Method, device and electronic equipment for generating a rock burst early warning model

Abstract

The present disclosure provides a method and device for generating a rock burst early warning model, and an electronic device, and relates to the technical field of coal mine early warning. The method comprises: obtaining a training data set and a verification data set corresponding to a target mining area; randomly selecting at least one index as a target index from each index corresponding to each parameter type multiple times; constructing a decision tree corresponding to the target index selected each time based on the target index selected each time and the training data set; verifying the early warning accuracy of each decision tree based on the verification data set; if the maximum early warning accuracy is less than a threshold, pruning a first decision tree corresponding to the maximum early warning accuracy, so that the early warning accuracy of the pruned decision tree is greater than the threshold; and determining the pruned decision tree as a rock burst early warning model corresponding to the target mining area. Thus, the index that has a greater impact on the target mining area can be accurately determined, thereby improving the accuracy of the generated rock burst early warning model.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of coal mine early warning, and in particular to a method and device for generating an impact ground pressure early warning model and an electronic device. BACKGROUND

[0002] At present, the online monitoring methods commonly used in mines include microseismic, ground sound, ground stress, and support resistance, etc. However, the judgment of the change trend of the impact danger early warning index relies on artificial experience, and the weight is manually set, which is low in efficiency and reliability. Meanwhile, due to the different dimensions reflected by each early warning index on the impact ground pressure gestation and evolution process, the results may conflict with each other, causing the misjudgment of the actual danger state by the relevant personnel. SUMMARY

[0003] The present disclosure aims to at least partially solve one of the technical problems in the related art.

[0004] A method for generating an impact ground pressure early warning model is provided in the first aspect of the present disclosure, comprising:

[0005] obtaining a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set includes a danger level label and a danger result of each index corresponding to each parameter type under the target mining area;

[0006] randomly selecting at least one index as a target index from each index corresponding to each parameter type for multiple times;

[0007] constructing a decision tree corresponding to each selected target index based on each selected target index and the training data set;

[0008] verifying the early warning accuracy of each decision tree based on the verification data set;

[0009] pruning the first decision tree corresponding to the maximum early warning accuracy to make the early warning accuracy of the pruned decision tree greater than the threshold value in the case that the maximum early warning accuracy is less than the threshold value;

[0010] determining the pruned decision tree as the impact ground pressure early warning model corresponding to the target mining area.

[0011] A device for generating an impact ground pressure early warning model is provided in the second aspect of the present disclosure, comprising:

[0012] a first obtaining module configured to obtain a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set includes a danger level label and a danger result of each index corresponding to each parameter type under the target mining area;

[0013] The second obtaining module is configured to randomly select at least one index as a target index from each parameter type corresponding index multiple times;

[0014] The constructing module is configured to construct a decision tree corresponding to each selected target index based on each selected target index and the training data set;

[0015] The verifying module is configured to verify the early warning accuracy of each decision tree based on the verification data set;

[0016] The pruning module is configured to, in a case where the maximum early warning accuracy is less than a threshold, prune the first decision tree corresponding to the maximum early warning accuracy, so that the early warning accuracy of the pruned decision tree is greater than the threshold;

[0017] The determining module is configured to determine the pruned decision tree as the rock burst early warning model corresponding to the target mining area.

[0018] The third aspect of the present disclosure provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, when the processor executes the program, the generation method of the rock burst early warning model is realized.

[0019] The fourth aspect of the present disclosure provides a computer readable storage medium, which stores a computer program, when the computer program is executed by a processor, the generation method of the rock burst early warning model is realized.

[0020] The fifth aspect of the present disclosure provides a computer program product, comprising a computer program, when the computer program is executed by a processor, the generation method of the rock burst early warning model is realized.

[0021] The generation method, device and electronic device of the rock burst early warning model provided by the present disclosure have the following beneficial effects:

[0022] In the embodiments of the present disclosure, first, the training data set and the verification data set corresponding to the target mining area are acquired, then at least one index is randomly selected as a target index from each index corresponding to each parameter type, and a decision tree corresponding to each selected target index is constructed based on each selected target index and the training data set. Then, the early warning accuracy of each decision tree is verified based on the verification data set. In the case that the maximum early warning accuracy is less than a threshold, the first decision tree corresponding to the maximum early warning accuracy is pruned so that the early warning accuracy of the pruned decision tree is greater than the threshold. Finally, the pruned decision tree is determined as the rock burst early warning model corresponding to the target mining area. Thus, the index having a greater impact on the target mining area can be accurately determined, so that the accuracy of the rock burst early warning model generated based on the index having a greater impact on the target mining area can be improved, and then the target mining area can be accurately provided with a danger warning.

