic emission signals during the dynamic failure process of the coal and rock samples.

[0008] (2) The acoustic emission signal was preprocessed using MATLAB to extract its time-frequency domain features, resulting in 10 acoustic emission indices and their evolution laws. The 10 acoustic emission indices are represented as: A(t), a, A...(b) S D , ΔF, η, Z map Q t AC, d s A(t) represents the total fault area. Theoretically, a high value anomaly in A(t) before a strong energy release indicates increased microseismic activity before the release. a represents the acoustic emission event activity rate of the coal and rock damage evolution rate. Its value is positively correlated with the AE event occurrence rate per unit time; the larger this index, the higher the probability of rockburst. (b) This represents the microseismic energy level conversion coefficient; the larger this index, the higher the probability of rockburst occurring. D The index represents the degree of microseismic activity; the larger the index, the higher the probability of rockburst. ΔF represents the microseismic activity scale; the larger the absolute value of this index, the higher the probability of rockburst. η represents the degree of deviation from the magnitude-frequency relationship curve; a convex curve indicates a higher probability of rockburst. Z map This represents the microseismic anomaly value; the larger the absolute value of this index, the higher the probability of a rockburst. Q t The entropy of microseismic activity information is represented by ; a decrease in this index indicates a higher probability of rockburst. AC represents algorithm complexity; a greater rate of decrease in this index indicates a higher probability of rockburst. s This indicates spatiotemporal extensibility; before a large-energy event is released, this indicator experiences a rapid increase.

[0009] (3) Normalize the above 10 acoustic emission indicators according to the calculation method of positive, negative or bidirectional indicators;

[0010] (4) Repeat steps (1) to (3) and conduct tests on no less than 50 coal and rock samples. Compare and analyze the evolution law of acoustic emission index with the stress evolution law of the sample, calculate the success rate of each acoustic emission index prediction, and select acoustic emission indexes with a success rate greater than the preset percentage threshold.

[0011] (5) Construct a feature matrix based on the selected acoustic emission index signals, and divide the training subset and the validation subset;

[0012] (6) Construct the SVM regression model architecture and complete the kernel function selection and structural parameter initialization;

[0013] (7) The Sparrow Search Algorithm (SSA) is used to optimize the penalty factor c and the parameter g of the radial basis kernel function, and the optimal hyperparameter combination is determined by cross-validation.

[0014] (8) By applying the established prediction model to predict energy changes in the future, real-time early warning of coal and rock dynamic disasters can be achieved.

[0015] Furthermore, in step (1), the acoustic emission system should be an acoustic emission system manufactured by PAC Corporation of the United States, with no fewer than 6 probes arranged in a spatially distributed manner, not on the same plane.

[0016] Furthermore, in step (2), the expressions for the 10 acoustic emission indicators are as follows:

[0017]

[0018] In the above formula, N(k) is the number of acoustic emission sources with energy level k within the time interval from time t to t+Δt; k is the energy level of each source.

[0019]

[0020] In the above formula, m is the magnitude of the earthquake; a and b are coefficients, whose statistical significance is as follows: a represents the level of earthquake activity, and b represents the proportional relationship between the number of earthquakes of different magnitudes.

[0021]

[0022] In the above formula, M i Let N be the energy level of the i-th microseismic event, and N be the number of microseismic events.

[0023]

[0024] In the above formula, M max This represents the maximum microseismic energy level.

[0025]

[0026] In the above formula, T is the time window and M is the acoustic emission source energy level.

[0027]

[0028] In the above formula, X i =M i -M0, where M0 is the initial energy level; N i Indicates the number of focal events; M i This represents the energy level of the i-th source event.

[0029]

[0030] In the above formula, N to n represents the total number of micro-earthquakes. sa σ represents the number of microseismic samples; M 2 σ represents the overall microseismic energy level variance; m 2 The variance of the sample microseismic energy levels; This represents the average energy level of the overall microseismic events. This represents the average energy level of the sample microseismic events.

[0031]

[0032] In the above formula, p i =(t i+1 -t i ) / (t N -t1), t i t represents the time when the i-th micro-earthquake occurs; N This refers to the moment when the Nth micro-earthquake occurs.

