CN107122860B - Rock burst danger level prediction method based on grid search and extreme learning machine - Google Patents
Rock burst danger level prediction method based on grid search and extreme learning machine Download PDFInfo
- Publication number
- CN107122860B CN107122860B CN201710290857.5A CN201710290857A CN107122860B CN 107122860 B CN107122860 B CN 107122860B CN 201710290857 A CN201710290857 A CN 201710290857A CN 107122860 B CN107122860 B CN 107122860B
- Authority
- CN
- China
- Prior art keywords
- rock burst
- learning machine
- extreme learning
- factor data
- influence factor
- 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.)
- Expired - Fee Related
Links
- 239000011435 rock Substances 0.000 title claims abstract description 149
- 238000000034 method Methods 0.000 title claims abstract description 69
- 230000006870 function Effects 0.000 claims abstract description 48
- 230000004913 activation Effects 0.000 claims abstract description 34
- 210000002569 neuron Anatomy 0.000 claims abstract description 30
- 239000003245 coal Substances 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 20
- 238000012544 monitoring process Methods 0.000 claims description 8
- 238000012795 verification Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 3
- 238000011425 standardization method Methods 0.000 claims description 3
- 238000009933 burial Methods 0.000 claims description 2
- 238000002790 cross-validation Methods 0.000 abstract description 5
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000010304 firing Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Biomedical Technology (AREA)
- Mining & Mineral Resources (AREA)
- Animal Husbandry (AREA)
- Agronomy & Crop Science (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Marine Sciences & Fisheries (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a rock burst danger level prediction method based on grid search and an extreme learning machine, which comprises the following steps: the method comprises the steps of taking influence factor data of known rock burst as input of an extreme learning machine, taking rock burst danger level as output of the extreme learning machine, optimizing combination of the number of neurons in a hidden layer and the type of an activation function of the extreme learning machine by adopting a grid search method, establishing a corresponding extreme learning machine according to each grid node, determining prediction accuracy of the corresponding grid node by adopting a cross-folding cross-validation method for each model, selecting the node with the highest prediction accuracy to determine the number of neurons in the hidden layer and the type of the activation function of the extreme learning machine, and establishing a rock burst danger level prediction model; and inputting the influence factor data of the rock burst to be predicted into the rock burst danger level prediction model to obtain the rock burst danger level prediction value. The method is simple and easy to implement, and simultaneously ensures that the model has good generalization performance.
Description
Technical Field
The invention belongs to the technical field of rock burst risk level prediction, and particularly relates to a rock burst risk level prediction method based on grid search and an extreme learning machine.
Background
Rock burst is a common dynamic phenomenon, the occurrence frequency of rock burst is in an increasing trend along with the increase of the coal mining depth in China, the damage degree of the rock burst is more serious, a great amount of casualties and property loss are caused, the safety production of coal mines is seriously threatened, and therefore the rock burst danger level needs to be effectively predicted.
At present, methods for predicting rock burst include methods such as a drilling cutting method and water content determination which adopt a single influence factor, however, the methods only consider the single influence factor and have the problem of low prediction accuracy, recently, with the development of artificial intelligence, many scholars adopt new technologies and new methods to predict the rock burst risk level, wherein the methods include an artificial neural network method, a GA-ELM method, a Fisher discriminant analysis method, an SVM model and the like, and the methods obtain a large amount of research results, but because the cause of the rock burst is complex, and rock burst data has characteristics such as nonlinearity, correlation and the like, the new methods need to be continuously explored to predict the rock burst risk level.