[0023] Additional aspects and advantages of the present disclosure will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0024] The above and / or additional aspects and advantages of the present disclosure will become apparent and be readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A flowchart of a method for generating a rock burst early warning model provided by an embodiment of the present disclosure;

[0026] Figure 2 A schematic diagram of sample data provided by an embodiment of the present disclosure;

[0027] Figure 3 A schematic diagram of a decision tree provided by an embodiment of the present disclosure;

[0028] Figure 4 A flowchart of a method for generating a rock burst early warning model provided by an embodiment of the present disclosure;

[0029] Figure 5 A flowchart of a method for generating a rock burst early warning model provided by an embodiment of the present disclosure

[0030] Figure 6 A structural schematic diagram of a rock burst early warning model generation device provided by another embodiment of the present disclosure;

[0031] Figure 7 A block diagram of an exemplary electronic device suitable for implementing an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0032] Embodiments of the present disclosure are described below in detail, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present disclosure, and cannot be understood as a limitation of the present disclosure.

[0033] A method for generating a rock burst early warning model, an apparatus, an electronic device, and a storage medium are described below with reference to the accompanying drawings for embodiments of the present disclosure.

[0034] Figure 1 A flowchart of the method for generating a rock burst early warning model provided by the embodiments of the present disclosure.

[0035] The embodiments of the present disclosure are exemplified by configuring the method for generating a rock burst early warning model in a rock burst early warning model generation apparatus, which can be applied to any electronic device, so that the electronic device can perform the function of generating a rock burst early warning model.

[0036] The electronic device can be a personal computer (PC), a cloud device, a mobile device, etc. The mobile device can be, for example, a mobile phone, a tablet computer, a personal digital assistant, a wearable device, a vehicle-mounted device, etc. The mobile device has various operating systems, touch screens, and / or display screens.

[0037] As shown in Figure 1 The method for generating a rock burst early warning model can include the following steps:

[0038] Step 101, obtaining a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set includes a danger level label and a danger result of each index corresponding to each parameter type under the target mining area.

[0039] The parameter types under the target mining area can include microseismic, ground sound, stress, support resistance, anchor rod and anchor cable data types.

[0040] Optionally, in the case of a microseismic parameter type, the microseismic corresponding indexes can include b value, A(B), η value, lack of seismic, Z-map value, etc. In the case of a stress parameter type, the stress corresponding indexes can include stress gradient, stress increase rate, etc. In the case of a support resistance parameter type, the support resistance corresponding indexes can include dynamic load coefficient, working cycle time, etc.

[0041] The danger result of each index can be a danger level determined according to the data of each index.

[0042] The danger level label can be an actual danger level corresponding to the sample data.

[0043] The number of sample data in the training data set is greater than the number of sample data in the verification data set.

[0044] Optionally, the training data set and the verification data set corresponding to the target mining area can be obtained by the following steps:

[0045] (1) Obtain historical data corresponding to each parameter type of the target mining area every day.

[0046] For example, historical data of each day in a preset time period (for example, three years, one year, etc.) before the current time can be obtained.

[0047] (2) Process the historical data corresponding to each parameter type every day based on the function corresponding to each index of each parameter type, to obtain a danger result corresponding to each index of each parameter type.

[0048] Table 1 shows each index, the function corresponding to each index, and the meaning of each index.

[0049]

[0050]

[0051]

[0052]

[0053] (3) Obtain a danger level label corresponding to the target mining area every day.

[0054] (4) Generate a plurality of sample data according to the danger result corresponding to each index every day and the danger level label.

[0055] (5) Divide the plurality of sample data into a training data set and a verification data set.

[0056] For example, 80% of the sample data can be used as the training data set, and 20% of the training data can be used as the verification data set. The disclosure does not limit the number of sample data in the verification data set and the training data set.

[0057] In the embodiments of the disclosure, after obtaining the historical data corresponding to each parameter type every day, the historical data corresponding to each day can also be preprocessed.

[0058] Optionally, in the case that historical data corresponding to any parameter type per day is missing, it is determined whether the time parameter is involved in the function of each index corresponding to any parameter type; if the time parameter is involved in the function of any index corresponding to any parameter type, the missing data is filled in by replacement or interpolation; or if the time parameter is not involved in the function of each index corresponding to any parameter type, the data of the day is deleted.

[0059] Optionally, the identification of outliers is usually processed by a univariate scatter plot or a box plot, such as a function of a univariate scatter plot, in which points far from the normal range are regarded as outliers. The outliers are marked (not deleted).

[0060] Figure 2 A schematic diagram of sample data provided by an embodiment of the present disclosure is shown in FIG. 1, in which each row is a sample data, the last element (low, general, larger) in each row of sample data is a danger level label, and “0”, “1”, “2” and “3” in each row of sample data are danger results corresponding to each index. Among them, 0 represents low risk, 1 represents general risk, 2 represents larger risk, and 3 represents major risk. Each column is a danger result corresponding to each index in different samples. Figure 2

[0061] Step 102, randomly selecting at least one index as a target index from each index corresponding to each parameter type multiple times.

