Rockburst intensity grade evaluation method for optimizing BP neural network based on sparrow search algorithm

By optimizing the BP neural network based on the sparrow search algorithm, the problems of slow convergence speed and local optimality in rock burst intensity prediction are solved, and efficient and objective evaluation of rock burst intensity levels are achieved, and prediction accuracy and rationality are improved. .

CN119025982APending Publication Date: 2024-11-26SOUTHWEAT UNIV OF SCI & TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202310605317.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-26
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Traditional BP neural networks converge slowly in rock burst intensity prediction and are prone to local optimality, affecting the effectiveness and accuracy of prediction, and are difficult to get rid of subjective influence.

Method used

The BP neural network is optimized based on the sparrow search algorithm. By determining the model structure and parameters, selecting appropriate feature vectors, normalizing and training the data, and using the sparrow search algorithm to accelerate the iteration process, jump out of the local minimum point, and improve prediction accuracy.

Benefits of technology

It realizes efficient and objective evaluation of rock burst intensity levels, improves prediction accuracy and effectiveness, avoids the influence of weight problems and subjective factors, and enhances the rationality of the evaluation index system and prediction rationality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025982A_ABST
    Figure CN119025982A_ABST
Patent Text Reader

Abstract

The invention discloses a rockburst intensity grade evaluation method for optimizing a BP (Back Propagation) neural network model based on a sparrow Search Algorithm (SSA), and belongs to the technical field of underground engineering and geotechnical engineering. The method comprises the following steps: selecting rockburst evaluation input vectors, and constructing a rockburst intensity grade evaluation index system; rock burst project instance data are collected, the initial weight and threshold of the BP neural network are optimized by using a sparrow search algorithm, and the rock burst intensity level is predicted through input data by using the powerful nonlinear mapping capability of the neural network. According to the evaluation method, a complex nonlinear relation can be fitted to the maximum extent, the influence of subjective factors on a data result is reduced to a certain extent, and the evaluation method has a certain practical application value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of underground engineering, and in particular to a rockburst intensity grade evaluation method based on a sparrow search algorithm to optimize a BP neural network. Background Art

[0002] Rockburst is a dynamic instability geological disaster in which the elastic strain energy stored in the rock mass is suddenly released and a series of destructive phenomena occur due to external disturbances that cause the surrounding rock stress to be redistributed under high ground stress conditions. Rockburst not only causes economic losses, but also directly threatens the safety of personnel and facilities. In severe cases, it can also induce earthquakes and cause a series of secondary disasters. Accurate and reliable prediction of rockburst has always been a hot issue in production safety. Many experts and scholars at home and abroad have conducted a lot of exploration and research on the mechanism of rockburst from different directions. At first, the research was mainly conducted from the aspects of rock strength, energy, brittleness and critical burial depth. However, since rockburst is a complex nonlinear problem, many methods consider the problem of indicator weights and are difficult to get rid of subjective influences. At the same time, due to the rapid development of computer technology and big data, research methods related to machine learning have emerged. Many relevant scholars have established different neural network models. However, traditional BP neural networks have problems such as slow convergence speed and easy to fall into local optimality, which affects the effectiveness and accuracy of prediction. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention proposes a rockburst intensity evaluation method based on a sparrow search algorithm to optimize a BP neural network, which can effectively and objectively evaluate the rockburst intensity in underground engineering.

[0004] A rockburst intensity evaluation method based on sparrow search algorithm optimization BP neural network includes the following steps.

[0005] Step 1: Determine the model structure and parameters and establish the SSA-BP neural network model.

[0006] Step 2: Select the feature vector of the model input and encode the output rockburst grade.

[0007] Step 3: Collect rockburst engineering case data.

[0008] Step 4: Normalize the sample matrix and divide the data into training set and test set.

[0009] Step 5: Input the training set data, conduct training, and obtain the training accuracy; establish the SSA-BP model.

[0010] Step 6: Input the test set data to test the model prediction effect.

[0011] Specifically, the model structure and parameters in step 1 include the number of neurons in the input layer, the number of hidden layers, the number of neurons in the hidden layer, the number of neurons in the output layer, the maximum number of training times, the learning rate, the target error, etc.; the parameters of the sparrow search algorithm: population size, number of evolutions, etc., to establish the SSA-BP model.

[0012] There are 6 evaluation feature vectors selected in step 2: maximum tangential stress σ θ , uniaxial compressive strength of rock σ c , the ratio of maximum shear stress to rock uniaxial compressive strength σ θ / σ c , the ratio of rock uniaxial compressive strength to rock uniaxial tensile strengthσ c / σ t , rock energy index Wet, burial depth H and other 6 indicators.

[0013] In step 2, the rockburst levels are divided into level I (no rockburst), level II (mild rockburst), level III (moderate rockburst) and level IV (severe rockburst), and the codes are "1", "2", "3" and "4" to represent level I, level II, level III and level IV respectively.

[0014] The specific process of normalization in step 4 is as follows: , In the formula, y is the standardized value, min is the minimum value after normalization, y max is the maximum value after normalization, X max is the maximum value in the original data, X min为 is the minimum value in the original data, X is the original data, 90% of the data is used as the training set, and 10% of the data is used as the test set.