The extreme learning machine is a novel learning algorithm with a single hidden layer structure, and compared with a traditional single hidden layer feedforward neural network, the extreme learning machine has the advantages of high learning speed, good generalization capability, less adjusting parameters and the like, the current application mainly adopts an optimization algorithm to optimize the input weight and hidden layer deviation of the extreme learning machine, for example, the shixijie and the like adopt a genetic algorithm to optimize the input weight and hidden layer deviation of the extreme learning machine to predict the rock burst, however, the performance of the extreme learning machine is greatly influenced by the number of hidden layer neurons and an activation function, and more parameters need to be optimized when the number of hidden layer neurons is more; in addition, Dinghua and the like adopt a genetic algorithm to preferably select the optimal number of hidden layer neurons, and compare and select a determined excitation function in a progressive mode to predict the power of the coal cutter, however, the type of the activation function is fixed when the number of the hidden layer neurons is optimized, and the weights of an input layer and a hidden layer and a threshold of the hidden layer are generated randomly, so that the uniqueness of an operation result is difficult to ensure; in addition, the overfitting problem of the model is not fully considered in the process of training the parameters of the extreme learning machine, so that the prediction performance of the model cannot be ensured.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a rock burst danger level prediction method based on grid search and an extreme learning machine.
A rock burst danger level prediction method based on grid search and an extreme learning machine comprises the following steps:
step 1: acquiring known rock burst monitoring data at different positions in a coal mine and influence factor data Z of the known rock burst [ Z ═ Z [ [ Z ]1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of rock burst to be predicted '1，z′2，......，z′p]TClassifying the known rock burst monitoring data according to the rock burst magnitude intensity classification standard to obtain rock burst danger levels corresponding to the rock burst influence factor data, wherein z isiIs the influence factor data of the known rock burst of the i th class, z'iThe method comprises the steps of obtaining impact factor data of the ith type of rock burst to be predicted, wherein i is 1, 2, p and p are the number of impact factor data of the rock burst;
the rock burst influencing factors comprise: coal seam thickness, coal seam inclination angle, burial depth, gas concentration and state parameters influencing rock burst;
the state parameters influencing rock burst comprise geological structure conditions, coal seam inclination angle changes, coal seam thickness changes, roof management, pressure relief states and coal blast sound.
Step 2: influence factor data Z ═ Z of known rock burst by adopting a zscore standardization method1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of the impact ground pressure to be predicted '1，z′2，......，z′p]TCarrying out standardization processing to obtain standardized influence factor data X ═ X of the known rock burst1，x2，......，xp]TAnd normalized data X ' ═ X ' of influence factors of the impact ground pressure to be predicted '1，x′2，……，x′p]T；
And step 3: the normalized data X of the influence factors of the known rock burst is determined as X1，x2，......，xp]TAnd the corresponding rock burst danger level is used as a training sample set;
and 4, step 4: the method comprises the steps of taking impact factor data of known rock burst after standardization in a training sample set as input of an extreme learning machine, taking corresponding rock burst danger levels in the training sample set as output of the extreme learning machine, optimizing combination of the number of hidden layer neurons and types of activation functions of the extreme learning machine by adopting a grid search method, establishing a corresponding extreme learning machine according to each grid node, determining prediction accuracy of the corresponding grid node by adopting a cross-folding cross verification method for each model, selecting the node with the highest prediction accuracy to determine the number of hidden layer neurons and types of activation functions of the extreme learning machine, and establishing a rock burst danger level prediction model;
step 4.1: setting intervals of a grid search method, setting a hidden layer neuron number interval according to the number of rock burst influencing factors, assigning the type of an activation function, setting the line number of the grid search method as the maximum value of the hidden layer neuron number, setting the column number of the grid search method as the maximum value of the activation function type assignment, and establishing a search grid;
and the assignment of the activation function type is an integer of 1-3, and the activation function type is respectively expressed as a sigmoid function, a sin function and a hardlim function.
