CN115130375A - Rock burst intensity prediction method - Google Patents

Rock burst intensity prediction method Download PDF

Info

Publication number
CN115130375A
CN115130375A CN202210686507.1A CN202210686507A CN115130375A CN 115130375 A CN115130375 A CN 115130375A CN 202210686507 A CN202210686507 A CN 202210686507A CN 115130375 A CN115130375 A CN 115130375A
Authority
CN
China
Prior art keywords
intensity
rock burst
rockburst
prediction
xgboost
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210686507.1A
Other languages
Chinese (zh)
Inventor
周喻
高永涛
朱强
孙贝贝
董翔
郭雨茜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Science and Technology Beijing USTB
Original Assignee
University of Science and Technology Beijing USTB
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by University of Science and Technology Beijing USTB filed Critical University of Science and Technology Beijing USTB
Priority to CN202210686507.1A priority Critical patent/CN115130375A/en
Publication of CN115130375A publication Critical patent/CN115130375A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00Computer-aided design [CAD]
    • G06F30/20Design optimisation, verification or simulation
    • G06F30/27Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • G06N20/20Ensemble learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/004Artificial life, i.e. computing arrangements simulating life
    • G06N3/006Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/14Force analysis or force optimisation, e.g. static or dynamic forces

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Medical Informatics (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Geometry (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Computer Hardware Design (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a rock burst intensity prediction method, and belongs to the technical field of geotechnical engineering and underground excavation engineering. The method comprises the following steps: determining the evaluation index of rock burst intensity prediction and rock burst intensity classification; establishing a rockburst intensity prediction sample data set according to the determined evaluation index of rockburst intensity prediction and the rockburst intensity classification; expanding a few classes in the dataset; based on the expanded data set, optimizing the hyper-parameters in the XGboost integration algorithm by using a cross validation and African bald eagle optimization algorithm, and establishing an AVOV-XGboost rockburst intensity prediction model; and inputting the rock burst sample to be predicted into the established AVOV-XGboost rock burst intensity prediction model to obtain a rock burst intensity grade prediction value. By adopting the method and the device, the accuracy of the rock burst intensity grade prediction result can be improved.