[0015] The specific process of step 5 is as follows: the finder position is updated as follows: , in, For the i-th sparrow in j Location information in the dimension, t is the current iteration number, iter max is the maximum number of iterations; a is a random number within (0,1]; R 2 is the alarm value, ST is the safety value (0.8 in this paper); Q is a random number belonging to the normal distribution; L is a 1×N matrix where each element is 1. The position of the follower is updated as follows: , X worst is the worst position during the iteration; X pis the best position of the current follower; n is the population size, A is a 1×n matrix, its internal elements are randomly 1 or -1, and A satisfies A + =A T (AA T ) -1 When i>n / 2, it means that the i-th follower with the lowest fitness is in a very hungry state, so it needs to change its position to get more food. The number of scouts generally accounts for 10% to 20% of the population, and its position is updated as follows: , Where: β is a random number that conforms to the standard normal distribution and is used to control the step length; K is the step length control parameter that represents the moving direction of the sparrow, which is a uniform random number in [-1,1]; Where: β is a random number that conforms to the standard normal distribution and is used to control the step length; K is the step length control parameter that represents the moving direction of the sparrow, which is a uniform random number in [-1,1]; X best is the best position in the iteration process; f i is the fitness value of the i-th individual; f g is the current global optimal fitness value; f w is the worst fitness value; ℇ is a constant, which is used to prevent the denominator from being 0. i >f g When , it means that the sparrow is at the edge of the population and is easily threatened by natural enemies. At this time, it needs to move closer to the center; when f i =f g When this happens, it means that the sparrow in the middle of the population is threatened, and it will randomly move closer to the sparrows in other directions.

[0016] Beneficial effects of the present invention:

[0017] 1. The present invention proposes a rockburst intensity evaluation method based on the sparrow search algorithm to optimize the BP neural network, which has strong effectiveness and high prediction accuracy. Taking into account the characteristics of rockburst and internal and external influencing factors, the maximum tangential stress σ is selected. θ , uniaxial compressive strength of rock σ c , the ratio of maximum shear stress to rock uniaxial compressive strength σ θ / σ c , the ratio of rock uniaxial compressive strength to rock uniaxial tensile strengthσ c / σ t , rock energy index Wet, burial depth H and other six indicators are used as input vectors of the rockburst grade model, which strengthens the rationality of the evaluation index system.

[0018] 2. The BP neural network technology used in the present invention can avoid weight problems and the influence of human subjective factors. The prediction is completely based on data, and the nonlinear relationship between rock burst and input vector is mined from the data to enhance the rationality of rock burst prediction.

[0019] 3. The sparrow search algorithm used in the present invention is a machine learning technology that can enhance local search capabilities, speed up iteration speed and prediction accuracy, jump out of local minimum points to a certain extent, and the evaluation results are accurate, which has strong application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The figure is a schematic diagram of the evaluation method of the present invention.

[0021] Figure 2 This is a schematic diagram of the BP neural network structure of the present invention. DETAILED DESCRIPTION

[0022] In order to clearly describe the calculation principle, working process and method advantages of the embodiment of the present invention, the technical scheme of the invention is described in detail below with reference to the accompanying drawings, and the rockburst evaluation steps of the present invention are described in detail and completely.

[0023] Step 1: Determine the model structure and parameters and establish the SSA-BP neural network model.

[0024] Step 2: Select the feature vector of the model input and encode the output rockburst grade.

[0025] Step 3: Collect rockburst engineering case data.

[0026] Step 4: Normalize the sample matrix and divide the data into training set and test set.

[0027] Step 5: Input the training set data, perform training, and obtain the training accuracy and error.

[0028] Step 6: Input the test set data to test the model prediction effect.

[0029] like Figure 1 As shown, the rockburst grade evaluation process of the present invention can be divided into the following steps:

[0030] Step 1: Determine the model structure and parameters, and establish the SSA-BP neural network model; including the number of input layer neurons, the number of hidden layers, the number of hidden layer neurons, the number of output layer neurons. Maximum number of training times, learning rate, target error, etc.; Sparrow search algorithm parameters: population size, number of evolutions, etc., to establish the network model of the Sparrow Search Algorithm-BP neural network.

[0031] Step 2: Select the maximum tangential stress σ θ, uniaxial compressive strength of rock σ c , the ratio of maximum shear stress to rock uniaxial compressive strength σ θ / σ c , the ratio of rock uniaxial compressive strength to rock uniaxial tensile strengthσ c / σ t , rock energy index Wet, burial depth H and other six indicators are used as the model input feature vectors. At the same time, according to the relevant research and classification standards of rock burst level, the rock burst level can be divided into four levels: Level I (no rock burst), Level II (mild rock burst), Level III (moderate rock burst) and Level IV (strong rock burst), which correspond to the output neurons "1", "2", "3" and "4" respectively.