Step 4.2: taking the number of lines where the nodes are located as the number of neurons in a hidden layer of the extreme learning machine, taking an activation function type corresponding to the number of columns where the nodes are located as an activation function type of the extreme learning machine, taking influence factor data of known rock burst after standardization in a training sample set as the input of the extreme learning machine, taking danger levels of the rock burst corresponding to the training sample set as the output of the extreme learning machine, establishing the extreme learning machine, and calculating the prediction accuracy of the extreme learning machine established by the current nodes by adopting a cross-folding verification method;
step 4.3: judging whether the maximum number of nodes is searched currently, if so, executing the step 4.4, otherwise, searching the next node and returning to the step 4.2;
step 4.4: selecting a node corresponding to a model with the maximum prediction accuracy in the extreme learning machine established according to all nodes as a search result, and establishing an extreme learning machine model according to the number of neurons in a hidden layer and the type of an activation function corresponding to the node to obtain a rock burst risk level prediction model;
and 5: predicting the danger level of the rock burst, and setting the standardized influence factor data X 'of the rock burst to be predicted to be [ X'1，x′2，......，x′p]TInputting a rock burst danger grade prediction model to obtain impactAnd (4) predicting the earth pressure danger level.
The invention has the beneficial effects that:
the invention provides a rock burst danger grade prediction method based on grid search and an extreme learning machine, and due to the fact that the mechanism of rock burst generation is complex and the number of influence factors is large, the method adopts a zscore method to standardize influence factor data to eliminate the influence of different dimensions on a model; the performance of the extreme learning machine is greatly influenced by the number of neurons in the hidden layer and the type of the activation function, the number of neurons in the hidden layer and the type of the activation function in the extreme learning machine are combined and optimized by adopting a grid search method and combining cross validation.
Drawings
Fig. 1 is a flowchart of a rock burst risk level prediction method based on grid search and an extreme learning machine according to an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings.
A rock burst danger level prediction method based on grid search and extreme learning machine is disclosed, as shown in figure 1, and comprises the following steps:
step 1: acquiring known rock burst monitoring data at different positions in a coal mine and influence factor data Z of the known rock burst [ Z ═ Z [ [ Z ]1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of rock burst to be predicted '1，z′2，......，z′p]TClassifying the known rock burst monitoring data according to the rock burst magnitude intensity classification standard to obtain rock burst danger levels corresponding to the rock burst influence factor data, wherein z isiIs the influence factor data of the known rock burst of the i th class, z'iFor the ith category of impact factor data of the rock burst to be predicted, i is 1, 2, …, and p is the number of impact factor of the rock burst.
In the present embodiment, the rock burst influence factor packComprises the following steps: thickness z of coal seam1Angle of inclination of coal seam z2Buried depth z3Gas concentration z4And state variables that influence rock burst.
The state variables influencing rock burst include the geological state z5Coal bed dip angle change z6Coal seam thickness variation z7And top plate management z8Pressure relief state z9Sound of coal-firing z10。
Assigning the state parameters influencing rock burst: corresponding assignments are set according to different states of the state parameters affecting rock burst, and the state parameters affecting rock burst are assigned as integer values, as shown in table 1:
TABLE 1 evaluation of the state variables influencing rock burst
In this embodiment, the rock burst monitoring data are classified according to the rock burst magnitude classification standard, and the obtained rock burst danger level is 4, which is level 1: micro-impact, grade 2: weak impact, grade 3: medium impact, grade 4: strong impact.
In this embodiment, rock burst monitoring data and corresponding influence factor data at different positions of the inkstone coal mine are obtained, and the risk level and the corresponding influence factor data of the rock burst are shown in table 2, wherein the influence factor data in the first 26 groups of data serve as the influence factor data of the known rock burst, the corresponding rock burst data serve as the risk level of the known rock burst, and the influence factor data in the last 10 groups of data serve as the influence factor data of the rock burst to be predicted.
TABLE 2 influence factor data of rock burst at different positions of the inkstone coal mine and corresponding rock burst danger level
Step 2: influence factor data Z ═ Z of known rock burst by adopting a zscore standardization method1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of the impact ground pressure to be predicted '1，z′2，......，z′p]TCarrying out standardization processing to obtain standardized influence factor data X ═ X of the known rock burst1，x2，......，xp]TAnd normalized data X ' ═ X ' of influence factors of the impact ground pressure to be predicted '1，x′2，……，x′p]T。
In this embodiment, the formula for normalizing the impact factor data of rock burst by the zscore normalization method is shown in formula (1):
wherein x isijFor the jth value, mu, of the normalized type i rock burst influence factor dataiIs the mean value, sigma, of the data of the ith type rock burst influencing factoriIs the standard deviation, z, of the data of the ith type rock burst influencing factoriiJ is 1, 2, …, and N, N is 36, which is the j-th value of the collected ith type of rock burst influencing factor data.