[0033]

[0034] In the above formula, M AC =M max -M min +1;

[0035]

[0036] In the above formula, The average distance between the sequentially occurring focal events; This represents the average time interval between sequentially occurring focal events. Therefore, the absolute values ​​of the data are very large. For aesthetic purposes in the graph, the maximum value in a sequence is used as the denominator, and the sequence is normalized. d s It is limited to the interval [0, 1].

[0037] Furthermore, in step (3), the normalization process is performed using the following expression:

[0038]

[0039] Among them, W ij λ represents the normalized microseismic index. ij (t) includes three cases, all of which are positive indicators: λ ij ↑ (t)=[(Q max -Q ij ) / (Q max -Q min )]; Negative index: λ ij ↓ (t)=[(Q min -Q ij ) / (Q max -Q min Two-way indicators:

[0040] Where, λ ij ↑ (t) is a positive index, λij ↓ (t) is a negative indicator. It is a two-way indicator; Q max Q is the maximum value of the indicator. min Q represents the minimum value of the indicator. ij Let be the value of the ij-th microseismic index.

[0041] Preferably, in step (3), the larger the positive index value, the higher the probability of coal and rock dynamic failure; the smaller the negative index value, the higher the probability of coal and rock dynamic failure; and the larger the absolute value of the two-way index, the higher the probability of coal and rock dynamic failure.

[0042] Furthermore, AC is a negative index; A(t), a, A (b) S D , ΔF, η, Q t d s A positive indicator; Z map It is a two-way indicator.

[0043] In step (7), the SSA algorithm includes initializing the population, calculating the fitness value, updating the discoverer, updating the joiner, updating the watcher, updating the optimal sparrow position, and determining whether the termination condition is met. The termination condition is whether the number of iterations is greater than the target value. The process of step (7) is as follows:

[0044] The fitness values ​​are calculated as follows:

[0045]

[0046] In the above formula, X i Let F represent the position of the i-th sparrow. i =f(X) i Let ) represent the fitness value of the i-th sparrow, Z be the number of sparrows, and d be the dimension of the fitness function variable to be solved;

[0047] The updated discoverers are as follows:

[0048]

[0049] In the above formula, t is the current iteration number; j = 1, 2, 3, ..., d represents the dimension; Let be the value of the i-th sparrow in the j-th dimension at the t-th iteration; α∈(0,1] is a random number; iter max is the maximum number of iterations; ST∈[0.5,1.0] and W∈[0,1] are the safety threshold and warning value for the current position, respectively; Z is a random number that follows a normal distribution; L is a 1×d-dimensional matrix where all elements are 1. When W≥ST, it indicates that a small number of sparrows have detected the danger and are alerting the flock to move to a safe location;

[0050] The following individuals have been added:

[0051]

[0052] The following are the updated vigilant individuals:

[0053]

[0054] In the formula: It is the optimal location for the previous generation of the entire flock; f i β is the current fitness value of the sparrow; β is a random number following a normal distribution with mean 0 and variance 1; K is a random number in the range [-1, 1]; ε is a constant, then f g and f w These are the best and worst fitness values ​​in the previous generation of birds; when f i >f g When f indicates that sparrows around the flock are beginning to move towards the center; when f i =f g This indicates that the sparrow in the center of the flock should move closer to the other sparrows;

[0055] The termination conditions are as follows:

[0056] N≥N * (Formula 16).

[0057] Furthermore, in step (8), during the training of the SVM model, the relationship between the predicted value and the actual value is compared, and the goodness of fit is calculated. If the goodness of fit is greater than 85%, the SVM model is used for multi-parameter early warning of coal and rock dynamic damage.

[0058] The beneficial effects of this invention are mainly reflected in the following aspects: it has outstanding global optimization ability, excellent time and dynamic adaptability, high efficiency of multi-source data fusion, enhanced noise suppression ability and low computational resource consumption. It makes full use of the acoustic emission data of coal and rock samples, which can reduce errors and improve the accuracy of early warning. It can make real-time prediction and early warning of coal and rock dynamic disasters. Moreover, the implementation method is simple, cost-saving and has a wider range of applications. Attached Figure Description

[0059] Figure 1 This is a flowchart of a multi-parameter early warning method for dynamic damage to coal and rock based on the SSA-SVM algorithm.

[0060] Figure 2 This is a graph showing the evolution of coal and rock stress and index A(t) over time.

[0061] Figure 3 This is a graph showing the evolution of coal and rock stress and index a over time.

[0062] Figure 4 This is a graph showing the evolution of coal and rock stress and index A(b) over time.

[0063] Figure 5 Coal and rock stress and index S D A diagram illustrating the evolution over time.

[0064] Figure 6 This is a graph showing the evolution of coal and rock stress and the index ΔF over time.

[0065] Figure 7 This is a graph showing the evolution of coal and rock stress and the index η over time.

[0066] Figure 8 Coal and rock stress and index Z map A diagram illustrating the evolution over time.

[0067] Figure 9 This is a graph showing the evolution of coal and rock stress and the index Qt over time.

[0068] Figure 10 This is a graph showing the evolution of coal and rock stress and the index AC over time.

[0069] Figure 11 This is a graph showing the evolution of coal and rock stress and the index ds over time.

[0070] Figure 12 This is a comparison chart of observed and predicted values ​​for acoustic emission energy. Detailed Implementation

[0071] The present invention will now be further described with reference to the accompanying drawings.

[0072] Reference Figures 1-12 A multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm includes the following steps:

[0073] (1) Select coal and rock samples that require early warning, conduct indoor uniaxial or triaxial compression tests, and use an acoustic emission monitoring system to collect acoustic emission signals during the dynamic failure process of the coal and rock samples.

[0074] In step (1), the acoustic emission system should be an acoustic emission system produced by PAC Corporation of the United States, with no less than 6 probes arranged in a spatially distributed manner, not on the same plane.

[0075] (2) The acoustic emission signal was preprocessed using MATLAB to extract its time-frequency domain features, resulting in 10 acoustic emission indices and their evolution laws. The 10 acoustic emission indices are represented as: A(t), a, A... (b) S D , ΔF, η, Z map Qt AC, d s A(t) represents the total fault area. Theoretically, a high value anomaly in A(t) before a strong energy release indicates increased microseismic activity before the release. a represents the acoustic emission event activity rate of the coal and rock damage evolution rate. Its value is positively correlated with the AE event occurrence rate per unit time; the larger this index, the higher the probability of rockburst. (b) This represents the microseismic energy level conversion coefficient; the larger this index, the higher the probability of rockburst occurring. D The index represents the degree of microseismic activity; the larger the index, the higher the probability of rockburst. ΔF represents the microseismic activity scale; the larger the absolute value of this index, the higher the probability of rockburst. η represents the degree of deviation from the magnitude-frequency relationship curve; a convex curve indicates a higher probability of rockburst. Z map This represents the microseismic anomaly value; the larger the absolute value of this index, the higher the probability of a rockburst. Q t The entropy of microseismic activity information is represented by ; a decrease in this index indicates a higher probability of rockburst. AC represents algorithm complexity; a greater rate of decrease in this index indicates a higher probability of rockburst. s This indicates spatiotemporal extensibility; before a large-energy event is released, this indicator experiences a rapid increase.

[0076] In step (2), the expressions for the 10 acoustic emission indicators are as follows:

[0077]

[0078] In the above formula, N(k) is the number of acoustic emission sources with energy level k within the time interval from time t to t+Δt; k is the energy level of each source.

[0079]

[0080] In the above formula, m is the magnitude of the earthquake; a and b are coefficients, whose statistical significance is as follows: a represents the level of earthquake activity, and b represents the proportional relationship between the number of earthquakes of different magnitudes.

[0081]

[0082] In the above formula, M i Let N be the energy level of the i-th microseismic event, and N be the number of microseismic events.

[0083]

[0084] In the above formula, M max This represents the maximum microseismic energy level.

[0085]

[0086] In the above formula, T is the time window and M is the acoustic emission source energy level.

[0087]

[0088] In the above formula, X i =M i -M0, where M0 is the initial energy level; N i Indicates the number of focal events; M i This represents the energy level of the i-th source event.

[0089]

[0090] In the above formula, N to n represents the total number of micro-earthquakes. sa σ represents the number of microseismic samples; M 2 σ represents the overall microseismic energy level variance; m 2 The variance of the sample microseismic energy levels; This represents the average energy level of the overall microseismic events. This represents the average energy level of the sample microseismic events.

[0091]

[0092] In the above formula, p i =(t i+1 -t i ) / (t N -t1), t i t represents the time when the i-th micro-earthquake occurs; N This refers to the moment when the Nth micro-earthquake occurs.