[0062] For example, if the indexes corresponding to microseism include b value, A(B), η value, lack of earthquake, and Z-map value; the indexes corresponding to stress can include stress gradient and stress increasing rate; and the indexes corresponding to support resistance include dynamic load coefficient and working cycle time. At least one index is selected from each index corresponding to each parameter type each time. For example, b value, A(B) and η value are selected from the indexes corresponding to microseism, dynamic load coefficient and working cycle time are selected from the indexes corresponding to support resistance, and stress gradient is selected as a target index selected once from the indexes corresponding to stress.

[0063] Step 103, constructing a decision tree corresponding to each target index selected each time based on the target index selected each time and the training data set.

[0064] Figure 3 A schematic diagram of a decision tree provided by an embodiment of the present disclosure, the method of constructing a decision tree in the embodiment of the present disclosure can include the following steps:

[0065] (1) determining the first information gain between each target index and the training data set.

[0066] (2) determining the first index with the maximum first information gain as the root node of the decision tree.​

[0067] As shown in Figure 3 , if the first information gain of the CUFIT model is the largest, the CUFIT model is the first index, and can be used as the root node of the decision tree.

[0068] (3) Based on the first index, the training data set is divided into a plurality of first subsets, wherein the risk results corresponding to the first index in each sample data of each first subset are the same.

[0069] Specifically, as shown in Figure 3 , in the case of the first index being the CUFIT model, the training data set is divided into four first subsets according to the CUFIT model. In the first first subset, the risk result corresponding to the CUFIT model in each sample data is 0, and the risk level label corresponding to each sample data is low. In the second first subset, the risk result corresponding to the CUFIT model in each sample data is 1; in the third first subset, the risk result corresponding to the CUFIT model in each sample data is 2; in the fourth first subset, the risk result corresponding to the CUFIT model in each sample data is 3, and the risk level label corresponding to each sample data is major.

[0070] (4) In the case that the risk level labels corresponding to each sample data in the first subset are the same, the risk level label corresponding to each sample data in the first subset is determined as a leaf node under the root node.

[0071] As shown in Figure 3 , in the case that the risk result corresponding to the CUFIT model in each sample data in the first first subset is 0, and the risk level label corresponding to each sample data is low, "low" is determined as a leaf node under the CUFIT model. In the case that the risk result corresponding to the CUFIT model in each sample data in the fourth first subset is 3, and the risk level label corresponding to each sample data is major, "major" is determined as a leaf node under the CUFIT model.

[0072] (5) In the case that the risk level labels corresponding to each sample data in the first subset are different, the second information gain between other target indexes and the first subset is determined in addition to the first index.

[0073] As shown in Figure 3 , in the case that the risk result corresponding to the CUFIT model in each sample data in the second first subset and the third first subset is 1, and the risk level labels corresponding to each sample data are different, the second information gain between other target indexes and the second first subset, and the second information gain between other target indexes and the third first subset are calculated in addition to the CUFIT model index.

[0074] (6) The second indicator with the largest second information gain is determined as an internal node under the root node.

[0075] like Figure 3 As shown, the second information gain indicator corresponding to the second first subset is the time series concentration Q 时 In the case of 时 Determine it as an internal node under the CUFIT model. In the case that the indicator with the largest second information gain corresponding to the third first subset is absence of earthquake, absence of earthquake is determined as an internal node under the CUFIT model.

[0076] (7) Based on the second indicator, the first subset is divided into multiple second subsets until the subset corresponding to each node cannot be further divided, so as to obtain a decision tree, wherein the risk result corresponding to the second indicator in each sample data of each second subset is the same.

[0077] like Figure 3 As shown, after determining the timing concentration Q 时 and after the absence of earthquake, respectively based on the time series concentration Q 时 Divide the second first subset to obtain the second subset, where the time series concentration Q of each sample data in the second subset is 时 The risk results are the same. Among them, the time series concentration Q of each sample data in the first and second subsets is 时 The risk result is 0, but the risk level label of each sample data is different, so it is necessary to continue dividing. The time series concentration Q of each sample data in the second subset is 时 The risk result is 1, and the risk level label of each sample data is "low", so "low" can be determined as the leaf node corresponding to the risk result 1. The time series concentration Q of each sample data in the third second subset is 时 The risk result is 2, and the risk level label of each sample data is "low", "low" can be determined as the leaf node corresponding to the risk result 2.

[0078] Optionally, if the danger level labels of each sample data in the subset are the same, or each target indicator is completely traversed, it is determined that the subset corresponding to each node cannot be further divided.

[0079] Step 104: Verify the warning accuracy corresponding to each decision tree based on the verification data set.

[0080] Optionally, the risk results of each index of each sample data in the verification data set are input into the decision tree to obtain a predicted risk level of each sample data in the verification data set, and then a first quantity of sample data whose predicted risk level is different from the corresponding risk level label is determined, and a ratio between the first quantity and a total quantity of sample data in the verification data set is determined as a corresponding early warning accuracy of the decision tree.

[0081] In a case where the maximum early warning accuracy is less than the threshold, the first decision tree corresponding to the maximum early warning accuracy is pruned to make the early warning accuracy of the pruned decision tree greater than the threshold.

[0082] Optionally, in a case where the maximum early warning accuracy is less than the threshold, it is identified that there is still a large error when the data is early warned by the decision tree with the highest early warning accuracy, and the first decision tree corresponding to the maximum early warning accuracy needs to be further adjusted to make the early warning accuracy of the adjusted decision tree greater than the threshold.

[0083] Optionally, the first decision tree corresponding to the maximum early warning accuracy can be pruned in the embodiments of the present disclosure to make the early warning accuracy of the pruned decision tree greater than the threshold.

[0084] Step 106, determining the pruned decision tree as the rock burst early warning model corresponding to the target mining area.

[0085] It should be noted that after the rock burst early warning model corresponding to the target mining area is determined, the data corresponding to each parameter type under the target mining area can be obtained in real time, and the data corresponding to each parameter type under the target mining area is predicted based on the rock burst early warning model to obtain the rock burst risk level corresponding to the target mining area in real time.

[0086] In the embodiments of the present disclosure, first, the training data set and the verification data set corresponding to the target mining area are obtained, then at least one index is randomly selected as a target index from each index corresponding to each parameter type, and a decision tree corresponding to each selected target index is constructed based on each selected target index and the training data set, and then the early warning accuracy of each decision tree is verified based on the verification data set, in a case where the maximum early warning accuracy is less than the threshold, the first decision tree corresponding to the maximum early warning accuracy is pruned to make the early warning accuracy of the pruned decision tree greater than the threshold, and finally the pruned decision tree is determined as the rock burst early warning model corresponding to the target mining area. Thus, the index that has a greater impact on the target mining area can be accurately determined, so that the accuracy of the rock burst early warning model generated based on the index that has a greater impact on the target mining area can be improved, and then the target mining area can be accurately provided with a risk warning.

[0087] Figure 4A flowchart of a method for generating a rock burst early warning model according to an embodiment of the present disclosure is shown in FIG. 13. The method for generating the rock burst early warning model can include the following steps: Figure 4

[0088] In step 401, a training data set and a validation data set corresponding to a target mining area are obtained, wherein each sample data in the training data set and the validation data set includes a danger level label and a danger result of each index corresponding to each parameter type in the target mining area.

[0089] In step 402, at least one index is randomly selected as a target index from each index corresponding to each parameter type.

[0090] In step 403, a decision tree corresponding to the target index is constructed based on the target index and the training data set.

[0091] In step 404, the early warning accuracy of each decision tree is verified based on the validation data set.

[0092] In the embodiments of the present disclosure, the specific implementation forms of steps 401 to 404 can refer to the detailed descriptions in other embodiments of the present disclosure, and will not be described in detail here.

[0093] In step 405, a first false judgment frequency mean and a false judgment frequency standard deviation corresponding to a last-end sub-tree in the first decision tree are determined.

[0094] As shown in FIG. 14, the last-end sub-tree is a sub-tree corresponding to the epicenter concentration index. Figure 3

[0095] Optionally, the number of samples N covered by each leaf node in the last-end sub-tree and the number of pre-judgment errors E are obtained, and the error rate of each leaf node is (E+0.5) / N, wherein 0.5 is a penalty factor.

[0096] For a sub-tree, if there are L leaf nodes, the false judgment rate p of the sub-tree is:

[0097]

[0098] wherein i is the i th leaf node, E i is the number of pre-judgment errors corresponding to the i th leaf node, N i is the number of samples covered by the i th leaf node. i i

[0099] The first false judgment frequency mean is the product of the number of samples covered by the sub-tree and the false judgment rate p of the sub-tree.

[0100] The false judgment frequency standard deviation is

[0101] ​​​​Step 406: Determine the target leaf node to be replaced corresponding to the last subtree.

[0102] Optionally, the hazard level with the most identical hazard level labels among the samples corresponding to the subtree may be determined as the target leaf node.

[0103] like Figure 3 As shown, if the number of samples with the hazard level label of "general" in the subtree where the epicenter concentration index is located is greater than the number of samples with the hazard level label of "low", the target leaf node is determined to be "general".

[0104] Step 407: Replace the last subtree in the first decision tree with the target leaf node to obtain a second decision tree.

[0105] Step 408: Determine the second mean number of misjudgments corresponding to the target leaf node.

[0106] Specifically, determine the misjudgment rate corresponding to the target leaf node and the number of samples corresponding to the target leaf node, and multiply the misjudgment rate corresponding to the target leaf node and the number of samples corresponding to the target leaf node to determine the second mean number of misjudgments.

[0107] Step 409 : When the difference between the second mean value of the number of false positives and the standard deviation of the number of false positives is smaller than the first mean value of the number of false positives, the warning accuracy of the second decision tree is determined.

[0108] It should be noted that if the difference between the second mean of the number of false positives and the standard deviation of the number of false positives is smaller than the first mean of the number of false positives, it means that the warning accuracy of the second decision tree is higher than that of the first decision tree.

[0109] Step 410: When the warning accuracy of the second decision tree is less than the threshold, continue to prune the second decision tree until the warning accuracy of the pruned decision tree is greater than the threshold.

[0110] Step 411: Determine the pruned decision tree as the rock burst warning model corresponding to the target mining area.

[0111] The embodiment of the present disclosure first acquires a training data set and a verification data set corresponding to a target mining area, then randomly selects at least one index as a target index from each index corresponding to each parameter type multiple times, and constructs a decision tree corresponding to each selected target index based on each selected target index and the training data set. Then, the warning accuracy corresponding to each decision tree is verified based on the verification data set. In the case where the maximum warning accuracy is less than a threshold value, the first misjudgment frequency mean and the misjudgment frequency standard deviation corresponding to the last end sub-tree of the first decision tree and the second misjudgment frequency mean corresponding to the target leaf node are determined. In the case where the difference between the second misjudgment frequency mean and the misjudgment frequency standard deviation is less than the first misjudgment frequency mean, the warning accuracy of the second decision tree is determined. In the case where the warning accuracy of the second decision tree is less than the threshold value, pruning of the second decision tree is continued until the warning accuracy of the pruned decision tree is greater than the threshold value. Thus, in the case where the maximum warning accuracy of the multiple decision trees is less than the threshold value, the first decision tree corresponding to the maximum warning accuracy can be accurately pruned, so that the accuracy of the generated rock burst warning model can be further improved, and the target mining area can be further accurately warned.

[0112] Figure 5 The flowchart of the method for generating a rock burst warning model provided by an embodiment of the present disclosure is shown in Figure 5 The method for generating a rock burst warning model can include the following steps:

[0113] Step 501, acquiring a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set includes a danger level label and a danger result of each index corresponding to each parameter type in the target mining area.

[0114] Step 502, randomly selecting at least one index as a target index from each index corresponding to each parameter type multiple times.

[0115] Step 503, constructing a decision tree corresponding to each selected target index based on each selected target index and the training data set.

[0116] Step 504, verifying the warning accuracy corresponding to each decision tree based on the verification data set.

[0117] Step 505, in the case where the maximum warning accuracy is greater than or equal to a threshold value, determining the first decision tree corresponding to the maximum warning accuracy as a rock burst warning model.

[0118] It should be noted that if the maximum early warning accuracy is greater than the threshold, it indicates that the performance of the first decision tree corresponding to the maximum early warning accuracy can meet the demand, and therefore the first decision tree corresponding to the maximum early warning accuracy can be directly determined as the impact and pressure early warning model.

[0119] In the embodiments of the present disclosure, first, the training data set and the verification data set corresponding to the target mining area are obtained, then at least one index is randomly selected as a target index from each index corresponding to each parameter type, and a decision tree corresponding to each selected target index is constructed based on each selected target index and the training data set, and then the early warning accuracy corresponding to each decision tree is verified based on the verification data set, and in the case that the maximum early warning accuracy is greater than or equal to the threshold, the first decision tree corresponding to the maximum early warning accuracy is determined as the impact and pressure early warning model. Thus, the index that has a greater impact on the target mining area can be accurately determined, so that the accuracy of the impact and pressure early warning model generated based on the index that has a greater impact on the target mining area can be improved, and then the target mining area can be accurately provided with a danger warning.

[0120] In order to realize the above-mentioned embodiments, the present disclosure further provides an impact and pressure early warning model generation device.

[0121] Figure 6 A structural schematic diagram of the impact and pressure early warning model generation device provided by the embodiments of the present disclosure.

[0122] As shown in Figure 6 , the impact and pressure early warning model generation device 600 can include:

[0123] The first obtaining module 610 is configured to obtain a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set includes a danger level label and a danger result of each index corresponding to each parameter type under the target mining area.

[0124] The second obtaining module 620 is configured to randomly select at least one index as a target index from each index corresponding to each parameter type.

[0125] The construction module 630 is configured to construct a decision tree corresponding to each selected target index based on each selected target index and the training data set.

[0126] The verification module 640 is configured to verify the early warning accuracy corresponding to each decision tree based on the verification data set.