In the present embodiment, 36 groups of influence factor data in table 2 are normalized, of which the first 26 groups of influence factor data are normalized to form X and the second 10 groups of influence factor data are normalized to form X'.
And step 3: the normalized data X of the influence factors of the known rock burst is determined as X1，x2，......，xp]TAnd its corresponding rock burst hazard ratingAs a training sample set.
And 4, step 4: the method comprises the steps of taking standardized influence factor data of known rock burst in a training sample set as input of an extreme learning machine, taking corresponding rock burst danger levels in the training sample set as output of the extreme learning machine, optimizing combination of the number of hidden layer neurons and types of activation functions of the extreme learning machine by adopting a grid search method, establishing a corresponding extreme learning machine according to each grid node, determining prediction accuracy of the corresponding grid node by adopting a cross-folding cross verification method for each model, selecting the node with the highest prediction accuracy to determine the number of hidden layer neurons and types of activation functions of the extreme learning machine, and establishing a rock burst danger level prediction model.
Step 4.1: setting the interval of a grid search method, setting a hidden layer neuron number interval according to the number of rock burst influencing factors, assigning the type of an activation function, setting the line number of the grid search method as the maximum value of the hidden layer neuron number, setting the column number of the grid search method as the maximum value of the activation function type assignment, and establishing a search grid.
In this embodiment, the interval of the grid search method is set to 1, the interval of the number of neurons in the hidden layer is set to [1, 100] according to the number of factors affecting rock burst, the type of the activation function is assigned to be an integer from 1 to 3, and the activation function is respectively expressed as a sigmoid function, a sin function, and a hardlim function, in this embodiment, the value of the sigmoid function is 1, the value of the sin function is 2, and the value of the hardlim function is 3.
In this embodiment, the number of rows in the grid search method is set to 100 rows, and the number of columns in the grid search method is set to 3 columns.
Step 4.2: taking the number of lines where the nodes are located as the number of neurons in a hidden layer of the extreme learning machine, taking the activation function type corresponding to the number of columns where the nodes are located as the activation function type of the extreme learning machine, taking the influence factor data of the known rock burst after standardization in the training sample set as the input of the extreme learning machine, taking the danger level of the rock burst corresponding to the training sample set as the output of the extreme learning machine, establishing the extreme learning machine, and calculating the prediction accuracy of the extreme learning machine established by the current nodes by adopting a cross-folding verification method.
In the embodiment, a ten-fold cross-validation method is adopted, the standardized influence factor data of the known rock burst in the training sample set is divided into ten parts, 9 parts of the ten parts are taken as training data in turn, 1 part of the ten parts is taken as test data and is taken as input of an extreme learning machine, the accuracy of the ten prediction results and the corresponding rock burst danger level in the training sample set is calculated through ten operations, and the prediction accuracy is taken as an evaluation index of the corresponding grid node.
Step 4.3: and judging whether the maximum number of nodes is searched currently, if so, executing the step 4.4, otherwise, searching the next node, and returning to the step 4.2.
In this embodiment, the maximum number of nodes is 97 rows and 1 column.
Step 4.4: and selecting a node corresponding to a model with the maximum prediction accuracy in the extreme learning machine established according to all the nodes as a search result, and establishing an extreme learning machine model according to the number of neurons in the hidden layer and the type of the activation function corresponding to the node to obtain a rock burst risk level prediction model.
In this embodiment, the rock burst risk level prediction model has a three-layer structure, as shown in formula (2):
wherein M is the number of hidden layer neurons, v is 1, 2, …, M, ωvIs the connection weight, beta, of the input layer and the hidden layervAs a connection weight of the hidden layer to the output layer, bvFor the thresholds of hidden layer neurons, g (—) is the activation function of the extreme learning machine obtained by optimization, okFor predicting the danger rating of rock burst, xkAnd inputting the impact ground pressure influence factor data after the k-th standardization in the training sample set, wherein k is 1, 2, …, N.