Description

Rock burst intensity prediction method
Technical Field
The invention relates to the technical field of geotechnical engineering and underground excavation engineering, in particular to a rockburst intensity prediction method.
Background
Rock burst is a geological disaster in which surrounding rocks accumulating high elastic strain energy are subjected to a rock ejection phenomenon due to excavation disturbance in underground engineering. The occurrence of the rockburst disaster usually damages mechanical equipment, delays the progress of a project, even threatens the life safety of personnel, and causes serious economic loss. With the rapid development of infrastructure construction in China, rockburst disasters widely occur in underground excavation projects such as mines, tunnels, hydropower stations and the like, and the occurrence of rockburst has outbreak, uncertainty and strong destructiveness. Therefore, studies on the prediction and prevention of rockburst disasters are not slow.
In recent decades, many domestic and foreign scholars have conducted a great deal of research on rock burst prediction methods in deep underground engineering. In particular, the research and application of machine learning algorithms in the field of rock burst prediction are increasing. For example, the machine learning algorithms have better effects to a certain extent by using a support vector machine, a decision tree, a least square support vector machine, naive Bayes, a random forest, a gradient elevator, an artificial neural network and the like. At present, the research of a machine learning algorithm on rock burst prediction mainly focuses on selection of an algorithm model, selection of rock burst evaluation indexes, optimization of model hyper-parameters and data set preprocessing. Machine learning algorithms can be divided into single models and integrated models based on the number of classifiers. The single model has low generalization capability, can not obtain optimal solutions to all problems, and the prediction performance of the single model changes along with the change of engineering environment or input parameters, so that the rock burst disaster prediction effect in the current underground excavation engineering is poor. In addition, the phenomenon that the quantity difference of rock burst data of different levels is large generally exists in data sets collected by scholars at home and abroad, and rock burst prediction is not well influenced by category imbalance.
Disclosure of Invention
The embodiment of the invention provides a rockburst intensity prediction method, which can improve the accuracy of rockburst intensity level prediction results. The method comprises the following steps:
determining the evaluation index of rock burst intensity prediction and rock burst intensity classification;
establishing a rockburst intensity prediction sample data set according to the determined evaluation index of rockburst intensity prediction and the rockburst intensity classification;
expanding a few classes in the dataset;
based on the expanded data set, optimizing the hyper-parameters in the XGboost integration algorithm by using a cross validation and African bald eagle optimization algorithm, and establishing an AVOV-XGboost rockburst intensity prediction model;
and inputting the rock burst sample to be predicted into the established AVOV-XGboost rock burst intensity prediction model to obtain the grade prediction value of the rock burst intensity.
Further, the evaluation indexes of the rockburst intensity prediction include: uniaxial compressive strength sigma c Uniaxial tensile strength σ t Stress coefficient σ θc Brittle coefficient σ ct And elastic energy index W et Wherein σ θ The maximum tangential stress is indicated.
Further, the classification of the rockburst intensity comprises: no rock burst i, slight rock burst ii, medium rock burst iii and strong rock burst iv.
Further, the expanding the few classes in the dataset includes:
and expanding the minority classes in the data set by adopting a synthesis minority class oversampling technology.
Further, after the expanding the few classes in the dataset, the method further comprises:
and carrying out standardization processing on the data in the expanded data set.
Further, the establishing of the AVOV-XGboost rockburst intensity prediction model based on the expanded data set by optimizing the hyperparameters in the XGboost integration algorithm by using the cross validation and the African bald eagle optimization algorithm comprises the following steps:
dividing the expanded data set into a training set and a testing set;
optimizing the hyper-parameters in the XGboost integration algorithm by using a 5-fold cross validation and African bald eagle optimization algorithm on a training set; wherein the hyper-parameters comprise: learning rate, minimum leaf node sample weight sum, maximum depth of tree, minimum loss reduction required for node splitting, and sub-sampling rate; in the optimizing process, the average value of the prediction accuracy rate obtained by 5-fold cross validation calculation is used as the judgment standard for determining the hyper-parameter.
Further, after establishing the AVOV-XGboost rockburst intensity prediction model, the method comprises the following steps:
developing an intelligent rockburst intensity prediction platform by using MATLAB software; wherein the prediction interface of the platform comprises: 5 evaluation index input boxes and 4 rockburst intensity early warning lamps.
The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:
1. the cross-validation strategy and the African bald eagle optimization algorithm are adopted to optimize the super-parameters of the XGboost integration algorithm, and the optimized XGboost integration algorithm is used for carrying out rockburst intensity grade prediction, so that the accuracy of rockburst intensity grade prediction results can be improved, and the problem of poor rockburst disaster prediction effect in the current underground excavation engineering is solved; the optimized model of the African bald eagle optimization algorithm has the advantages of rapidness and high efficiency;
2. a few classes in the established rockburst intensity prediction sample data set are expanded by adopting a synthetic few-class oversampling technology so as to solve the problem of imbalance in the data set; then, carrying out standardization processing on the data in the expanded data set, eliminating dimension difference and reducing calculation expense;
3. the method does not need deep professional knowledge and rich engineering experience possessed by field workers, does not need complicated mechanical calculation or numerical simulation analysis, can obtain the rock burst intensity grade prediction value by inputting the evaluation index corresponding to the rock burst sample to be predicted in the engineering into the AVOV-XGboost rock burst intensity prediction model provided by the embodiment, has the advantages of strong practicability and high efficiency, and provides better guiding significance for rock burst prediction of underground excavation engineering.
4. The AVOV-XGboost rockburst intensity prediction model provided by the embodiment utilizes MATLAB software to develop a rockburst intensity prediction platform, and can be conveniently used by field workers.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic flow diagram of a rockburst intensity prediction method according to an embodiment of the present invention;
fig. 2 is a detailed flowchart schematic diagram of a rockburst intensity prediction method according to an embodiment of the present invention;
fig. 3 is a frequency histogram of 5 evaluation indicators according to an embodiment of the present invention;
FIG. 4 is a pie chart of a distribution of a data set provided by an embodiment of the present invention;
FIG. 5 is a schematic flow chart of an iteration of the African bald eagle optimization algorithm provided by the embodiment of the invention;
FIG. 6 is a diagram illustrating a fitness iteration curve according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a confusion matrix on a training set according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of a confusion matrix on a test set according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of a log-in interface of the intelligent rockburst intensity prediction platform according to the embodiment of the present invention;
fig. 10 is a schematic diagram of a prediction interface of the intelligent rockburst intensity prediction platform according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
As shown in fig. 1 and fig. 2, an embodiment of the present invention provides a method for predicting a rockburst intensity, including:
s101, determining an evaluation index of rock burst intensity prediction and rock burst intensity grading;
the factors influencing the occurrence of rock burst mainly include rock properties, stress state of surrounding rock, excavation mode, underground water, buried depth, ground stress and the like, and mutually influence each other in time and space. The relationship between the rockburst intensity and the rockburst effect factor is highly nonlinear. The two reasons enable the selected index to reflect important factors influencing the rock burst as much as possible when the rock burst intensity prediction evaluation index is selected. Selecting uniaxial compressive strength sigma according to the rock burst occurrence mechanism and on the basis of single-index and multi-index criteria adopted for predicting the rock burst intensity at present c Uniaxial tensile strength σ t Stress coefficient sigma θ σ c Brittle coefficient sigma c σ t And elastic energy index W et The 5 indexes are used as evaluation indexes for rock burst intensity prediction, wherein sigma θ The maximum tangential stress is indicated. The indexes not only can comprehensively react to influence internal causes and external causes of rock burst generation, but also are easy to obtain.
In the embodiment, the rockburst intensity can be divided into 4 grades according to the rockburst destruction intensity, the destruction mode and the scale, wherein the 4 grades are respectively a non-rockburst I, a slight rockburst II, a medium rockburst III and a strong rockburst IV (coded by 0-3), and the specific grading standard is shown in table 1.
TABLE 1 rockburst severity grading Standard
Figure BDA0003699877430000041
S102, establishing a rockburst intensity prediction sample data set according to the determined evaluation index and the rockburst intensity classification of the rockburst intensity prediction;
in the embodiment, typical rock burst case data of tunnels, mines and hydropower station engineering at home and abroad are obtained according to the determined evaluation index and rock burst intensity classification of rock burst intensity prediction, and a rock burst intensity prediction sample data set is established; the acquired rock burst case data are shown in table 2.
TABLE 2 rockburst case data
Figure BDA0003699877430000042
Figure BDA0003699877430000051
Figure BDA0003699877430000061
In this embodiment, to solve the distribution of each index, a frequency histogram is plotted and statistical features are calculated, as shown in fig. 3 and table 3. As can be seen from FIG. 3 and Table 3, the uniaxial compressive strength σ c And uniaxial tensile strength σ t Brittle coefficient sigma ct And elastic energy index W et All basically show a tendency of right deviation and stress coefficient sigma θc The distribution is more symmetrical.
TABLE 3 statistical characterization of the indices
Figure BDA0003699877430000071
S103, expanding a few classes in the data set;
in the embodiment, a few classes in the data set are expanded by adopting a composite few class oversampling technology so as to solve the problem of imbalance in the data set; then, the data in the expanded data set is standardized, dimension difference is eliminated, and calculation expense is reduced.
In this embodiment, the proportions of the four rock burst intensity levels in the established rock burst intensity prediction sample data set are respectively 27 rock burst-free samples (accounting for 18%), 47 slight rock bursts (accounting for 31.3%), 57 medium rock bursts (accounting for 38%), and 19 strong rock bursts (accounting for 12.7%), as shown in fig. 4. Obviously, the data set has the problem of class imbalance, and the proportion of rock burst-free and strong rock burst samples is only 18% and 12.7%. Imbalance of classes often results in excessive attention to the majority classes of the model, resulting in poor training of the model on the minority classes. Therefore, in this embodiment, a few types of oversampling technologies are synthesized to expand two types of samples, namely, a non-rockburst sample and a strong rockburst sample, and the specific operation flow is as follows:
(1) calculating the euclidean distance from each sample x to all samples in the respective few types of sample sets by using equation (1) for each sample x in the two types of samples of the non-rockburst and the strong rockburst to obtain K (for example, K is 5) neighbors of the sample x;
Figure BDA0003699877430000072
in the formula (1), x 1i Is x 1 The ith characteristic (i.e., evaluation index); x is the number of 2i Is x 2 The ith feature of (1); n is a characteristic number, D 12 Is x 1 And x 2 The euclidean distance between them.
(2) Setting sampling multiplying power of two minority classes of rockburst-free rock burst and strong rock burst as 2 times and 3 times respectively;
(3) for a few types of samples in the non-rockburst and the strong rockburst, respectively randomly selecting 2 neighbors and 3 neighbors from 5 neighbors;
(4) for each selected neighbor sample, a new sample is generated using equation (2):
x new =x+rand()*|x-x k | (2)
in the formula (2), x new A newly generated minority sample; rand () is a random number of (0, 1); x is the number of k K first neighbor samples of x;
finally, the number of classes of non-and strong rock bursts in the data set is expanded to 54 and 57, respectively.
After the problem of unbalanced category of the original data set is solved, the data in the expanded data set is standardized by using equation (3) so as to eliminate dimension difference and reduce calculation overhead.
Figure BDA0003699877430000081
In formula (3): x ij A value normalized to the jth sample in the ith index; x is the number of ij The actual value of the jth sample in the ith index is taken as the actual value of the jth sample in the ith index;
Figure BDA0003699877430000082
and
Figure BDA0003699877430000083
the maximum value and the minimum value of the ith index are respectively.
In this embodiment, the synthesis of few classes of oversampling techniques and normalization belong to the preprocessing operation.
S104, optimizing a super parameter in an extreme gradient boost (XGboost) integration algorithm by using a cross validation and African bald eagle optimization Algorithm (AVOV) based on the expanded data set, and establishing an AVOV-XGboost rockburst intensity prediction model; the method specifically comprises the following steps:
a1, dividing the expanded data set into a training set and a testing set;
in the embodiment, the data set after the standardization processing is imported into MATLAB software, and random sequences of 1-215 are generated by using randderm () function to randomly disorder the data set. Then, the first 80% of the data set is divided into training sets and the last 20% into test sets.
A2, optimizing the hyper-parameters in the XGboost integration algorithm by using 5-turn cross validation (5-CV) and an African bald eagle optimization algorithm on a training set; wherein the hyper-parameters comprise: learning rate (learning _ rate), minimum leaf node sample weight sum (min _ child _ weight), maximum depth of tree (max _ depth), minimum loss reduction amount required for node splitting (gamma), and sub-sampling rate (subsample); in the optimizing process, the average value of the prediction accuracy calculated by 5-fold cross validation is used as the judgment standard for determining the hyper-parameter.
In this embodiment, the XGBoost integration algorithm is implemented on the MATLAB platform by two functions, XGBoost _ train () and XGBoost _ test () compiled by Jeffrey van Prehn.
In this embodiment, 5-fold cross validation is adopted to avoid overfitting of the model in the process of determining the hyper-parameters, the data set is divided into 5 parts, one part of the data set is used as a test set, the other 4 parts of the data set are used as a training set, the training set and the hyper-parameters are used for training the model, the test set is used for testing the model, and the prediction accuracy is obtained. After 5 repetitions, the prediction accuracy average was calculated. The prediction accuracy average value is used as a basis for measuring the quality of the hyperparameter, namely an adaptability value, and an optimized hyperparameter is determined by combining an African bald eagle optimization algorithm, as shown in FIG. 5, j is an individual number, and the specific process is as follows:
(1) initializing a population and related parameters, wherein the parameters comprise: population size Pop, maximum number of iterations, and probability parameter (e.g., L) 1 And L 2 );
(2) Calculating the fitness of the population by using 5-fold cross validation, and determining bald hawks ranked first and second in the population; wherein, the fitness formula is as follows:
Figure BDA0003699877430000091
wherein i represents the ith cross-validation; acc (i) is the accuracy of the ith cross-validation;
in this embodiment, the calculating step of the fitness includes: and dividing the training set into 5 parts again, and calculating by adopting a cross-validation strategy. And 4 parts of the test set are taken as a training set and combined with a hyper-parametric training model, and the test set (the rest 1 part) is used for testing the model to obtain the accuracy. After 5 repetitions, the average accuracy was calculated. The average accuracy value is used as a basis for measuring the quality of the super-parameter, namely an adaptability value.
In this embodiment, after the fitness of each individual (feasible solution) in the population is calculated, the individuals are ranked to obtain the first and second ranked individuals (referred to as bald eagles), i.e., the first and second best bald eagles.
(3) Selecting the direction of movement of the individual (first best bald eagle or second best bald eagle);
Figure BDA0003699877430000092
in the formula (4), R (i) is the best bald eagle selected for the current population; p is a radical of i The selection probability of the individuals of the current population is determined by a roulette mode; l is a radical of an alcohol 1 And L 2 Respectively, the probability parameter (L) for selecting the first best and the second best bald hawk 1 =0.8,L 2 0.2), first and second result are the current first and second best bald eagles; i is the current number of iterations.
(4) Calculating the hunger degree of the current population individuals;
Figure BDA0003699877430000093
in the formula (5), F is the hunger degree of bald hawk; n is the total number of iterations, z is at [ -1,1 [ ]]The random number of (2). h is in [ -2,2]The random number of (2); rand 1 Is a random number at (0, 1); w is a parameter for determining the interruption of the exploration and exploitation stages, and 2.5 is taken; t is an intermediate expression.
(5) Judging the stage (exploration or development) of the individual according to the hunger degree F, and respectively adopting different position updating strategies;
searching stage (F | > 1)
Probability (P) of selecting mechanism in setting exploration phase 1 0.6), and the generated random number rand P1 The location update mode is selected by comparison.
Figure BDA0003699877430000094
In the formula, P (i +1) is the updated position of the population individual; ub and lb optimize the upper and lower boundaries of the space, respectively (ub ═ 1,1,15, 1)],lb=[0,0,0,0,0.4]) The 5 values 1,1,15,1,1 respectively represent the upper limit value of the optimization of the hyper-parameter, and the 5 values 0,0,0,0,0.4 respectively represent the hyper-parameterThe searching optimization lower limit value of (2), the over-parameter is searched within a specified value range; p 1 Probability of selecting a mechanism for an exploration phase; x is a random motion vector of an individual; rand 2 And rand 3 Are both random numbers of (0, 1).
② development stage (0.5 ≦ F | <1)
Probability (P) of selecting a mechanism at a stage of setting development 2 0.4), and the generated random number rand P2 The location update mode is selected by comparison.
Figure BDA0003699877430000101
Figure BDA0003699877430000102
Wherein S is 1 And S 2 Both represent mathematical models simulating individual rotary flight; rand 5 And rand 6 Random numbers that are both (0, 1);
development two stage (0.5 ≤ | F | <1)
Probability (P) of two-stage selection mechanism in setting development 3 0.6), and the generated random number rand P3 Comparing to select a location update mode.
Figure BDA0003699877430000103
Figure BDA0003699877430000104