[0032] Step 3: Normalize the data using the following formula: , In the formula, y is the standardized value, min is the minimum value after normalization, y max is the maximum value after normalization, X max is the maximum value in the original data, X min为 is the minimum value in the original data, X is the original data, 90% of the data is used as the training set, and 10% of the data is used as the test set.

[0033] a. Load the training set data into the SSA-BP neural network model (such as Figure 2 ) is used for model training. During the training process, the fitness of each sparrow is calculated by the sparrow search algorithm, and the positions of the explorer, follower and scout are updated until the optimal position and the optimal fitness value are obtained, which are assigned to the BP neural network model as the weight and threshold respectively, so as to obtain the optimal BP neural network model; repeat step a for model training until the preset model training stop condition is reached, and the BP neural network model obtained at this time is used as the trained sparrow search algorithm optimized BP neural network model; the output rock burst prediction value is statistically analyzed to obtain the final prediction accuracy and error to evaluate the prediction performance of this model.

[0034] Step 5: Substitute the remaining 10% of the data into the model established in step 5 and output the prediction accuracy and error to determine whether the model is effective.

Claims

1. A rockburst intensity evaluation method based on the SSA-BP neural network model, characterized by: The following steps are involved: Step 1: Determine the model structure and parameters and establish the SSA-BP neural network model; Step 2: Select the model input feature vector and encode the output rockburst level; Step 3: Collect rockburst engineering case data; Step 4: Normalize the sample matrix and divide the data into training set and test set; Step 5: Input the training set data, perform training, and obtain the training accuracy; Establish the SSA-BP model; Step 6: Input the test set data to test the model prediction effect; The method for evaluating the grade of the SSA-BP neural network model according to claim 1 is characterized in that: in step 1, the BP neural network parameters are set: the number of neurons in the input layer, the number of hidden layers, the number of neurons in the hidden layer, the number of neurons in the output layer, the maximum number of training times, the learning rate, the target error, etc.; the sparrow search algorithm parameters are set: the population size, the number of evolution times, etc., and the sparrow search algorithm is used to optimize the BP neural network model.

2. The SSA-BP neural network model-based grading method according to claim 1, characterized in that: In the step 1, the BP neural network parameters are set: the number of neurons in the input layer, the number of hidden layers, the number of neurons in the hidden layer, the number of neurons in the output layer, the maximum number of training times, the learning rate, the target error, etc.; the sparrow search algorithm parameters are set: the population size, the number of evolution times, etc., and the sparrow search algorithm is used to optimize the BP neural network model.

3. The SSA-BP neural network model-based grading method according to claim 1, characterized in that: In the step 2, the rockburst levels are divided into level I (no rockburst), level II (mild rockburst), level III (moderate rockburst) and level IV (severe rockburst), and the coding is "1", "2", "3" and "4" to represent level I, level II, level III and level IV respectively.

4. The SSA-BP neural network model-based grade evaluation method according to claim 1, characterized in that: When the sample matrix is ​​normalized in step 4, the evaluation index weights may be offset due to differences in the properties and dimensions of the evaluation indexes. To avoid this problem, the sample index matrix needs to be normalized. The normalized expression is as follows: , In the formula, y is the standardized value, min is the minimum value after normalization, y max is the maximum value after normalization, X max is the maximum value in the original data, X min为 is the minimum value in the original data, X is the original data, 90% of the data is used as the training set, and 10% of the data is used as the test set.

5. The SSA-BP neural network model-based grade evaluation method according to claim 1, characterized in that: The finder position is updated as follows: , in, For the i-th sparrow in j Location information in the dimension, t is the current iteration number, iter max is the maximum number of iterations; a is a random number in (0,1]; R2 is the alarm value, ST is the safety value (0.8 in this paper); Q is a random number belonging to the normal distribution; L is a 1×N matrix where each element is 1. The position of the follower is updated as follows: , X worst is the worst position during the iteration; X p is the best position of the current follower; n is the population size, A is a 1×n matrix, its internal elements are randomly 1 or -1, and A satisfies A + =A T (AA T ) -1 When i>n / 2, it means that the i-th follower with the lowest fitness is in a very hungry state, so it needs to change its position to get more food. The number of scouts generally accounts for 10% to 20% of the population, and its position is updated as follows: , Where: β is a random number that conforms to the standard normal distribution and is used to control the step length; K is the step length control parameter that represents the moving direction of the sparrow, which is a uniform random number in [-1,1]; Where: β is a random number that conforms to the standard normal distribution and is used to control the step length; K is the step length control parameter that represents the moving direction of the sparrow, which is a uniform random number in [-1,1]; X best is the best position in the iteration process; f i is the fitness value of the i-th individual; f g is the current global optimal fitness value; f w is the worst fitness value; ℇ is a constant, which is used to prevent the denominator from being 0. i >f g When , it means that the sparrow is at the edge of the population and is easily threatened by natural enemies. At this time, it needs to move closer to the center; when f i =f g When this happens, it means that the sparrow in the middle of the population is threatened, and it will randomly move closer to the sparrows in other directions.

Citation Information

Cited By

  • Rock burst intensity grade prediction model training method and grade prediction method

    CN119272176A