In this embodiment, the obtained optimal nodes are 97 rows and 1 column, that is, the number of neurons in the hidden layer is 97, the type of the activation function is a sigmoid function, and the weights of part of the input layer and the hidden layer of the extreme learning machine model and the threshold b of the hidden layer are obtained as shown in table 3:
TABLE 3 weight values of partial input layer and hidden layer of extreme learning machine model and hidden layer threshold b
And (3) establishing an extreme learning machine model according to the corresponding parameters of the node, wherein the correct identification rate of the model is 0.84615 after ten-fold cross validation.
In this embodiment, in order to compare with the proposed method, a naive bayes method and an adaboost m1 method are respectively adopted to establish a rock burst risk level prediction model, and through ten-fold cross validation, the correct recognition rates of the models are 0.7692 and 0.6154 respectively, which are both lower than 0.84615, indicating that the rock burst prediction model established according to the method has better performance.
And 5: predicting the danger level of the rock burst, and setting the standardized influence factor data X 'of the rock burst to be predicted to be [ X'1，x′2，......，x′p]TAnd inputting the rock burst danger grade prediction model to obtain a rock burst danger grade prediction value.
The method, the naive Bayes method and the Adaboost M1 method are adopted to predict the risk level of the corresponding rock burst according to the data standardized by the last 10 groups of influence factor data in the table 2, and the prediction results are shown in the table 4:
TABLE 4 prediction results
It can be seen from the table that the prediction model established by the method accurately predicts 9 groups of rock burst danger levels in data, and only the medium rock burst pressure of the 10 th group of data is judged as weak rock burst by mistake, and 8 groups of levels are accurately predicted by a naive bayes method and an Adaboost M1 method, wherein the medium rock burst pressure of the 2 nd group and the 7 th group is judged as strong rock burst by the naive bayes method, and the medium rock burst pressure of the 4 th group and the medium rock burst pressure of the 7 th group are judged as strong rock burst by the Adaboost M1 method.
Claims (3)
1. A rock burst danger level prediction method based on grid search and an extreme learning machine is characterized by comprising the following steps:
step 1: acquiring known rock burst monitoring data at different positions in a coal mine and influence factor data Z of the known rock burst [ Z ═ Z [ [ Z ]1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of rock burst to be predicted '1，z′2，......，z′p]TClassifying the known rock burst monitoring data according to the rock burst magnitude intensity classification standard to obtain rock burst danger levels corresponding to the rock burst influence factor data, wherein z isiIs the influence factor data of the known rock burst of the i th class, z'iThe method comprises the steps of obtaining impact factor data of the ith type of rock burst to be predicted, wherein i is 1, 2, p and p are the number of impact factor data of the rock burst;
step 2: influence factor data Z ═ Z of known rock burst by adopting a zscore standardization method1，z2，......，zp]TAnd influence factor data Z ' ═ Z ' of the impact ground pressure to be predicted '1，z′2，......，z′p]TCarrying out standardization processing to obtain standardized influence factor data X ═ X of the known rock burst1，x2，......，xp]TAnd normalized data X ' ═ X ' of influence factors of the impact ground pressure to be predicted '1，x′2，......，x′p]T；
And step 3: the normalized data X of the influence factors of the known rock burst is determined as X1，x2，......，xp]TAnd their corresponding rock burst hazardsTaking the grade as a training sample set;
and 4, step 4: the method comprises the steps of taking impact factor data of known rock burst after standardization in a training sample set as input of an extreme learning machine, taking corresponding rock burst danger levels in the training sample set as output of the extreme learning machine, optimizing combination of the number of hidden layer neurons and types of activation functions of the extreme learning machine by adopting a grid search method, establishing a corresponding extreme learning machine according to each grid node, determining prediction accuracy of the corresponding grid node by adopting a cross-folding cross verification method for each model, selecting the node with the highest prediction accuracy to determine the number of hidden layer neurons and types of activation functions of the extreme learning machine, and establishing a rock burst danger level prediction model;
step 4.1: setting intervals of a grid search method, setting a hidden layer neuron number interval according to the number of rock burst influencing factors, assigning the type of an activation function, setting the line number of the grid search method as the maximum value of the hidden layer neuron number, setting the column number of the grid search method as the maximum value of the activation function type assignment, and establishing a search grid;
step 4.2: taking the number of lines where the nodes are located as the number of neurons in a hidden layer of the extreme learning machine, taking an activation function type corresponding to the number of columns where the nodes are located as an activation function type of the extreme learning machine, taking influence factor data of known rock burst after standardization in a training sample set as the input of the extreme learning machine, taking danger levels of the rock burst corresponding to the training sample set as the output of the extreme learning machine, establishing the extreme learning machine, and calculating the prediction accuracy of the extreme learning machine established by the current nodes by adopting a cross-folding verification method;
step 4.3: judging whether the maximum number of nodes is searched currently, if so, executing the step 4.4, otherwise, searching the next node and returning to the step 4.2;
step 4.4: selecting a node corresponding to a model with the maximum prediction accuracy in the extreme learning machine established according to all nodes as a search result, and establishing an extreme learning machine model according to the number of neurons in a hidden layer and the type of an activation function corresponding to the node to obtain a rock burst risk level prediction model;
and 5: predicting the danger level of the rock burst, and setting the standardized influence factor data X 'of the rock burst to be predicted to be [ X'1，x′2，......，x′p]TAnd inputting the rock burst danger grade prediction model to obtain a rock burst danger grade prediction value.
2. The grid search and extreme learning machine-based rock burst hazard level prediction method according to claim 1, wherein the rock burst influencing factors comprise: coal seam thickness, coal seam inclination angle, burial depth, gas concentration and state parameters influencing rock burst;
the state parameters influencing rock burst comprise geological structure conditions, coal seam inclination angle changes, coal seam thickness changes, roof management, pressure relief states and coal blast sound.
3. The method for predicting the danger level of rock burst based on grid search and extreme learning machine according to claim 1, wherein the activation function type is assigned with an integer of 1-3 and is respectively represented by a sigmoid function, a sin function and a hardlim function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710290857.5A CN107122860B (en) | 2017-04-28 | 2017-04-28 | Rock burst danger level prediction method based on grid search and extreme learning machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710290857.5A CN107122860B (en) | 2017-04-28 | 2017-04-28 | Rock burst danger level prediction method based on grid search and extreme learning machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107122860A CN107122860A (en) | 2017-09-01 |
| CN107122860B true CN107122860B (en) | 2020-01-07 |
Family
ID=59725143
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710290857.5A Expired - Fee Related CN107122860B (en) | 2017-04-28 | 2017-04-28 | Rock burst danger level prediction method based on grid search and extreme learning machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107122860B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110630330B (en) * | 2019-09-23 | 2021-01-05 | 辽宁工程技术大学 | Rock burst classification and judgment method based on energy release main body |
| CN111325461B (en) * | 2020-02-18 | 2022-03-08 | 山东科技大学 | Real-time evaluation method for coal seam impact risk based on vibration monitoring technology |
| CN111764963B (en) * | 2020-07-06 | 2021-04-02 | 中国矿业大学(北京) | Rock burst prediction method based on fast-RCNN |