[0093]

[0094] In the above formula, M AC =M max -M min +1;

[0095]

[0096] In the above formula, The average distance between the sequentially occurring focal events; This represents the average time interval between sequentially occurring focal events. Therefore, the absolute values ​​of the data are very large. For aesthetic purposes in the graph, the maximum value in a sequence is used as the denominator, and the sequence is normalized. d s It is limited to the interval [0, 1].

[0097] (3) Normalize the above 10 acoustic emission indicators according to the calculation method of positive, negative or bidirectional indicators;

[0098] In step (3), the normalization process is performed using the following expression:

[0099]

[0100] Among them, W ij λ represents the normalized microseismic index. ij (t) includes three cases, all of which are positive indicators: λ ij ↑ (t)=[(Q max -Q ij ) / (Q max -Q min Negative index: λ ij ↓ (t)=[(Q min -Q ij ) / (Q max -Q min Two-way indicators:

[0101] Where, λ ij ↑ (t) is a positive index, λ ij ↓ (t) is a negative indicator. It is a two-way indicator; Q max Q is the maximum value of the indicator. min Q represents the minimum value of the indicator. ij Let be the value of the ij-th microseismic index.

[0102] Preferably, in step (3), the larger the positive index value, the higher the probability of coal and rock dynamic failure; the smaller the negative index value, the higher the probability of coal and rock dynamic failure; and the larger the absolute value of the two-way index, the higher the probability of coal and rock dynamic failure.

[0103] Furthermore, AC is a negative index; A(t), a, A (b) S D , ΔF, η, Q t d s A positive indicator; Z map It is a two-way indicator.

[0104] (4) Repeat steps (1) to (3) and conduct tests on no less than 50 coal and rock samples. Compare and analyze the evolution law of acoustic emission index with the stress evolution law of the sample, calculate the success rate of each acoustic emission index prediction, and select acoustic emission indexes with a success rate greater than the preset percentage threshold.

[0105] (5) Construct a feature matrix based on the selected acoustic emission index signals, and divide the training subset and the validation subset;

[0106] (6) Construct the SVM regression model architecture and complete the kernel function selection and structural parameter initialization;

[0107] (7) The Sparrow Search Algorithm (SSA) is used to optimize the penalty factor c and the parameter g of the radial basis kernel function, and the optimal hyperparameter combination is determined by cross-validation.

[0108] In step (7), the SSA algorithm includes initializing the population, calculating the fitness value, updating the discoverer, updating the joiner, updating the watcher, updating the optimal sparrow position, and determining whether the termination condition is met. The termination condition is whether the number of iterations is greater than the target value. The process of step (7) is as follows:

[0109] The fitness values ​​are calculated as follows:

[0110]

[0111] In the above formula, X i Let F represent the position of the i-th sparrow. i =f(X) i Let ) represent the fitness value of the i-th sparrow, Z be the number of sparrows, and d be the dimension of the fitness function variable to be solved;

[0112] The updated discoverers are as follows:

[0113]

[0114] In the above formula, t is the current iteration number; j = 1, 2, 3, ..., d represents the dimension; Let be the value of the i-th sparrow in the j-th dimension at the t-th iteration; α∈(0,1] is a random number; iter max is the maximum number of iterations; ST∈[0.5,1.0] and W∈[0,1] are the safety threshold and warning value for the current position, respectively; Z is a random number that follows a normal distribution; L is a 1×d-dimensional matrix where all elements are 1. When W≥ST, it indicates that a small number of sparrows have detected the danger and are alerting the flock to move to a safe location;

[0115] The following individuals have been added:

[0116]

[0117] The following are the updated vigilant individuals:

[0118]

[0119] In the formula: It is the optimal location for the previous generation of the entire flock; fi β is the current fitness value of the sparrow; β is a random number following a normal distribution with mean 0 and variance 1; K is a random number in the range [-1, 1]; ε is a constant, then f g and f w These are the best and worst fitness values ​​in the previous generation of birds; when f i >f g When f indicates that sparrows around the flock are beginning to move towards the center; when f i =f g This indicates that the sparrow in the center of the flock should move closer to the other sparrows;

[0120] The termination conditions are as follows:

[0121] N≥N * (Formula 16).

[0122] (8) By applying the established prediction model to predict energy changes in the future, real-time early warning of coal and rock dynamic disasters can be achieved.

[0123] In step (9), during the training of the SVM model, the relationship between the predicted value and the actual value is compared, and the goodness of fit is calculated. If the goodness of fit is greater than 85%, the SVM model is used for multi-parameter early warning of coal and rock dynamic damage.

[0124] The following case study illustrates the process using a uniaxial compression test of one coal sample as an example:

[0125] (1) Acoustic emission probe and photos before and after the experiment;

[0126] (2) Measure and analyze the acoustic emission parameters, referring to... Figures 2 to 11 ;

[0127] (3) Based on an accuracy rate greater than 85%, seven indicators were obtained, namely:

[0128] A(t), a, A (b) S D , ΔF, Z map d s ;

[0129] (4) The SSA algorithm is used to calculate the parameters c and g in the SVM model to obtain the energy value of the coal sample destruction process, such as... Figure 3 As shown in the figure, the predicted value is very close to the observed value, with a goodness of fit of 0.88. This indicates that the method proposed in this invention makes full use of the acoustic emission data of coal and rock samples, which can reduce errors, improve the accuracy of early warning, and enable real-time prediction and early warning of coal and rock dynamic disasters. Moreover, the method is simple to implement, saves costs, and has a wider range of applications.

[0130] The embodiments described in this specification are merely examples of implementations of the inventive concept and are for illustrative purposes only. The scope of protection of this invention should not be considered limited to the specific forms described in these embodiments; rather, it extends to equivalent technical means conceived by those skilled in the art based on the inventive concept.

Claims

1. A multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm, characterized in that, Includes the following steps: (1) Select coal and rock samples that require early warning, conduct indoor uniaxial or triaxial compression tests, and use an acoustic emission monitoring system to collect acoustic emission signals during the dynamic failure process of the coal and rock samples. (2) The acoustic emission signal was preprocessed using MATLAB to extract its time-frequency domain features, resulting in 10 acoustic emission indices and their evolution laws. The 10 acoustic emission indices are represented as: A(t), a, A... (b) S D , ΔF, η, Z map Q t AC, d s A(t) represents the total fault area. Theoretically, a high value anomaly in A(t) before a strong energy release indicates increased microseismic activity before the release. a represents the acoustic emission event activity rate of the coal and rock damage evolution rate. Its value is positively correlated with the AE event occurrence rate per unit time; the larger this index, the higher the probability of rockburst. (b) This represents the microseismic energy level conversion coefficient; the larger this index, the higher the probability of rockburst occurring. D The index represents the degree of microseismic activity; the larger the index, the higher the probability of rockburst. ΔF represents the microseismic activity scale; the larger the absolute value of this index, the higher the probability of rockburst. η represents the degree of deviation from the magnitude-frequency relationship curve; a convex curve indicates a higher probability of rockburst. Z map This represents the microseismic anomaly value; the larger the absolute value of this index, the higher the probability of a rockburst. Q t The entropy of microseismic activity information is represented by ; a decrease in this index indicates a higher probability of rockburst. AC represents algorithm complexity; a greater rate of decrease in this index indicates a higher probability of rockburst. s This indicates spatiotemporal extensibility; before a large-energy event is released, this indicator experiences a rapid increase. (3) Normalize the above 10 acoustic emission indicators according to the calculation method of positive, negative or bidirectional indicators; (4) Repeat steps (1) to (3) and conduct tests on no less than 50 coal and rock samples. Compare and analyze the evolution law of acoustic emission index with the stress evolution law of the sample, calculate the success rate of each acoustic emission index prediction, and select acoustic emission indexes with a success rate greater than the preset percentage threshold. (5) Construct a feature matrix based on the selected acoustic emission index signals, and divide the training subset and the validation subset; (6) Construct the SVM regression model architecture and complete the kernel function selection and structural parameter initialization; (7) The Sparrow Search Algorithm (SSA) is used to optimize the penalty factor c and the parameter g of the radial basis kernel function, and the optimal hyperparameter combination is determined by cross-validation. (8) By applying the established prediction model to predict energy changes in the future, real-time early warning of coal and rock dynamic disasters can be achieved.

2. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 1, characterized in that, In step (1), the acoustic emission system should be an acoustic emission system produced by PAC Corporation of the United States, with no less than 6 probes arranged in a spatially distributed manner, not on the same plane.

3. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 1 or 2, characterized in that, In step (2), the expressions for the 10 acoustic emission indicators are as follows: In the above formula, N(k) is the number of acoustic emission sources with energy level k within the time interval from time t to t+Δt; k is the energy level of each source. In the above formula, m is the magnitude of the earthquake; a and b are coefficients, whose statistical significance is as follows: a represents the level of earthquake activity, and b represents the proportional relationship between the number of earthquakes of different magnitudes. In the above formula, M i Let N be the energy level of the i-th microseismic event, and N be the number of microseismic events. In the above formula, M max This represents the maximum microseismic energy level. In the above formula, T is the time window, and M is the acoustic emission source energy level; In the above formula, X i =M i -M0, where M0 is the initial energy level; N i Indicates the number of focal events; M i This represents the energy level of the i-th focal event; In the above formula, N to n represents the total number of micro-earthquakes. sa σ represents the number of microseismic samples; M 2 σ represents the overall microseismic energy level variance; m 2 The variance of the sample microseismic energy levels; This represents the average energy level of the overall microseismic events. This represents the average energy level of the sample microseismic events. In the above formula, p i =(t i+1 -t i ) / (t N -t1), t i t represents the time when the i-th micro-earthquake occurs; N This refers to the moment when the Nth micro-earthquake occurs. In the above formula, M AC =M max -M min +1; In the above formula, The average distance between the sequentially occurring focal events; This represents the average time interval between sequentially occurring focal events. Using the maximum value in a sequence as the denominator, the sequence is normalized, and d... s It is limited to the interval [0, 1].

4. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 3, characterized in that, In step (3), the normalization process is performed using the following expression: Among them, W ij λ represents the normalized microseismic index. ij (t) includes three cases, all of which are positive indicators: λ ij ↑ (t)=[(Q max -Q ij ) / (Q max -Q min Negative index: λ ij ↓ (t)=[(Q min -Q ij ) / (Q max -Q min Two-way indicators: Where, λ ij ↑ (t) is a positive index, λ ij ↓ (t) is a negative indicator. It is a two-way indicator; Q max Q is the maximum value of the indicator. min The minimum value of the indicator; Q ij Let be the value of the ij-th microseismic index.

5. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 3, characterized in that, In step (3), AC is a negative index; A(t), a, A (b) S D , ΔF, η, Q t d s A positive indicator; Z map It is a two-way indicator.

6. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 1 or 2, characterized in that, The process of step (7) is as follows: The fitness values ​​are calculated as follows: In the above formula, X i Let F represent the position of the i-th sparrow. i =f(X) i Let ) represent the fitness value of the i-th sparrow, Z be the number of sparrows, and d be the dimension of the fitness function variable to be solved; The updated discoverers are as follows: In the above formula, t is the current iteration number; j = 1, 2, 3, ..., d represents the dimension; Let be the value of the i-th sparrow in the j-th dimension at the t-th iteration; α∈(0,1] is a random number; iter max The maximum number of iterations is denoted by ; ST∈[0.5,1.0] and W∈[0,1] are the safety threshold and warning value of the current position, respectively; Z is a random number that conforms to a normal distribution; L is a 1×d-dimensional matrix in which all elements are 1. When W≥ST, it means that a small number of sparrows have discovered the danger and are reminding the flock to move to a safe position. The following individuals have been added: The updated vigilant personnel are as follows: In the formula: It is the optimal location for the previous generation of the entire flock; f i β is the current fitness value of the sparrow; β is a random number following a normal distribution with mean 0 and variance 1; K is a random number in the range [-1, 1]; ε is a constant, then f g and f w These are the best and worst fitness values ​​in the previous generation of birds; when f i >f g When f indicates that sparrows around the flock are beginning to move towards the center; when f i =f g This indicates that the sparrow in the center of the flock should move closer to the other sparrows; The termination conditions are as follows: N≥N * (Formula 16).

7. The multi-parameter early warning method for coal and rock dynamic failure based on the SSA-SVM algorithm as described in claim 1 or 2, characterized in that, In step (8), during the training of the SVM model, the relationship between the predicted value and the actual value is compared, and the goodness of fit is calculated. If the goodness of fit is greater than 85%, the SVM model is used for multi-parameter early warning of coal and rock dynamic damage.

Images