[0127] The pruning module 650 is configured to, in the case that the maximum early warning accuracy is less than the threshold, prune the first decision tree corresponding to the maximum early warning accuracy, so that the early warning accuracy of the pruned decision tree is greater than the threshold.

[0128] The determining module 660 is configured to determine the pruned decision tree as the rock burst early warning model corresponding to the target mining area.

[0129] Optionally, the constructing module 630 is configured to:

[0130] determine a first information gain between each target index and the training data set;

[0131] determine a first index with the maximum first information gain as a root node of the decision tree;

[0132] divide the training data set into a plurality of first subsets based on the first index, wherein the risk results corresponding to the first index in each sample data of each first subset are the same;

[0133] in a case where the risk level labels corresponding to each sample data in the first subset are different, determine a second information gain between other target indexes and the first subset in addition to the first index;

[0134] determine a second index with the maximum second information gain as an internal node under the root node;

[0135] divide the first subset into a plurality of second subsets based on the second index until each node corresponds to a subset that cannot be further divided, to obtain a decision tree, wherein the risk results corresponding to the second index in each sample data of each second subset are the same.

[0136] Optionally, the constructing module 630 is further configured to:

[0137] in a case where the risk level labels corresponding to each sample data in the first subset are the same, determine the risk level labels corresponding to each sample data in the first subset as a leaf node under the root node.

[0138] Optionally, the pruning module 650 is configured to:

[0139] determine a first false judgment frequency mean and a false judgment frequency standard deviation corresponding to a last end subtree in the first decision tree;

[0140] determine a target leaf node to be replaced corresponding to the last end subtree;

[0141] replace the last end subtree in the first decision tree with the target leaf node to obtain a second decision tree;

[0142] determine a second false judgment frequency mean corresponding to the target leaf node;

[0143] determine the early warning accuracy of the second decision tree in a case where a difference between the second false judgment frequency mean and the false judgment frequency standard deviation is less than the first false judgment frequency mean;

[0144] In a case where the early warning accuracy of the second decision tree is less than the threshold, pruning of the second decision tree is continued until the early warning accuracy of the pruned decision tree is greater than the threshold.

[0145] Optionally, the verification module 640 is configured to:

[0146] input the risk results of the indexes of each piece of sample data in the verification data set into the decision tree to obtain a predicted risk level of each piece of sample data in the verification data set;

[0147] determine a first quantity of sample data whose predicted risk level is different from the corresponding risk level label;

[0148] determine, as the corresponding early warning accuracy of the decision tree, a ratio between the first quantity and a total quantity of sample data in the verification data set.

[0149] Optionally, the first obtaining module 610 is configured to:

[0150] obtain historical data corresponding to each parameter type of the target mining area every day;

[0151] process the historical data corresponding to each parameter type of the target mining area every day based on the functions corresponding to the indexes of each parameter type to obtain risk results corresponding to the indexes of each parameter type;

[0152] obtain risk level labels corresponding to the target mining area every day;

[0153] generate a plurality of pieces of sample data according to the risk results corresponding to the indexes of each day and the risk level labels;

[0154] divide the plurality of pieces of sample data into a training data set and a verification data set.

[0155] Optionally, the preprocessing module is configured to:

[0156] in a case where the historical data corresponding to any parameter type every day is missing, determine whether the functions corresponding to the indexes of any parameter type involve a time parameter;

[0157] in a case where the functions of any index of any parameter type involve the time parameter, fill in the missing data by using a replacement or interpolation method, or in a case where the functions of the indexes of any parameter type do not involve the time parameter, delete the data of the day by using a deletion method.

[0158] Optionally, the determination module 660 is further configured to:

[0159] in a case where the maximum early warning accuracy is greater than or equal to the threshold, determine the first decision tree corresponding to the maximum early warning accuracy as the impact and pressure early warning model.

[0160] The functions and specific implementation principles of the above modules in the embodiments of the present disclosure can be referred to the above method embodiments and will not be repeated here.

[0161] The device for generating the rock burst warning model of the disclosed embodiment first obtains a training data set and a verification data set corresponding to the target mining area, and then randomly selects at least one indicator as the target indicator from each indicator corresponding to each parameter type multiple times, and constructs a decision tree corresponding to each selected target indicator based on the target indicator and the training data set each time, and then verifies the warning accuracy corresponding to each decision tree based on the verification data set, and when the maximum warning accuracy is less than a threshold, prunes the first decision tree corresponding to the maximum warning accuracy so that the warning accuracy of the pruned decision tree is greater than the threshold, and finally determines the pruned decision tree as the rock burst warning model corresponding to the target mining area. In this way, the indicators that have a greater impact on the target mining area can be accurately determined, thereby improving the accuracy of the rock burst warning model generated based on the indicators that have a greater impact on the target mining area, and further accurately providing a danger warning for the target mining area.

[0162] In order to implement the above embodiments, the present disclosure also proposes an electronic device, including: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the method for generating the impact ground pressure warning model proposed in the above embodiments of the present disclosure.

[0163] In order to implement the above embodiments, the present disclosure also proposes a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for generating a rock burst warning model proposed in the above embodiments of the present disclosure is implemented.

[0164] In order to implement the above embodiments, the present disclosure further proposes a computer program product, including a computer program, which, when executed by a processor, implements the method for generating a rock burst warning model as proposed in the above embodiments of the present disclosure.

[0165] Figure 7 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present disclosure is shown. Figure 7 The electronic device 12 shown is only an example and should not limit the functionality and scope of use of the embodiments of the present disclosure.

[0166] like Figure 7 As shown, electronic device 12 is implemented as a general-purpose computing device. Components of electronic device 12 may include, but are not limited to, one or more processors or processing units 16, system memory 28, and a bus 18 that connects various system components (including system memory 28 and processing unit 16).

[0167] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of such architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnection (PCI) bus.

[0168] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0169] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be configured to read and write non-removable, non-volatile magnetic media ( Figure 7 Not shown, often called a "hard drive"). Although Figure 7 Although not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a Compact Disc Read Only Memory (hereinafter referred to as: CD-ROM), a Digital Video Disc Read Only Memory (hereinafter referred to as: DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the various embodiments of the present disclosure.

[0170] Program / utility 40 having a set of program modules 42 can be stored in memory 28 by way of example, such program modules 42 include an operating system, one or more application programs, other program modules, and program data, each or some combination thereof, which can include implementation of the network environment in each or some combination thereof. Program modules 42 generally carry out the functions and / or methodologies described in embodiments of the disclosure.

[0171] Electronic device 12 can also communicate with one or more external devices 14 such as a keyboard or pointing device, a display 24, etc. which can be used in connection with the electronic device 12. Furthermore, electronic device 12 can communicate with one or more devices that enable a user to interact with electronic device 12 and / or one or more devices that enable electronic device 12 to communicate with one or more other computing devices. Such communication can be via input / output (I / O) interface 22. Still yet, electronic device 12 can communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or the Internet) through network adapter 20. As an example, network adapter 20 can include a modem, a network card (wireless or wired), or other well-known interface devices. As shown, network adapter 20 communicates with the other components of electronic device 12 through bus 18. It should be appreciated that although not shown, other hardware and / or software components could be used in conjunction with electronic device 12. These include, but are not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0172] Processing unit 16 executes the various functions and processes of the embodiments by running programs stored in system memory 28. These programs include, for example, program modules for implementing the methods described in the foregoing embodiments.

[0173] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In the specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Also, the specific features, structures, materials or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0174] Moreover, the terms "first", "second", "third", etc. are used herein only to describe different instances, and do not imply or suggest relative importance or a number of indicated technical features. Thus, features defined with "first", "second" can include at least one of such features, either explicitly or implicitly. In the description of the disclosure, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise explicitly and specifically limited.

[0175] Any process or method descriptions or blocks in flow charts herein, and elsewhere, can be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps in the process. Alternate implementations are included within the scope of the preferred embodiments of this disclosure in which additional functionality can be added or some functionality can be removed, by, for example, adding one or more steps performing a similar or reciprocal function, combining two or more steps into a single step, or splitting one step into two or more steps.

[0176] The logic and / or steps represented in flow charts herein, and elsewhere, can be considered as a sequence of executable instructions, for implementing logic functions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch instructions from a instruction execution system, apparatus, or device and execute instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can specifically include a hardware apparatus, such as a wired or wireless communication link, a portable computer diskette, a RAM, a ROM, an EPROM, a FLASH memory, or a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be used by the instruction execution system, apparatus, or device.

[0177] It should be understood that portions of the present disclosure can be realized with hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be realized with software or firmware stored in a memory and executed by a suitable instruction execution system. As such, if realized with hardware and in another embodiment, any one or a combination of the following technologies known in the art can be used: discrete logic circuit with logic gate circuit for implementing logic functions on data signals, application specific integrated circuit with suitable combination logic gate circuit, programmable gate array (PGA), field programmable gate array (FPGA), etc.

[0178] Those skilled in the art of the present technology can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by a program instructing the relevant hardware, and the program can be stored in a computer readable storage medium. When the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0179] In addition, each functional unit in each embodiment of the present disclosure can be integrated into one processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The above-mentioned integrated module can be realized in the form of hardware or in the form of a software functional module. When the integrated module is realized in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0180] The above-mentioned storage medium can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and cannot be understood as limiting the present disclosure, and those skilled in the art can make changes, modifications, replacements and variations to the above-mentioned embodiments within the scope of the present disclosure.

Claims

1. A method for generating a model for predicting rock burst, characterized in that, The method comprises the following steps: obtaining a training data set and a verification data set corresponding to a target mining area, wherein each sample data in the training data set and the verification data set comprises a danger level label and a danger result of each index corresponding to each parameter type in the target mining area; randomly selecting at least one index as a target index from each index corresponding to each parameter type multiple times; based on each selected target index and the training data set, constructing a decision tree corresponding to each selected target index, specifically comprising: determining a first information gain between each target index and the training data set; determining a first index with the largest first information gain as a root node of the decision tree; based on the first index, dividing the training data set into multiple first subsets, wherein the danger result of the first index corresponding to each sample data in each first subset is the same; in the case that the danger level label corresponding to each sample data in the first subset is different, determining a second information gain between other target indexes except the first index and the first subset; determining a second index with the largest second information gain as an internal node under the root node; based on the second index, dividing the first subset into multiple second subsets until the subset corresponding to each node cannot be further divided, to obtain a decision tree, wherein the danger result of the second index corresponding to each sample data in each second subset is the same; verifying the early warning accuracy of each decision tree based on the verification data set; determining a first false judgment frequency mean value and a false judgment frequency standard deviation corresponding to the last end of the first decision tree; determining a target leaf node to be replaced corresponding to the last end of the tree; replacing the last end of the tree in the first decision tree with the target leaf node to obtain a second decision tree; determining a second false judgment frequency mean value corresponding to the target leaf node; in the case that the difference between the second false judgment frequency mean value and the false judgment frequency standard deviation is less than the first false judgment frequency mean value, determining the early warning accuracy of the second decision tree; in the case that the early warning accuracy of the second decision tree is less than a threshold value, continuing to prune the second decision tree until the early warning accuracy of the pruned decision tree is greater than the threshold value; determining the pruned decision tree as the rock burst early warning model corresponding to the target mining area; after dividing the training data set into multiple first subsets based on the first index, further comprising: in the case that the danger level label corresponding to each sample data in the first subset is the same, determining the danger level label corresponding to each sample data in the first subset as a leaf node under the root node.

2. The method of claim 1, wherein, verifying the early warning accuracy of each decision tree based on the verification data set, comprising: inputting the danger result of each index of each sample data in the verification data set into the decision tree to obtain the predicted danger level of each sample data in the verification data set; determining a first number of sample data with a predicted danger level different from the corresponding danger level label. A ratio between the first quantity and a total quantity of sample data in the verification data set is determined as a pre-warning accuracy corresponding to the decision tree.

3. The method of claim 1, wherein, The training data set and the verification data set corresponding to the target mining area are obtained, including: The historical data corresponding to each parameter type of the target mining area every day is obtained. The historical data corresponding to each parameter type every day is processed based on the function corresponding to each index of each parameter type, to obtain the dangerous result corresponding to each index of each parameter type. The dangerous level label corresponding to the target mining area every day is obtained. The multiple pieces of sample data are generated according to the dangerous result corresponding to each index every day and the dangerous level label. The multiple pieces of sample data are divided into the training data set and the verification data set.

4. The method of claim 3, wherein, Further comprising: In the case that the historical data corresponding to any parameter type every day is missing, it is determined whether the function corresponding to each index of the any parameter type involves a time parameter; In the case that the function corresponding to any index of the any parameter type involves the time parameter, the missing data is filled in by replacement or interpolation; or in the case that the function corresponding to each index of the any parameter type does not involve the time parameter, the data of the day is deleted.

5. The method of claim 1, wherein, After verifying the pre-warning accuracy corresponding to each decision tree based on the verification data set, further comprising: In the case that the maximum pre-warning accuracy is greater than or equal to the threshold value, the first decision tree corresponding to the maximum pre-warning accuracy is determined as the impact and pressure pre-warning model.

6. A device for generating a model for predicting rock burst, characterized by comprising: The device is used to implement the generation method of the rock burst pre-warning model according to claim 1, and the device comprises: A first obtaining module is configured to obtain a training data set and a verification data set corresponding to a target mining area, wherein each piece of sample data in the training data set and the verification data set comprises a dangerous level label and a dangerous result corresponding to each index of each parameter type under the target mining area. A second obtaining module is configured to randomly select at least one index as a target index from each index corresponding to each parameter type multiple times. A construction module is configured to construct a decision tree corresponding to each selected target index based on each selected target index and the training data set. A verification module is configured to verify a pre-warning accuracy corresponding to each decision tree based on the verification data set. A pruning module is configured to prune a first decision tree corresponding to the maximum pre-warning accuracy in the case that the maximum pre-warning accuracy is less than a threshold value, so that the pre-warning accuracy of the pruned decision tree is greater than the threshold value. A determination module is configured to determine the pruned decision tree as an impact and pressure pre-warning model corresponding to the target mining area.

7. An electronic device, comprising: The device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the generation method of the rock burst pre-warning model according to any one of claims 1-5 when executing the program.

Images