| CN113009077B (en) * | 2021-02-18 | 2023-05-02 | 南方电网数字电网研究院有限公司 | Gas detection method, gas detection device, electronic equipment and storage medium |
| CN113298299A (en) * | 2021-05-13 | 2021-08-24 | 华北科技学院(中国煤矿安全技术培训中心) | BP neural network-based coal bed impact risk intelligent evaluation method |
| CN113469342A (en) * | 2021-07-08 | 2021-10-01 | 北京科技大学 | Rock burst early warning method based on deep learning microseismic monitoring data |
| CN113901939B (en) * | 2021-10-21 | 2022-07-01 | 黑龙江科技大学 | Rock burst danger level prediction method based on fuzzy correction, storage medium and equipment |
| CN114877925B (en) * | 2022-03-31 | 2023-08-22 | 上海交通大学 | Comprehensive energy system sensor fault diagnosis method based on extreme learning machine |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103256073A (en) * | 2013-04-28 | 2013-08-21 | 中国矿业大学 | Underground coal mine pressure bump zoning grading predication method |
| CN105785471A (en) * | 2016-02-14 | 2016-07-20 | 辽宁工程技术大学 | Impact danger evaluation method of mine pre-exploiting coal seam |
-
2017
- 2017-04-28 CN CN201710290857.5A patent/CN107122860B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103256073A (en) * | 2013-04-28 | 2013-08-21 | 中国矿业大学 | Underground coal mine pressure bump zoning grading predication method |
| CN105785471A (en) * | 2016-02-14 | 2016-07-20 | 辽宁工程技术大学 | Impact danger evaluation method of mine pre-exploiting coal seam |
Non-Patent Citations (2)
| Title |
|---|
| 《基于GA-ELM的冲击地压危险性预测研究》;朱志洁等;《中国安全生产科学技术》;20140831;第10卷(第8期);正文第46-51页 * |
| 《基于改进GS-SVM的煤矿冲击地压预测研究》;李烨等;《世界科技研究与发展》;20160831;第38卷(第4期);正文第758-762页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107122860A (en) | 2017-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107122860B (en) | Rock burst danger level prediction method based on grid search and extreme learning machine | |
| CN107122861B (en) | Gas emission quantity prediction method based on PCA-PSO-ELM | |
| CN112529341B (en) | Drilling well leakage probability prediction method based on naive Bayesian algorithm | |
| CN108388921B (en) | Overflow loss real-time identification method based on random forest | |
| CN109214460B (en) | Power transformer fault diagnosis method based on relative transformation and nuclear entropy component analysis | |
| CN103617147A (en) | Method for identifying mine water-inrush source | |
| CN108280289B (en) | Rock burst danger level prediction method based on local weighted C4.5 algorithm | |
| CN116341901B (en) | Integrated evaluation method for landslide surface domain-monomer hazard early warning | |
| CN104570109B (en) | A kind of method of reservoir hydrocarbons prediction | |
| CN113806889A (en) | Processing method, device and equipment of TBM cutter head torque real-time prediction model | |
| CN115130375A (en) | Rock burst intensity prediction method | |
| Zhang et al. | Geological Type Recognition by Machine Learning on In‐Situ Data of EPB Tunnel Boring Machines | |
| CN109594967A (en) | A kind of lock of tool drilling detection method for early warning based on well logging big data | |
| KR102155101B1 (en) | System and method for predicting ground layer information, and a recording medium having computer readable program for executing the method | |
| CN117272841B (en) | Shale gas dessert prediction method based on hybrid neural network | |
| Chen et al. | Comparison and application of neural networks in LWD lithology identification | |
| CN116822971B (en) | Well wall risk level prediction method | |
| CN115438823A (en) | Borehole wall instability mechanism analysis and prediction method and system | |
| CN104978484A (en) | Fuzzy forecasting model based method for detecting pulp concentration in ore grinding process of dressing plant | |
| CN112819087A (en) | Effluent BOD sensor abnormity detection method based on modular neural network | |
| CN113946790A (en) | Method, system, equipment and terminal for predicting height of water flowing fractured zone | |
| CN110674868A (en) | Stratum lithology identification system and method based on high-dimensional drilling parameter information | |
| CN111274736A (en) | Water flowing fractured zone prediction method based on supervised learning neural network algorithm | |
| Roustazadeh et al. | Estimating oil and gas recovery factors via machine learning: Database-dependent accuracy and reliability | |
| Feder | Machine-learning approach determines spatial variation in shale decline curves |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200107 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |