A rockburst characteristic analysis method based on three indicators

Abstract

The present invention provides a rockburst characteristic analysis method based on three indicators, taking the rock mechanics uniaxial compression test as the object, and calculating the elastic energy index, elastic energy criterion index, and initial kinetic energy of rock debris ejection of different rocks. The Pearson correlation coefficient is selected to analyze the correlation between the indicators, and the random forest is used to find the characteristic importance between the three indicators and the destruction state / degree and determine the weight of each indicator. A universal formula for judging the degree of rock damage or rockburst is proposed to guide the prevention and control of rock dynamics disasters. It is suitable for effectively evaluating the destruction of rock samples in rock mechanics tests, and at the same time effectively collecting the ejected rock debris, and efficiently collecting them in different areas according to the specific ejection distance required.

Description

Technical Field

[0001] The invention relates to the technical field of rock mechanics experimental analysis, and in particular to a rockburst characteristic analysis method based on three indicators. Background Art

[0002] As rock projects such as mining or construction gradually move deeper, the frequent occurrence of engineering disasters and their increasingly serious disaster-causing degree have become a major challenge restricting engineering safety and sustainable development due to the combined effects of multiple factors such as high stress and high ground temperature. Rockburst, as one of the most typical energy release phenomena in deep rock engineering, its occurrence mechanism has not yet been fully clarified. The occurrence of rockburst is often accompanied by varying degrees of rock damage, and is accompanied by the ejection or throwing of a large number of rock fragments (rock chips) with significant kinetic energy. These fragments are not only highly sudden and highly destructive, but also directly threaten the lives of construction workers and the stable operation of engineering equipment.

[0003] In order to facilitate the study of complex deep engineering rock disasters, laboratory experiments are an important means to deeply explore the physical and mechanical properties of rocks and understand their behavior patterns and failure mechanisms under complex stress environments. Among them, the uniaxial compression test is a basic and key part of rock mechanics experiments. By simulating the mechanical response of rocks under a single axial pressure, it can accurately measure the key parameters of rocks such as uniaxial compressive strength and elastic modulus, providing an important basis for engineering design, geological disaster prevention and rock stability assessment. In the process of exploring the mechanism of rock dynamic disasters, a series of indicators have been proposed for the degree of rock damage or the severity of rock bursts, such as elastic energy index, elastic energy criterion index, initial kinetic energy of rock debris ejection, etc. Among them, the elastic energy index (W ET ) is a classification index of rockburst tendency, which reflects the ratio of elastic strain energy to plastic strain energy during the unloading process of rock after being loaded to a certain stress level; the elastic energy criterion index (PES) is the elastic energy stored in the rock before it reaches the peak strength, which is the ratio of the square of the uniaxial compressive strength of the rock to twice the unloading tangent modulus of the rock; the initial kinetic energy of rock fragments ejection (E0) reflects the severity of rock failure and the amount of energy released, and refers to the initial kinetic energy of rock fragments ejected from the rock matrix when the rock is destroyed.

[0004] At the same time, the collection and management of rock cuttings are often neglected during rock mechanics experiments, but their importance cannot be ignored. Rock cuttings, as fragments produced during the process of rock damage, not only carry key information such as the internal microstructure of the rock, crack propagation path and failure mode, but also directly reflect the mechanical properties and failure characteristics of the rock. Therefore, efficient and accurate collection of rock cuttings is of great significance for in-depth analysis of rock failure mechanisms, evaluation of rock mechanical properties and prediction of rock failure behavior.

[0005] Although these three indicators reflect the change of rock elastic strain energy, failure tendency and energy release during failure from different perspectives, they still cannot achieve a universal effect in judging the degree of rock failure. It can be said that each has its own reasons. In addition, traditional rock cuttings collection methods often have many shortcomings, such as mixed rock cuttings that are difficult to distinguish and large human errors. These problems seriously restrict the accuracy and reliability of rock mechanics experimental data. Summary of the invention

[0006] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a rockburst characteristic analysis method based on three indicators, which is suitable for effectively evaluating the damage of rock samples in rock mechanics experiments, and at the same time effectively catching the ejected rock debris, and efficiently collecting them in different regions according to the specific required ejection distance, so as to improve the accuracy of rock mechanics experimental data.

[0007] To achieve the above object, the present invention adopts the following technical solution:

[0008] A rockburst characteristic analysis method based on three indicators comprises the following steps:

[0009] Step 1: Build a rock cuttings collection device; perform uniaxial compression loading and unloading tests on rocks using an MTS hydraulic servo testing machine and record stress-strain curves;

[0010] Step 2: Calculate the elastic energy index W based on the stress-strain curve ET , elastic energy criterion index PES and initial kinetic energy of rock cuttings ejection E0;

[0011] Step 3: Use Pearson correlation coefficient to analyze elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock cuttings ejection;

[0012] Step 4: Use random forest to analyze the elastic energy index W ET , the characteristic importance between the elastic energy criterion index PES and the initial kinetic energy of rock fragments ejection E0 and the rock damage state / degree, and determine the index weight;

[0013] Step 5: Construct a linear regression model to obtain a universal formula for determining the degree of rock explosion;

[0014] Step 6: Use SPSS to perform multivariate linear regression analysis to verify the universal formula obtained by machine learning, and obtain the distribution histogram of the regression standardized residuals and the normal PP diagram of the regression standardized residuals.

[0015] Preferably, in step 2, the elastic energy index W ETIt is calculated by the ratio of elastic deformation energy to permanent deformation dissipation energy, and the calculation formula is as follows:

[0016]

[0017] In the formula, Φ c is the total strain energy, Φ sp is the elastic deformation energy, Φ st is the plastic deformation energy, f(ε) is the stress-strain curve expression, and ε is the strain.

[0018] Preferably, in step 2, the elastic energy criterion index PES is the elastic energy stored before the rock reaches the peak strength, and the calculation formula is the ratio of the square of the uniaxial compressive strength of the rock to the twice unloading tangent modulus, specifically:

[0019]

[0020] In the formula, σ c is the uniaxial compressive strength of rock; E s Unloading tangent modulus for rock.

[0021] Preferably, in step 2, the sample height is used as the initial height for horizontal projection calculation, and the distance from the center of each partition to the sample surface is used as the horizontal distance of the horizontal projection landing point; according to the weighed mass, the initial velocity v0 of the rock cuttings projection is calculated using the horizontal projection formula and the kinetic energy calculation formula:

[0022]

[0023] Where, h is the initial height of the cuttings from the test bench of the MTS hydraulic servo testing machine, g is the gravitational acceleration, t is the time it takes for the cuttings to fall into the cuttings collection device when thrown horizontally, v0 is the initial throwing speed of the horizontal throwing, and x is the horizontal distance between the point where the cuttings fall into the cuttings collection device and the specimen;

[0024] The initial kinetic energy E0 of rock cuttings ejection is calculated according to the kinetic energy calculation formula:

[0025]

[0026] Where, h is the initial height of the cuttings from the test bench of the MTS hydraulic servo testing machine, g is the gravitational acceleration, t is the time it takes for the cuttings to fall into the cuttings collection device when thrown horizontally, v0 is the initial throwing velocity, x is the horizontal distance from the point where the cuttings fall into the cuttings collection device to the specimen, and E0 is the initial kinetic energy of the cuttings ejection.

[0027] Preferably, in step three, the elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock fragments ejection is calculated as follows:

[0028]

[0029] Wherein, variable X and variable Y are elastic energy index W ET Any two of the following: elastic energy criterion index PES, initial kinetic energy of rock debris ejection E0; X i and Y i are the observed values ​​of variables X and Y, respectively. and are the means of variables X and Y respectively, and n is the number of observations.

[0030] Preferably, in step five, machine learning is used to construct a linear regression model to obtain a universal formula for determining the degree of rock explosion.

[0031] Preferably, the rock cuttings collection device comprises a pillar, a support platform, a collection module and a steel ring; the pillar is fixed on the bottom end surface of the support platform; a first through hole is provided in the middle of the support platform; the collection module is laid on the upper part of the support platform; a second through hole corresponding to the first through hole is provided in the middle of the collection module, and the axes of the first through hole and the second through hole coincide with each other; the steel ring is installed on the upper part of the collection module; the axis of the steel ring coincides with the axis of the first through hole; a plurality of steel rings are configured; the steel rings are arranged in the same horizontal plane, and the diameter gradually increases from the axis to the outside, and the area between two adjacent steel rings is a connecting area; reinforcing ribs are fixedly arranged between two adjacent steel rings.

[0032] Preferably, it also includes a shielding ring; the shielding ring is sleeved on the outermost side of the steel ring, and the lower end surface is in contact with the collection module; the shielding ring and the steel ring are fixedly connected by reinforcing ribs.

[0033] Preferably, the collecting module comprises polar fleece cloth and Velcro; the second through hole is arranged on the polar fleece cloth; and the polar fleece cloth is mounted on the supporting platform by Velcro.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] This analysis method takes the uniaxial compression test of rock mechanics as the object, and calculates the elastic energy index, elastic energy criterion index, and initial kinetic energy of rock chip ejection of different rocks. The Pearson correlation coefficient is selected to analyze the correlation between the indicators, and the random forest is used to find the characteristic importance between the three indicators and the damage state / degree and determine the weight of each indicator. A universal formula for judging the degree of rock damage or rock burst is proposed to guide the prevention and control of rock dynamic disasters. At the same time, the present invention has developed a rock chip collection device, the construction basis of which is derived from the fine capture of the dynamic characteristics of rock chips in the pre-experimental stage. The use of the multi-region segmented rock chip collection device of the present invention significantly improves the efficiency and accuracy of rock chip collection, while effectively weakening the potential deviation of experimental results caused by human operation factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of a rockburst characteristic analysis method based on three indicators and a matching partitioned rock debris collection device of the present invention;

[0037] Figure 2 To calculate the elastic energy index (W ET ) schematic diagram;

[0038] Figure 3 Schematic diagram for calculating the elastic energy (PES) stored in rock before reaching peak strength;

[0039] Figure 4 The schematic diagram for calculating the initial ejection kinetic energy of rock fragments in different regions;

[0040] Figure 5 Data visualization of random forest analysis of feature importance for each indicator;

[0041] Figure 6 This is the distribution histogram of the regression standardized residuals obtained by multiple linear regression analysis using SPSS;

[0042] Figure 7 This is the normal PP plot of the regression standardized residuals obtained by multiple linear regression analysis using SPSS;

[0043] Figure 8 It is a schematic diagram of the structure of the rock cuttings collection device in the present invention.

[0044] in:

[0045] 1. Pillar; 2. Steel ring; 3. Reinforcement ribs; 4. Polar fleece cloth; 5. Shielding ring; 6. Support platform. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings.

[0047] like Figure 1 As shown, a rockburst characteristic analysis method based on three indicators includes the following steps:

[0048] Step 1: Build a rock debris collection device; design uniaxial compression loading and unloading tests on four different rocks according to calculation requirements; use an MTS hydraulic servo testing machine to perform uniaxial compression loading and unloading tests on rocks, and record the stress-strain curve during the entire test process; since the uniaxial compressive strength of the same rock cannot be determined, choose to unload at 70%-80% of the estimated strength, and load to failure when unloading to 0%-1%;

[0049] Step 2: Calculate the elastic energy index W based on the stress-strain curve ET, elastic energy criterion index PES and initial kinetic energy of rock cuttings ejection E0;

[0050] Step 3: Use Pearson correlation coefficient to analyze elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock cuttings ejection;

[0051] Step 4: Use random forest to analyze the elastic energy index W ET , the characteristic importance between the elastic energy criterion index PES and the initial kinetic energy of rock fragments ejection E0 and the rock damage state / degree, and determine the index weight;

[0052] The test results and calculation results are organized in an Excel file. The first row is the feature and target variable, followed by the failure load, elastic energy index, stored elastic energy, initial kinetic energy of rock fragments in different regions, and degree of damage (0 represents a low degree of damage, 1 represents a medium degree of damage, and 2 represents a high degree of damage. This index can also correspond to the degree of rock burst), and the following are specific data. It is important to ensure that all data are in digital format so that the feature importance calculation can be performed directly.

[0053] Step 2: Use Python to perform feature analysis on the dataset. First, import the required libraries such as random forest library, machine learning library, numpy library, pandas library, etc. Then load the dataset and read all the data. Separate the features and target variables. Preliminary processing has been done here. The last column is the target variable, and the rest are features.

[0054] X = data.iloc[:,:-1]; Y = data.iloc[:,-1] (here X is the feature and Y is the target variable).

[0055] Then create a random forest classifier and train the model to obtain the feature importance and the corresponding feature names.

[0056] rf=RandomForestClassifier(n_estimators=100,random_state=42)

[0057] rf.fit(X,Y)

[0058] feature_names=X.columns.tolist()

[0059] feature_importances_df=pd.DataFrame({'Feature':feature_names,'Importance':feature_importances})

[0060] Secondly, you can sort the most important features as needed, and after output, you can visualize the importance of the features as needed, and output a more intuitive bar chart. The purpose of this step is to further determine the importance between several features and the target variable, and give an intuitive weight to verify the correctness of the Peason correlation coefficient.

[0061] Step 5. According to the results of correlation analysis and importance analysis, they are sorted into an Excel table. Each column is a corresponding characteristic indicator, and the last column is the degree of damage (here 0 represents a low degree of damage, 1 represents a medium degree of damage, and 2 represents a large degree of damage. This indicator can also correspond to the degree of rock burst). Use Python machine learning to build a linear regression model and obtain a universal formula for determining the degree of rock explosion to guide rock dynamics disaster prevention and control work.

[0062] Define the features and target variables, using the data in the prepared Excel file as the basis. Create a linear regression model and conduct training.

[0063] X=data[['Failure load','Elastic energy index','Stored elastic energy','Total kinetic energy of rock debris ejection','1E','2E','3E','4E']]

[0064] y = data['extent of damage']

[0065] lr=LinearRegression()

[0066] lr.fit(X,y)

[0067] Then the coefficients of the linear regression equation can be obtained as needed.

[0068] print("Linear Regress ion Coefficients(including intercept):")

[0069] print(f"Intercept:{lr.intercept_}")

[0070] print(f"Coefficients:{lr.coef_}")

[0071] The linear regression equation can be expressed as: y = intercept + coef_0*x0 + coef_1*x1 + coef_2*x2 + coef_3*x3 + coef_4*x4 + coef_5*x5 + coef_6*x6 + coef_7*x7

[0072] x0-7 in the equation are the corresponding features.

[0073] Of course, if you need to derive a polynomial equation, you can also use the corresponding polynomial model for analysis. Here, a quadratic polynomial is used for regression simulation:

[0074] poly=PolynomialFeatures(degree=2)

[0075] X_poly=poly.fit_transform(X)

[0076] lr_poly=LinearRegression()

[0077] lr_poly.fit(X_poly,y)

[0078] Step 6. Based on the linear regression equation obtained in step 6, use big data analysis to further verify its correctness; use SPSS to perform multivariate linear regression analysis to verify the universal formula obtained by machine learning, and at the same time obtain the distribution histogram of the regression standardized residuals and the normal PP diagram of the regression standardized residuals to facilitate more intuitive judgment.

[0079] In this embodiment, in step 2, the elastic energy index (WET) is used to indicate the rock impact tendency (see Appendix Figure 2 ); elastic energy index W ET By calculating the ratio of elastic deformation energy to permanent deformation dissipation energy, during the loading stage of the rock, as the elastic strain energy accumulated inside the rock increases, the energy dissipated due to irreversible deformation decreases accordingly, resulting in the rock showing a stronger impact tendency. The calculation formula is as follows:

[0080]

[0081] In the formula, Φ c is the total strain energy, Φ sp is the elastic deformation energy, Φ st is the plastic deformation energy, f(ε) is the stress-strain curve expression, and ε is the strain.

[0082] In this embodiment, in step 2, the elastic energy (PES) stored in the rock before reaching the peak strength under uniaxial compression is calculated (see Appendix Figure 3 ). The occurrence of rockburst usually depends on two core conditions: first, the surrounding environment of the rock mass must have the energy reserve conditions required to trigger rockburst; second, the rock mass itself must have a high energy storage capacity. The combined effect of these two conditions provides the necessary conditions for the occurrence of rockburst. The calculation formula is the ratio of the square of the uniaxial compressive strength of the rock to the twice unloading tangent modulus, specifically:

[0083]

[0084] In the formula, σ c is the uniaxial compressive strength of rock; E s Unloading tangent modulus for rock.

[0085] In this embodiment, the sample height is used as the initial height for horizontal projection calculation, and the distance from the center of each partition to the sample surface is used as the horizontal distance of the horizontal projection landing point; according to the weighed mass, the initial velocity v0 of the rock cuttings projection is calculated using the horizontal projection formula and the kinetic energy calculation formula (see the attached manual). Figure 4 ):

[0086]

[0087] Where, h is the initial height of the cuttings from the test bench of the MTS hydraulic servo testing machine, g is the gravitational acceleration, t is the time it takes for the cuttings to fall into the cuttings collection device when thrown horizontally, v0 is the initial throwing speed of the horizontal throwing, and x is the horizontal distance between the point where the cuttings fall into the cuttings collection device and the specimen;

[0088] The initial kinetic energy E0 of rock cuttings ejection is calculated according to the kinetic energy calculation formula:

[0089]

[0090] Where, h is the initial height of the cuttings from the test bench of the MTS hydraulic servo testing machine, g is the gravitational acceleration, t is the time it takes for the cuttings to fall into the cuttings collection device when thrown horizontally, v0 is the initial throwing velocity, x is the horizontal distance from the point where the cuttings fall into the cuttings collection device to the specimen, and E0 is the initial kinetic energy of the cuttings ejection.

[0091] In this embodiment, in step three, the Pearson correlation coefficient is used, also known as the Pearson correlation coefficient or the product-moment correlation coefficient, which is a statistical measurement tool specifically used to evaluate the strength and direction of the linear relationship between two variables X and Y. The value range of this coefficient is from -1 to 1 (including -1 and 1), where the closer the absolute value of the coefficient is to 1, the more significant the linear relationship between X and Y; conversely, if the absolute value of the coefficient is closer to 0, it implies that the linear relationship between the two is weaker. Specifically, a positive value indicates a positive correlation, that is, an increase in one variable is accompanied by an increase in another variable; and a negative value indicates a negative correlation, that is, an increase in one variable is accompanied by a decrease in another variable. Determine the elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock fragments ejection is calculated as follows:

[0092]

[0093] Wherein, variable X and variable Y are elastic energy index W ET , elastic energy criterion index PES, initial kinetic energy of rock cuttings ejection

[0094] ——Any two of E0; X i and Y i are the observed values ​​of variables X and Y, respectively, X and Y are the means of variables X and Y, respectively, and n is the number of observations.

[0095] In this embodiment, in step five, machine learning is used to construct a linear regression model to obtain a universal formula for determining the degree of rock explosion.

[0096] In order to specifically introduce the innovative effect of the first aspect of the present invention, a specific embodiment is described, taking the use of the device of the present invention in a real rock mechanics experiment as a specific object.

[0097] Step 1: Through a series of uniaxial compression loading and unloading tests, 77 sets of data available for analysis were obtained, including the failure load, elastic energy index, stored elastic energy, initial kinetic energy of rock fragments in different regions, and intuitively judged degree of damage.

[0098] Step 2: Use Pearson's correlation coefficient to conduct pairwise correlation analysis of each data indicator, mainly to see the correlation between the degree of damage and other parameters.

[0099] Table 1 Pearson correlation coefficient analysis between indicators

[0100]

[0101] Here we can see that the ranking by correlation coefficient is: WET-0.613999, E-0.496643359, σ-0.495004, PES-0.430407, IVE-0.429706, IIIE-0.376919, IE-0.253583, IIE-0.209115

[0102] It can be seen that the correlation coefficients between all indicators and the degree of damage are greater than 0.2, indicating that each characteristic indicator has an intuitive impact on the judgment of the degree of rock damage.

[0103] Step 3: Since one of the innovative points of the present invention is to analyze the initial kinetic energy of the cuttings in different regions, the initial total kinetic energy E of the cuttings is discarded for analysis. The random forest method is used to find the weight of each characteristic index. The results are as follows (see the attached manual for the importance of visual features). Figure 5 ):

[0104] Table 2 Random forest analysis of feature importance of each indicator

[0105]

[0106] From the calculation results of feature importance weights, we can see that in the whole process of judging the degree of rock damage, the importance of several indicators from large to small are W ET , IVE, IIIE, IIE, σ, PES, ⅠE, which can be reflected in the regression equation below.

[0107] Step 4: Use Python machine learning to build a linear regression model and obtain a universal formula for estimating the degree of damage / rockburst severity.

[0108] Python calculates the result:

[0109] Intercept:-0.8036962961058463

[0110] Coefficients:[3.07482257e-03 1.87755185e-01 6.94584511e-031.69089281e+02

[0111] 9.91260805e+01 5.87593133e+01 1.23477019e+01]

[0112] That is, the universal linear regression equation is:

[0113] A=-0.8036962961058463+3.07482257e-03×σ+1.87755185e-01×W ET +6.94584511e-03×

[0114] PES+1.69089281e+02×1E+9.91260805e+01×2E+5.87593133e+01×3E+

[0115] 1.23477019e+01×4E

[0116] Where A is the degree of damage; σ is the damage load, MPa; WET is the elastic energy index; PES is the stored elastic energy, kJ / m3; 1-4E is the initial ejection kinetic energy of the cuttings in the 1st to 4th partitions, J.

[0117] Step 5: Use SPSS to perform multiple linear regression analysis to verify the universal formula obtained by machine learning and obtain the distribution histogram of the regression standardized residuals (see Appendix Figure 6 ) and the normal PP plot of the regression standardized residuals (see Appendix Figure 7 ).

[0118] Table 3 SPSS multiple linear regression analysis verification results

[0119]

[0120]

[0121] Obviously, the results obtained are basically consistent with the results of machine learning in step 4.

[0122] Step 6: You can use real test data for re-verification. The real test data is as follows:

[0123] Table 4 Calculation of actual test data using regression equation

[0124]

[0125] Obviously, the calculated value based on the universal formula is basically consistent with the actual degree of damage, so it can be judged that the regression formula is correct.

[0126] In this embodiment, if Figure 8 As shown, the rock cuttings collection device includes a pillar 1, a support platform, a collection module and a steel ring 2; the support platform is made of a strong and stable metal material to ensure that it can withstand the overall weight of the device and the impact force during the test; the pillar 1 is fixed on the bottom surface of the support platform; a first through hole is provided in the middle of the support platform; the collection module is laid on the upper part of the support platform; a second through hole corresponding to the first through hole is provided in the middle of the collection module, and the axes of the first through hole and the second through hole coincide with each other; the steel ring 2 is installed on the upper part of the collection module; the axis of the steel ring 2 coincides with the axis of the first through hole; multiple steel rings 2 are configured; the steel rings 2 are arranged in the same horizontal plane, and the diameter gradually increases from the axis to the outside, and the area between two adjacent steel rings 2 is the connection area; reinforcing ribs 3 are fixedly provided between two adjacent steel rings 2; the pillar 1 is made of an existing high-strength, corrosion-resistant metal material (stainless steel) and is an adjustable height pillar 1, so as to adjust the position and angle of the collection area module according to the test requirements.

[0127] In this embodiment, a shielding ring is also included; the shielding ring sleeve 5 is connected to the outermost side of the steel ring 2, and the lower end surface is in contact with the collection module; the shielding ring and the steel ring 2 are fixedly connected by reinforcing ribs 3.

[0128] In this embodiment, the collection module includes a polar fleece cloth 4 and a Velcro; the second through hole is provided on the polar fleece cloth 4; the polar fleece cloth 4 is mounted on the support platform by the Velcro. The polar fleece cloth 4 has the characteristics of energy absorption and splash prevention, which can effectively slow down the impact speed of the rock cuttings and prevent them from rebounding, reduce the noise and damage when the rock cuttings impact, and further improve the collection effect; the size of the polar fleece cloth 4 is larger than the steel ring 2, which is convenient for bonding.

[0129] When in use, several areas are pre-divided on the polar fleece cloth 4 according to the expected range of rock debris ejection, and the size of the area can be set according to experience or preliminary estimation; the polar fleece cloth 4 with divided areas is laid on the test bench of the MTS hydraulic servo test machine to ensure that the rock debris can fall on the flocking cloth and be effectively collected during the test. At the same time, ensure that the edges of the flocking cloth are firmly fixed to avoid movement or deformation during the test; conduct preliminary tests: start the MTS hydraulic servo test machine to load the rock sample until it is destroyed. Observe and record the distribution of rock debris on the polar fleece cloth 4, especially pay attention to which areas the rock debris is mainly concentrated in and the relative density of each area. By measuring the distance from the rock debris to the center of the sample (rock sample), make a refined partition of the polar fleece cloth 4. Install the collection module at the MTS hydraulic servo test machine, adjust the height of the pillar 1, place the sample on the test bench of the MTS hydraulic servo test machine from the first through hole, install the collection module and the steel ring 2 and fix them, and start the test. After the test, handle the collected rock debris as needed, remove the polar fleece cloth 4 to carry out other subsequent analyses such as weighing.

[0130] The specific object is to use the device of the present invention in a real rock mechanics experiment.

[0131] Step 1: Divide the receiving area into four receiving areas: Area I 0-4 cm, Area II 4-8 cm, Area III 8-16 cm, and Area IV>16 cm; the overall height of the cuttings collection device is 7 cm.

[0132] Step 2: Set the rock chip collection device in the center of the test bench of the MTS hydraulic servo test machine and adjust the height to match the test bench height. Select a cylindrical rock sample with a diameter of 50 mm and a height of 50 mm, place it in the center of the test bench, and load it with a suitable and reasonable experimental plan. Stop loading when the sample is damaged by loading, restore the hydraulic column, and remove the sample and the receiving device.

[0133] Step three: weigh the mass of the rock cuttings in each area separately, using an electronic scale with an accuracy of 0.1g.

[0134] Step 4: Calculate according to the method for calculating the initial kinetic energy of rock cuttings mentioned above in the present invention.

[0135] The experimental calculation results of this specific embodiment are presented in the form of a table below:

[0136] Table 5. Analysis of the damage of coarse sandstone

[0137]

[0138]

Claims

1. A rockburst characteristic analysis method based on three indicators, characterized in that: The following steps are involved: Step 1: Build a rock cuttings collection device; perform uniaxial compression loading and unloading tests on rocks using an MTS hydraulic servo testing machine and record stress-strain curves; Step 2: Calculate the elastic energy index W based on the stress-strain curve ET , elastic energy criterion index PES and initial kinetic energy of rock cuttings ejection E0; Step 3: Use Pearson correlation coefficient to analyze elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock cuttings ejection; Step 4: Use random forest to analyze the elastic energy index W ET , the characteristic importance between the elastic energy criterion index PES and the initial kinetic energy of rock fragments ejection E0 and the rock damage state / degree, and determine the index weight; Step 5: Construct a linear regression model to obtain a universal formula for determining the degree of rock explosion; Step 6: Use SPSS to perform multiple linear regression analysis to verify the universal formula obtained by machine learning, and obtain the distribution histogram of the regression standardized residuals and the normal PP diagram of the regression standardized residuals; In step 2, the elastic energy index W ET It is calculated by the ratio of elastic deformation energy to permanent deformation dissipation energy, and the calculation formula is as follows: ， ， ； ； In the formula, Φ c is the total strain energy, Φ sp is the elastic deformation energy, Φ st is the plastic deformation energy, f(ε) is the stress-strain curve expression, and ε is the strain; In step 2, the elastic energy criterion index PES is the elastic energy stored in the rock before it reaches the peak strength. The calculation formula is the ratio of the square of the uniaxial compressive strength of the rock to the twice unloading tangent modulus, specifically: ； In the formula, σ c is the uniaxial compressive strength of rock; E s Unloading tangent modulus for rock.

2. A rockburst characteristic analysis method based on three indicators as claimed in claim 1, characterized in that: In step 2, the sample height is used as the initial height for horizontal projection calculation, and the distance from the center of each partition to the sample surface is used as the horizontal distance of the horizontal projection landing point; according to the weighed mass, the initial velocity v0 of the rock chip projection is calculated using the horizontal projection formula and the kinetic energy calculation formula: ， ， ； Where, h is the initial height of the cuttings from the test bench of the MTS hydraulic servo testing machine, g is the gravitational acceleration, t is the time it takes for the cuttings to fall into the cuttings collection device when thrown horizontally, v0 is the initial throwing speed of the horizontal throwing, and x is the horizontal distance between the point where the cuttings fall into the cuttings collection device and the specimen; The initial kinetic energy E0 of rock cuttings ejection is calculated according to the kinetic energy calculation formula: ； Where E0 is the initial kinetic energy of rock cuttings ejection.

3. A rockburst characteristic analysis method based on three indicators as claimed in claim 1, characterized in that: In step 3, the elastic energy index W ET , the correlation between the elastic energy criterion index PES and the initial kinetic energy E0 of rock fragments ejection is calculated as follows: ； Wherein, variable X and variable Y are elastic energy index W ET Any two of the following: elastic energy criterion index PES, initial kinetic energy of rock debris ejection E0; X i and Y i are the observed values ​​of variables X and Y, respectively. and are the means of variables X and Y respectively, and n is the number of observations.

4. A rockburst characteristic analysis method based on three indicators as claimed in claim 1, characterized in that: In step five, machine learning is used to construct a linear regression model to obtain a universal formula for determining the degree of rock explosion.

5. A rockburst characteristic analysis method based on three indicators as claimed in any one of claims 1 to 4, characterized in that: The rock debris collection device comprises a pillar (1), a support platform, a collection module and a steel ring (2); the pillar (1) is fixed on the bottom end surface of the support platform; a first through hole is arranged in the middle of the support platform; the collection module is laid on the upper part of the support platform; a second through hole corresponding to the first through hole is arranged in the middle of the collection module, and the axes of the first through hole and the second through hole coincide with each other; the steel ring (2) is installed on the upper part of the collection module; the axis of the steel ring (2) coincides with the axis of the first through hole; a plurality of steel rings (2) are arranged; the steel rings (2) are arranged in the same horizontal plane, and the diameter gradually increases from the axis to the outside, and the area between two adjacent steel rings (2) is a receiving area; reinforcing ribs (3) are fixedly arranged between two adjacent steel rings (2).

6. A rockburst characteristic analysis method based on three indicators as claimed in claim 5, characterized in that: It also includes a shielding ring; the shielding ring sleeve (5) is connected to the outermost side of the steel ring (2), and the lower end surface is in contact with the collection module; the shielding ring and the steel ring (2) are fixedly connected by reinforcing ribs (3).

7. A rockburst characteristic analysis method based on three indicators as claimed in claim 5, characterized in that: The collecting module comprises a polar fleece cloth (4) and Velcro; the second through hole is arranged on the polar fleece cloth (4); and the polar fleece cloth (4) is mounted on the supporting platform via the Velcro.

Images