In which levy (d) mode is used to improve the effectiveness of the algorithm; BestVulture 1 (i) Bald hawk ranked first at present; BestVulture 2 (i) Bald eagle ranked second at present; a. the 1 And A 2 Are equations of motion used to model individual competition.
(6) And (5) repeating the steps (2) to (5) until the maximum iteration number of the algorithm is met.
In the above-described hyper-parametric optimization procedure, the population size is 20, and the maximum number of iterations is 100. The iterative process of the algorithm and the optimization result are shown in fig. 6. After the super parameter optimizing process is finished, the determined super parameters (learning _ rate is 0.15, min _ child _ weight is 0.64, max _ depth is 4, gamma is 0.12 and subsample is 0.6) are substituted into the AVOV-XGboost rockburst intensity prediction model again to update the super parameters, and the optimized AVOV-XGboost rockburst intensity prediction model is obtained by retraining with the training set. Then, the training set is input into the model again, so that the prediction effect of the model on the training set is verified. As can be seen from fig. 7, the accuracy of the model on the training set reached 94.2%. Finally, the remaining 20% of the test set is input into the model, resulting in the predicted effect of the model on the test set, as shown in fig. 8. As can be seen from fig. 8, the accuracy on the test set reached 93%.
According to the prediction results of the model in the training set and the test set, the AVOV-XGboost rockburst intensity prediction model obtained by training has better prediction performance on the training set and the test set, and has excellent generalization capability.
In this embodiment, after the AVOV-XGBoost rockburst intensity prediction model is established, an intelligent rockburst intensity prediction platform is designed through an App Designer module in MATLAB software, and the platform includes: a login interface and a forecast interface, as shown in fig. 9 and 10; wherein, the interface of logging in includes: a user name and password; the prediction interface comprises: 5 evaluation index input boxes and 4 rockburst intensity early warning lamps.
In the embodiment, in the login interface, the user enters the prediction interface by inputting a user name and a password; and inputting numerical values of 5 evaluation indexes in a prediction interface, and turning a warning lamp corresponding to the rockburst intensity predicted by the platform into red after clicking an operation button.
And S105, inputting the rock burst sample to be predicted into the established AVOV-XGboost rock burst intensity prediction model to obtain a rock burst intensity grade prediction value.
In the embodiment, the evaluation indexes of the rockburst sample to be predicted obtained in engineering are input into the established AVOV-XGboost rockburst intensity prediction model to obtain the grade prediction value of the rockburst intensity.
In order to further verify the effectiveness of the model, the AVOV-XGboost rockburst intensity prediction model is applied to the mulberry ridge tunnel engineering, the prediction result is good, and only one case of prediction error exists. The complete prediction results are shown in table 4.
TABLE 4 Sanzhu Ridge Tunnel engineering rockburst prediction results
Figure BDA0003699877430000111
The rock burst intensity prediction method provided by the embodiment of the invention at least has the following beneficial effects:
1. the cross verification strategy and the African bald eagle optimization algorithm are adopted to optimize the super-parameters of the XGboost integration algorithm, and the optimized XGboost integration algorithm is used for carrying out rock burst intensity grade prediction, so that the accuracy of rock burst intensity grade prediction results can be improved, and the problem of poor rock burst disaster prediction effect in the current underground excavation engineering is solved; the optimized model of the African bald eagle optimization algorithm has the advantages of rapidness and high efficiency;
2. a few classes in the established rockburst intensity prediction sample data set are expanded by adopting a synthetic few-class oversampling technology so as to solve the problem of imbalance in the data set; then, carrying out standardization processing on the data in the expanded data set, eliminating dimension difference and reducing calculation expense;
3. the method does not need deep professional knowledge and rich engineering experience possessed by field workers, does not need complicated mechanical calculation or numerical simulation analysis, can obtain the rock burst intensity grade prediction value by inputting the evaluation index corresponding to the rock burst sample to be predicted in the engineering into the AVOV-XGboost rock burst intensity prediction model provided by the embodiment, has the advantages of strong practicability and high efficiency, and provides better guiding significance for rock burst prediction of underground excavation engineering.
4. The AVOV-XGboost rockburst intensity prediction model provided by the embodiment utilizes MATLAB software to develop a rockburst intensity prediction platform, and can be conveniently used by field workers.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (7)

1. A rock burst intensity prediction method is characterized by comprising the following steps:
determining the evaluation index of rock burst intensity prediction and rock burst intensity classification;
establishing a rockburst intensity prediction sample data set according to the determined evaluation index of rockburst intensity prediction and the rockburst intensity classification;
expanding a few classes in the dataset;
based on the expanded data set, optimizing the hyper-parameters in the XGboost integration algorithm by using a cross validation and African bald eagle optimization algorithm, and establishing an AVOV-XGboost rockburst intensity prediction model;
and inputting the rock burst sample to be predicted into the established AVOV-XGboost rock burst intensity prediction model to obtain the grade prediction value of the rock burst intensity.
2. The method for predicting the intensity of a rockburst according to claim 1, wherein the evaluation index of the rockburst intensity prediction includes: uniaxial compressive strength sigma c Uniaxial tensile strength σ t Stress coefficient sigma θc Brittle coefficient σ ct And elastic energy index W et Wherein σ is θ The maximum tangential stress is indicated.
3. The method of predicting rockburst intensity according to claim 1, wherein said grading of rockburst intensity includes: no rock burst i, slight rock burst ii, medium rock burst iii and strong rock burst iv.
4. The method of rock burst intensity prediction according to claim 1, wherein the expanding a few classes in a dataset comprises:
and expanding the minority classes in the data set by adopting a synthesis minority class oversampling technology.
5. The method of rock burst severity prediction according to claim 1, wherein after extending the minority class in the data set, the method further comprises:
and carrying out standardization processing on the data in the expanded data set.
6. The method for predicting the rockburst intensity according to claim 1, wherein the establishing an AVOV-XGboost rockburst intensity prediction model based on the extended data set by optimizing the hyperparameters in the XGboost integration algorithm by using cross validation and the African bald eagle optimization algorithm comprises:
dividing the expanded data set into a training set and a test set;
optimizing the hyper-parameters in the XGboost integration algorithm by using a 5-fold cross validation and African bald eagle optimization algorithm on a training set; wherein the hyper-parameters comprise: learning rate, minimum leaf node sample weight sum, maximum depth of tree, minimum loss reduction amount required by node splitting and sub-sampling rate; in the optimizing process, the average value of the prediction accuracy rate obtained by 5-fold cross validation calculation is used as the judgment standard for determining the hyper-parameter.
7. The method for predicting the rockburst intensity according to claim 1, wherein after the AVOV-XGboost rockburst intensity prediction model is established, the method includes:
developing an intelligent rockburst intensity prediction platform by using MATLAB software; wherein the predictive interface of the platform comprises: 5 evaluation index input boxes and 4 rockburst intensity early warning lamps.
CN202210686507.1A 2022-06-17 2022-06-17 Rock burst intensity prediction method Pending CN115130375A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210686507.1A CN115130375A (en) 2022-06-17 2022-06-17 Rock burst intensity prediction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210686507.1A CN115130375A (en) 2022-06-17 2022-06-17 Rock burst intensity prediction method

Publications (1)

Publication Number Publication Date
CN115130375A true CN115130375A (en) 2022-09-30

Family

ID=83378192

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210686507.1A Pending CN115130375A (en) 2022-06-17 2022-06-17 Rock burst intensity prediction method

Country Status (1)

Country Link
CN (1) CN115130375A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116050663A (en) * 2023-03-09 2023-05-02 北京建筑大学 Rock burst intensity level prediction method based on GD-DNN model
CN116307299A (en) * 2023-05-23 2023-06-23 国网天津市电力公司营销服务中心 Photovoltaic power generation power short-term prediction method, system, equipment and storage medium
CN117093919A (en) * 2023-10-19 2023-11-21 深圳市岩土综合勘察设计有限公司 Geotechnical engineering geological disaster prediction method and system based on deep learning
CN117332240A (en) * 2023-12-01 2024-01-02 中铁四局集团有限公司 Rock burst prediction model construction method, storage medium, rock burst prediction method and system

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116050663A (en) * 2023-03-09 2023-05-02 北京建筑大学 Rock burst intensity level prediction method based on GD-DNN model
CN116307299A (en) * 2023-05-23 2023-06-23 国网天津市电力公司营销服务中心 Photovoltaic power generation power short-term prediction method, system, equipment and storage medium
CN117093919A (en) * 2023-10-19 2023-11-21 深圳市岩土综合勘察设计有限公司 Geotechnical engineering geological disaster prediction method and system based on deep learning
CN117093919B (en) * 2023-10-19 2024-02-02 深圳市岩土综合勘察设计有限公司 Geotechnical engineering geological disaster prediction method and system based on deep learning
CN117332240A (en) * 2023-12-01 2024-01-02 中铁四局集团有限公司 Rock burst prediction model construction method, storage medium, rock burst prediction method and system
CN117332240B (en) * 2023-12-01 2024-04-16 中铁四局集团有限公司 Rock burst prediction model construction method, storage medium, rock burst prediction method and system

Similar Documents

Publication Publication Date Title
CN104732070B (en) A kind of rock burst grade prediction technique based on information vector machine
CN115130375A (en) Rock burst intensity prediction method
CN112735097A (en) Regional landslide early warning method and system
Feng et al. Modeling non-linear displacement time series of geo-materials using evolutionary support vector machines
CN108280289B (en) Rock burst danger level prediction method based on local weighted C4.5 algorithm
CN112232522B (en) Intelligent recommendation and dynamic optimization method for deep roadway support scheme
CN109934398A (en) A kind of drill bursting construction tunnel gas danger classes prediction technique and device
CN112949202A (en) Bayesian network-based rockburst probability prediction method
CN112506990A (en) Hydrological data anomaly detection method based on spatiotemporal information
CN112147432A (en) BiLSTM module based on attention mechanism, transformer state diagnosis method and system
CN112966871A (en) Traffic jam prediction method and system based on convolution long-short term memory neural network
CN110889440A (en) Rockburst grade prediction method and system based on principal component analysis and BP neural network
Yin et al. A novel tree-based algorithm for real-time prediction of rockburst risk using field microseismic monitoring
CN115980826A (en) Rock burst intensity prediction method based on weighted meta-heuristic combined model
CN116992522A (en) Deep foundation pit support structure deformation prediction method, device, equipment and storage medium
CN114662405A (en) Rock burst prediction method based on few-sample measurement and ensemble learning
CN113408190A (en) Bayes-LSTM model-based surrounding rock deformation prediction method during construction of highway tunnel
CN118191967B (en) Intelligent early warning system and method for full-space three-dimensional monitoring of water damage risk of coal seam roof
Li et al. Rockburst Prediction Based on the KPCA‐APSO‐SVM Model and Its Engineering Application
Zhang et al. A semi-Naïve Bayesian rock burst intensity prediction model based on average one-dependent estimator and incremental learning
CN115018161A (en) Intelligent rock burst prediction method based on African bald eagle optimization random forest model
CN116933920A (en) Prediction and early warning method and system for underground mine debris flow
Qiu et al. Rockburst prediction based on distance discrimination method and optimization technology-based weight calculation method
CN113283173B (en) Comprehensive inverse analysis system and method for underground engineering energy and parameters
CN115618610A (en) Underground engineering rockburst intensity evaluation method based on information variable weight

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination