Support parameter prediction model training method, determination method, device and equipment

Abstract

The embodiment of the application discloses a kind of support parameter prediction model training method, support parameter determination method, device and electronic equipment, belong to the surrounding rock support technical field of geotechnical engineering and mining engineering, can solve the problem that the determination of roadway support parameter in prior art relies on a large number of manpower and the determined support parameter is inaccurate.The method comprises the following steps: constructing an initial support parameter prediction model;Obtain a plurality of first sample data, each first sample data comprising: actual roadway parameters and actual support parameters;The plurality of first sample data is divided into a training set and a test set;The initial support parameter prediction model is trained by the training set, and a trained support parameter prediction model is obtained;The test set is input into the trained support parameter prediction model, and a predicted support parameter is obtained;The trained support parameter prediction model is optimized based on the predicted support parameter, and a well-trained support parameter prediction model is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of surrounding rock support in geotechnical engineering and mining engineering, and particularly relates to a training method of a support parameter prediction model, a support parameter determination method and device, and an electronic device. BACKGROUND

[0002] Coal mines in China are mainly mined underground, and a large number of roadways need to be excavated underground. It is of great significance to keep the roadways unblocked and the surrounding rock stable for coal mine construction and production. With the continuous increase of mining depth, mining range and mining intensity, the geological conditions are becoming more complex, the conditions faced by roadway support are becoming more and more complex, the requirements and standards for support are becoming higher and higher, and the difficulty of support is increasing.

[0003] Since the roadway support parameters are difficult to be expressed by accurate calculation formulas, the existing technology mainly determines the support parameters by engineering classification method, dynamic information design method and numerical simulation method. This process requires construction technicians to have rich field construction experience, but there are many uncertain factors caused by manual work. The support parameters are usually determined only through rough calculation and analysis, which leads to the fact that the quality of the coal mine roadway support determined according to the support parameters cannot meet the engineering needs. SUMMARY

[0004] The embodiments of the present application provide a training method of a support parameter prediction model, a support parameter determination method and device, and an electronic device to solve the problem that the determination of the roadway support parameters in the prior art relies on a large amount of manpower and the determined support parameters are inaccurate.

[0005] In a first aspect, the present application provides a training method of a support parameter prediction model. The method comprises: constructing an initial support parameter prediction model; obtaining a plurality of first sample data, each first sample data comprising: an actual roadway parameter and an actual support parameter; dividing the plurality of first sample data into a training set and a test set; training the initial support parameter prediction model through the training set to obtain a trained support parameter prediction model; inputting the test set into the trained support parameter prediction model to obtain a predicted support parameter; and optimizing the trained support parameter prediction model based on the predicted support parameter to obtain a trained support parameter prediction model.

[0006] In some embodiments of the present application, before obtaining the plurality of first sample data, the method further comprises: constructing a roadway support database based on obtained historical roadway support data; and obtaining the plurality of first sample data comprises: obtaining the plurality of first sample data from the roadway support database.

[0007] In some embodiments of the present application, before obtaining the plurality of first sample data from the roadway support database, the method further comprises: constructing a plurality of random separation models by a random forest algorithm, each support parameter corresponding to a random separation model; obtaining a plurality of second sample data, each second sample data comprising: an actual roadway parameter and an actual support parameter; inputting the plurality of second sample data into the plurality of random separation models respectively to obtain a plurality of first rankings, each first ranking indicating an importance degree ranking of the plurality of roadway parameters for one support parameter; determining a target roadway parameter based on the plurality of first rankings; and obtaining the plurality of first sample data from the roadway support database, comprising: obtaining the plurality of first sample data corresponding to the target roadway parameter from the roadway support database.

[0008] In some embodiments of the present application, the initial support parameter prediction model is trained by the training set to obtain a trained support parameter prediction model, comprising: training the initial support parameter prediction model by the training set to obtain a plurality of trained support parameter prediction sub-models, each support parameter corresponding to a trained support parameter prediction sub-model; and fusing the plurality of trained support parameter prediction sub-models to obtain the trained support parameter prediction model.

[0009] In some embodiments of the present application, the type of support parameter comprises: factor type and numerical type; and the plurality of trained support parameter prediction sub-models comprise: a classification prediction model and a regression prediction model, the type of support parameter corresponding to the classification prediction model being the factor type, and the type of support parameter corresponding to the regression prediction model being the numerical type.

[0010] In some embodiments of the present application, the trained support parameter prediction model is optimized based on the predicted support parameter to obtain a trained support parameter prediction model, comprising: determining the accuracy of the trained support parameter prediction model based on the predicted support parameter and the actual support parameter; obtaining the trained support parameter prediction model in the case that the accuracy is within an accuracy threshold range; and adjusting the parameters of the initial support parameter prediction model for further training until the accuracy is within the accuracy threshold range to obtain the trained support parameter prediction model in the case that the accuracy is not within the accuracy threshold range.

[0011] In the second aspect of the embodiments of the present application, a support parameter determination method is provided, comprising: obtaining a roadway parameter of a roadway to be supported; and inputting the roadway parameter of the roadway to be supported into a trained support parameter prediction model to obtain a support parameter of the roadway to be supported; wherein the trained support parameter prediction model is obtained by training the support parameter prediction model by the training method of the support parameter prediction model according to the first aspect.

[0012] In a third aspect, the application provides a device for training a support parameter prediction model, comprising: a construction module, an acquisition module, a division module, a training model, an input module, and an optimization module; the construction module is configured to construct an initial support parameter prediction model; the acquisition module is configured to acquire a plurality of first sample data, each of which comprises actual roadway parameters and actual support parameters; the division module is configured to divide the plurality of first sample data into a training set and a test set; the training model is configured to train the initial support parameter prediction model based on the training set to obtain a trained support parameter prediction model; the input module is configured to input the test set into the trained support parameter prediction model to obtain predicted support parameters; and the optimization module is configured to optimize the trained support parameter prediction model based on the predicted support parameters to obtain a trained support parameter prediction model.

[0013] In some embodiments of the application, the construction module is further configured to, before acquiring the plurality of first sample data, construct a roadway support database based on acquired historical roadway support data; and the acquisition module is specifically configured to acquire the plurality of first sample data from the roadway support database.

[0014] In some embodiments of the application, the device further comprises a determination module; the construction module is further configured to, before acquiring the plurality of first sample data from the roadway support database, construct a plurality of random separation models by using a random forest algorithm, each support parameter corresponding to a random separation model; the acquisition module is further configured to acquire a plurality of second sample data, each of which comprises actual roadway parameters and actual support parameters; the input module is further configured to input the plurality of second sample data into the plurality of random separation models respectively to obtain a plurality of first rankings, each of which indicates an importance ranking of the plurality of roadway parameters with respect to a support parameter; and the determination module is configured to determine target roadway parameters based on the plurality of first rankings; and the acquisition module is specifically configured to acquire the plurality of first sample data corresponding to the target roadway parameters from the roadway support database.

[0015] In some embodiments of the application, the device further comprises a fusion module; the training module is specifically configured to train the initial support parameter prediction model based on the training set to obtain a plurality of trained support parameter prediction sub-models, each support parameter corresponding to a trained support parameter prediction sub-model; and the fusion module is configured to fuse the plurality of trained support parameter prediction sub-models to obtain the trained support parameter prediction model.

[0016] In some embodiments of the application, the types of support parameters include factor types and numerical types; and the plurality of trained support parameter prediction sub-models include classification prediction models and regression prediction models, the types of support parameters corresponding to the classification prediction models being factor types, and the types of support parameters corresponding to the regression prediction models being numerical types.

[0017] In some embodiments of the present application, the optimization module is specifically configured to determine the accuracy of the trained support parameter prediction model based on the predicted support parameters and the actual support parameters; in the case that the accuracy is within the accuracy threshold range, the trained support parameter prediction model is obtained; in the case that the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted for further training until the accuracy is within the accuracy threshold range, and the trained support parameter prediction model is obtained.

[0018] In a fourth aspect, the present application provides a support parameter determination device, which comprises an acquisition module and an input module; the acquisition module is configured to acquire the roadway parameters of a roadway to be supported; the input module is configured to input the roadway parameters of the roadway to be supported into a trained support parameter prediction model to obtain the support parameters of the roadway to be supported; wherein the trained support parameter prediction model is obtained by training the support parameter prediction model according to the training method of the support parameter prediction model.

[0019] In a fifth aspect, the present application provides an electronic device, which comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, and when the program or instruction is executed by the processor, the steps of the training method of the support parameter prediction model according to the first aspect or the steps of the support parameter determination method are implemented.

[0020] In a sixth aspect, the present application provides a readable storage medium, which stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the training method of the support parameter prediction model according to the first aspect or the steps of the support parameter determination method are implemented.

[0021] The technical scheme provided by the embodiments of the present application has the following advantages compared with the prior art:

[0022] In the embodiment of the present application, an initial support parameter prediction model is constructed; a plurality of first sample data are obtained, each first sample data comprising: actual roadway parameters and actual support parameters; the plurality of first sample data are divided into a training set and a test set; the initial support parameter prediction model is trained through the training set to obtain a trained support parameter prediction model; the test set is input into the trained support parameter prediction model to obtain predicted support parameters; the trained support parameter prediction model is obtained by optimizing the trained support parameter prediction model based on the predicted support parameters. The finally obtained trained support parameter prediction model is trained based on a large amount of training set and optimized through the test set, which ensures the accuracy of support parameter prediction. At the same time, the trained support parameter prediction model can be used for the prediction of roadway support parameters, reduces the dependence on the experience of technical personnel (reduces the uncertainty caused by the experience of different technical personnel), greatly simplifies the process of determining support parameters, and is beneficial to more reasonable, rapid and sufficient stable support of coal mine roadway surrounding rock, thereby improving the design efficiency of the entire coal mine roadway support scheme. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments and the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained from these drawings.

[0024] Figure 1 One of the flowcharts of the training method of the support parameter prediction model provided by the embodiments of the present application;

[0025] Figure 2 The second flowchart of the training method of the support parameter prediction model provided by the embodiments of the present application;

[0026] Figure 3 The third flowchart of the training method of the support parameter prediction model provided by the embodiments of the present application;

[0027] Figure 4 The fourth flowchart of the training method of the support parameter prediction model provided by the embodiments of the present application;

[0028] Figure 5 The fifth flowchart of the training method of the support parameter prediction model provided by the embodiments of the present application;

[0029] Figure 6 The flowchart of the support parameter determination method provided by the embodiments of the present application;

[0030] Figure 7A structural block diagram of a training device of a support parameter prediction model provided by an embodiment of the present application is provided.

[0031] Figure 8 A structural block diagram of a support parameter determination device provided by an embodiment of the present application is provided.

[0032] Figure 9 A hardware structure schematic diagram of an electronic device provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0034] The terms "first", "second", and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be exchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally a category and do not limit the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / ", generally indicates that the front and rear associated objects are in a "or" relationship.

[0035] The electronic device in the embodiments of the present application can be a mobile electronic device or a non-mobile electronic device. The mobile electronic device can be a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. The non-mobile electronic device can be a personal computer (PC), a television (TV), a cashier machine, or a self-service machine, etc. The embodiments of the present application are not limited specifically.

[0036] The execution subject of the support parameter prediction model training method or the support parameter determination method provided in the embodiments of the present application can be the electronic device (including a mobile electronic device and a non-mobile electronic device) described above, or can be a functional module and / or a functional entity capable of implementing the support parameter prediction model training method or the support parameter determination method in the electronic device. The specific implementation can be determined according to actual use requirements, and the embodiments of the present application are not limited in this regard.

[0037] The support parameter prediction model training method and the support parameter determination method provided in the embodiments of the present application will be described in detail below in combination with the accompanying drawings, specific embodiments and application scenarios.

[0038] As shown in FIG. 1, Figure 1 The embodiments of the present application provide a support parameter prediction model training method, and the following exemplary description of the support parameter prediction model training method provided in the embodiments of the present application is based on the execution subject being an electronic device. The method can include the following steps 101 to 106.

[0039] 101. Construct an initial support parameter prediction model.

[0040] It can be understood that the initial support parameter prediction model is obtained based on a machine learning algorithm, and the specific machine learning algorithm used is not limited in the embodiments of the present application. For example, it can be a support vector machine (SVM) algorithm, a neural network algorithm, a gradient boosting regression tree, a random forest algorithm, etc.

[0041] 102. Obtain a plurality of first sample data.

[0042] Each first sample data includes an actual roadway parameter and an actual support parameter.

[0043] It can be understood that the roadway parameter includes at least one of the following: coal seam thickness, coal seam inclination, coal seam number, coal seam strength, immediate roof thickness, immediate roof strength, basic roof thickness, basic roof strength, immediate floor thickness, immediate floor strength, old floor thickness, old floor strength, surrounding rock fracture development, roadway service life, cross section shape, roadway burial depth, roadway height, and roadway width. The roadway environment of a coal mine is relatively complex, and the roadway parameter can also include other parameters, which are not limited in the embodiments of the present application.

[0044] It can be understood that the support parameter includes at least one of the following: diameter, length, spacing and row spacing of roof anchor rod and two-side anchor rod, diameter, length, spacing, row spacing and arrangement of roof anchor cable, and the support parameter can also include other parameters known in the art, which are not limited in the embodiments of the present application.

[0045] It can be understood that the plurality of first sample data can be based on the actual coal mine engineering obtained original data, also can be based on the third party database obtained data, specific according to actual circumstances to determine, the embodiment of the application is not limited.

[0046] It can be understood that in the data analysis process, the dimension of the variable has a great influence on the calculation of the coefficient, the distance and the weight, resulting in that the variable with larger number occupies a larger proportion, therefore, it is crucial to standardize the data. The standardization is dimensionless processing to the data, so that the data with different dimensions can be compared horizontally on the same order of magnitude, and the error caused by the difference in data level is reduced.

[0047] Optionally, the plurality of first sample data is obtained by: obtaining a plurality of original sample data; and performing standardization processing on the plurality of original sample data to obtain the plurality of first sample data. In this way, each first sample data is obtained by standardization processing, and the standardization processing method can be a range method, a Z-score standardization method or the like, as long as the purpose of enabling the data with different dimensions to be compared on the same order of magnitude to reduce the error caused by the difference in data level is achieved. The embodiment of the application is not limited.

[0048] 103. The plurality of first sample data is divided into a training set and a test set.

[0049] It can be understood that the plurality of first sample data can be divided into a training set and a test set according to a preset ratio, or the plurality of first sample data can be randomly divided into a training set and a test set. The specific division manner is not limited in the embodiment of the application.

[0050] Exemplarily, 100 first sample data are divided according to a ratio of 7:3 between the training set and the test set, and a plurality of first sample data are randomly and non-retrievably collected from the 100 first sample data multiple times (one first sample data can be collected at a time, or a plurality of first sample data can be collected at a time), 70 first sample data obtained are taken as the training set, and the remaining 30 first sample data not collected are taken as the test set.

[0051] 104. The initial support parameter prediction model is trained through the training set to obtain a trained support parameter prediction model.

[0052] 105. The test set is input into the trained support parameter prediction model to obtain a predicted support parameter.

[0053] 106. The trained support parameter prediction model is optimized based on the predicted support parameter to obtain a trained support parameter prediction model.

[0054] It can be understood that the initial support parameter prediction model training process includes the selection of a kernel function used in the machine learning algorithm and the setting of a hyperparameter, etc.

[0055] Exemplarily, taking the initial support parameter prediction model based on the SVM algorithm as an example, the target of the initial support parameter prediction model training includes the kernel function and the hyperparameter used in the SVM algorithm, and the hyperparameter includes the penalty coefficient of the objective function and the coefficient of the kernel function. The kernel function can be selected from a radial basis kernel function, a linear kernel function, a polynomial kernel function and a Sigmoid kernel function. The four kernel functions are sequentially substituted into the initial support parameter prediction model, the errors of the model are calculated respectively, and the kernel function with the minimum error is selected as the optimal kernel function of the initial support parameter prediction model. The penalty coefficient of the objective function and the coefficient of the kernel function can be determined by any algorithm such as a genetic algorithm, a grid search method or an artificial ant colony algorithm. The final selected optimization algorithm is based on the penalty coefficient and the kernel function coefficient that make the model error minimum as the optimal hyperparameter of the final initial support parameter prediction model. Through the above optimization training, the nonlinear fitting capability of the initial support parameter prediction model is better.

[0056] In the embodiments of the present application, an initial support parameter prediction model is constructed, a plurality of first sample data are obtained, each first sample data including an actual roadway parameter and an actual support parameter, the plurality of first sample data are divided into a training set and a test set, the initial support parameter prediction model is trained through the training set to obtain a trained support parameter prediction model, the test set is input into the trained support parameter prediction model to obtain a predicted support parameter, and the trained support parameter prediction model is optimized based on the predicted support parameter to obtain a trained support parameter prediction model. The final trained support parameter prediction model is trained based on a large amount of training set and optimized through the test set, which ensures the accuracy of the support parameter prediction. At the same time, the trained support parameter prediction model can be used for the prediction of the roadway support parameter, reduces the dependence on the experience of technical personnel (reduces the uncertainty caused by the experience of different technical personnel), greatly simplifies the process of determining the support parameter, and is beneficial to more reasonable, rapid and sufficient stable support of the surrounding rock of the coal mine roadway, thereby improving the design efficiency of the entire coal mine roadway support scheme.

[0057] In some embodiments of the present application, the initial support parameter prediction model is trained based on the historical roadway support data. Figure 1 As shown in FIG. 1, before the step 102, the method further includes the following step 107, and the step 102 can be implemented by the following step 102a. Figure 2

[0058] 107. Construct a roadway support database based on the obtained historical roadway support data.

[0059] ​It can be understood that the database is the basis for data analysis and model establishment. In order to facilitate the training and optimization of the model, the existing coal mine supported roadway engineering technical data is collected to form an original data set containing a large number of samples. The collected supported roadway engineering technical data specifically includes roadway parameters and support parameters. All parameters can be expressed in the form of numbers and words. The parameters expressed in the form of words need to be quantified or converted into digital form to facilitate subsequent data processing and analysis. Among them, the case samples in the original data set collected must be coal mine roadways that have been supported and have good support effect.

[0060] It can be understood that in order to ensure the effectiveness of the data stored in the database, the original data set needs to be cleaned to obtain coal roadway support data and stored in the roadway support database. Specifically, the data distribution in the original data set is determined, and the data included in the original data set is cleaned according to the distribution. Among them, the cleaning method includes at least one of deleting abnormal data, replacing abnormal data, and supplementing missing data.

[0061] It can be understood that in order to facilitate the use of the subsequent model, the coal roadway support data can be standardized and stored in the roadway support database, or it can be standardized when used. The specific determination is made according to the actual needs, and the embodiments of the present application are not limited.

[0062] 102a, obtain a plurality of first sample data from the roadway support database.

[0063] In the embodiments of the present application, based on the obtained historical roadway support data, a roadway support database is constructed, and a plurality of first sample data is obtained from the roadway support database. By establishing the roadway support database, subsequent use can be facilitated, and new support parameters and roadway parameters can be stored in the roadway support database to continuously train the support parameter prediction model using new data, so that the prediction of the model is more accurate.

[0064] In some embodiments of the present application, the method further includes the following steps 102b to 102e in combination with Figure 2 As shown in Figure 3 Before the above step 102a, the method further includes the following steps 102b to 102e, and the above step 102a can be implemented by the following step 102f.

[0065] 102b, a plurality of random separation models are constructed by a random forest algorithm.

[0066] Each support parameter corresponds to a random separation model.

[0067] It can be understood that the number of support parameters to be predicted is the same as the number of random separation models.

[0068] Exemplarily, the support parameters include diameters, lengths, spacings and row spacings of the roof anchor rods and the two-side anchor rods, diameters, lengths, spacings, row spacings and arrangement modes of the roof anchor cables; and the random separation models include a random separation model for the roof anchor rod diameters, a random separation model for the roof anchor rod lengths, a random separation model for the roof anchor rod spacings, a random separation model for the roof anchor rod row spacings, a random separation model for the two-side anchor rod diameters, a random separation model for the two-side anchor rod lengths, a random separation model for the two-side anchor rod spacings, a random separation model for the two-side anchor rod row spacings, a random separation model for the roof anchor cable diameters, a random separation model for the roof anchor cable lengths, a random separation model for the roof anchor cable spacings, a random separation model for the roof anchor cable row spacings, and a random separation model for the roof anchor cable arrangement modes.

[0069] 102c, acquire a plurality of second sample data.

[0070] Each of the second sample data includes actual roadway parameters and actual support parameters.

[0071] It can be understood that the plurality of second sample data can be the same as or different from the plurality of first sample data. The plurality of second sample data can be part of the data acquired from the roadway support database, or can be all the data in the roadway support database, which is not limited by the embodiments of the present application.

[0072] 102d, input the plurality of second sample data into the plurality of random separation models respectively to obtain a plurality of first rankings.

[0073] Each of the first rankings indicates an importance ranking of the plurality of roadway parameters for one support parameter.

[0074] It can be understood that, for the random separation model corresponding to one support parameter, the output result obtained by inputting the plurality of second sample data into the model is an importance ranking of the plurality of roadway parameters for the one support parameter.

[0075] Exemplarily, the second sample data is: a first group of data: roadway width: 5, roadway height: 4, roof anchor rod diameter: 20; a second group of data: roadway width: 6, roadway height: 4, roof anchor rod diameter: 20; and a third group of data: roadway width: 5, roadway height: 5, roof anchor rod diameter: 20; the sample data is input into the random classification model for the roof anchor rod diameter, the first group of data: the prediction accuracy of the model is 75%; the second group of data: the prediction accuracy of the model is 74.5%; and the third group of data: the prediction accuracy of the model is 74%; then the error change obtained by changing the roadway width is 0.5%, the error change obtained by changing the roadway height is 1%, and for the roof anchor rod diameter, the importance of the roadway width is lower than that of the roadway height.

[0076] It can be understood that the specific error analysis can be mean square error (the result obtained by using mean square error is more accurate in actual engineering), and the Gini coefficient method can also be used, and the embodiments of the present application are not limited.

[0077] 102e, determining a target roadway parameter based on the plurality of first rankings.

[0078] It can be understood that among the roadway parameters, some parameters have a greater impact on the support parameters, and some parameters have a smaller impact on the support parameters. For the roadway parameters that have a smaller impact on the support parameters, the training can be performed without inputting in the model training process. In this way, the parameters in the support parameter prediction model training process can be reduced, so that the model converges faster and the training process is simpler.

[0079] It can be understood that for each support parameter, a first ranking of the roadway parameters is obtained, and based on the plurality of first rankings, the target roadway parameter is determined. Specifically, a score can be given to each roadway parameter, and the target roadway parameter is determined according to the sum of the scores of each support parameter. The target roadway parameter can also be determined based on the plurality of first rankings according to a target rule. The target rule can be weighted summation, average, etc., and the embodiments of the present application are not limited.

[0080] It can be understood that the roadway parameters are ranked from high to low according to the importance of the support parameters, and the first preset roadway parameters are determined as the target roadway parameters, or the roadway parameters with a score greater than or equal to a score threshold are determined as the target roadway parameters.

[0081] Exemplarily, for the random separation model of the roof anchor rod diameter, the importance of the roadway parameters is ranked from high to low as follows: coal seam thickness, coal seam inclination, coal seam number, coal seam strength, and direct roof thickness. The scores are as follows: coal seam thickness: 5, coal seam inclination: 4, coal seam number: 3, coal seam strength: 2, and direct roof thickness: 1. For the random separation model of the roof anchor rod length, the importance of the roadway parameters is ranked from high to low as follows: coal seam thickness: 5, coal seam number: 4, direct roof thickness: 3, coal seam inclination: 2, and coal seam strength: 1. The total scores are as follows: coal seam thickness: 10, coal seam inclination: 6, coal seam number: 7, coal seam strength: 3, and direct roof thickness: 4. Therefore, the importance is ranked from high to low as follows: coal seam thickness, coal seam number, coal seam inclination, direct roof thickness, and coal seam strength. If three roadway parameters are needed to be selected, the target roadway parameters are: coal seam thickness, coal seam number, and coal seam inclination.

[0082] 102f, obtaining a plurality of first sample data corresponding to the target roadway parameter from the roadway support database.

[0083] It can be understood that only the data corresponding to the target roadway parameter is obtained from the roadway support database, and the data of the remaining roadway parameters does not participate in the model training. At this time, each first sample data includes the target roadway parameter and the support parameter.

[0084] In the embodiment of the application, a plurality of random separation models are constructed through the random forest algorithm, and each support parameter corresponds to a random separation model. A plurality of second sample data are obtained, and each second sample data includes an actual roadway parameter and an actual support parameter. The plurality of second sample data are input into the plurality of classification models respectively to obtain a plurality of first rankings, and each first ranking indicates the importance ranking of the plurality of roadway parameters to one support parameter. The target roadway parameter is determined based on the plurality of first rankings, and a plurality of first sample data corresponding to the target roadway parameter are obtained from the roadway support database. In this way, the roadway parameters that have less influence on the support parameter are not used as sample data, which can reduce the input parameters of the support parameter prediction model, thereby effectively reducing the calculation amount in the model training process, reducing the complexity of the model training, and further improving the accuracy of the model prediction.

[0085] In some embodiments of the application, the target roadway parameter is determined based on the importance ranking of the plurality of roadway parameters to one support parameter. Figure 1 As shown in FIG. 1, the above step 104 can be implemented through the following steps 104a and 104b. Figure 4

[0086] 104a, training the initial support parameter prediction model through the training set to obtain a plurality of trained support parameter prediction sub-models.

[0087] Each support parameter corresponds to a trained support parameter prediction sub-model.

[0088] It can be understood that for each support parameter, an initial support parameter prediction sub-model is determined, and for the sub-model, the output result is only one, the training process is simpler, and the model is more easily converged.

[0089] 104b, fusing the plurality of trained support parameter prediction sub-models to obtain the trained support parameter prediction model.

[0090] In the embodiment of the application, the initial support parameter prediction model is trained through the training set to obtain a plurality of trained support parameter prediction sub-models. Each trained support parameter prediction sub-model corresponds to one support parameter, and the plurality of trained support parameter prediction sub-models are fused to obtain the trained support parameter prediction model. Compared with the model that inputs a plurality of parameters and outputs a plurality of parameters, the output of the initial support parameter prediction sub-model has only one result, so that the training of the model is simpler, the calculation amount is small, and the model is more easily converged.

[0091] ​In some embodiments of the present application, the types of the support parameters include factor types and numerical types; and the plurality of trained support parameter prediction sub-models include a classification prediction model and a regression prediction model, the type of the support parameter corresponding to the classification prediction model is a factor type, and the type of the support parameter corresponding to the regression prediction model is a numerical type.

[0092] It can be understood that the type of the support parameter is a factor type, such as the arrangement mode of the roof anchor cable, which is not a specific numerical value, and the output result of the trained support parameter prediction sub-model is a classification result, such as the arrangement mode of the roof anchor cable being mode 1; the type of the support parameter is a numerical type, such as the row spacing of the roof anchor cable, and the output result of the trained support parameter prediction sub-model is a specific numerical value, such as 3 meters.

[0093] Optionally, the initial support parameter prediction model is trained by using the training set to obtain a trained first sub-model and a trained second sub-model, the support parameter corresponding to the first sub-model is of a numerical type, and the support parameter corresponding to the second sub-model is of a factor type; and the trained first sub-model and the second sub-model are fused to obtain the trained support parameter prediction model. In this way, the number of sub-models and the type of the output result in the training process can be comprehensively considered.

[0094] In the embodiments of the present application, the types of the support parameters include factor types and numerical types; and the plurality of trained support parameter prediction sub-models include a classification prediction model and a regression prediction model, the type of the support parameter corresponding to the classification prediction model is a factor type, and the type of the support parameter corresponding to the regression prediction model is a numerical type. In this way, according to the different types of the support parameters, the output results of the models are also different, and the prediction accuracy of the model is further improved.

[0095] In some embodiments of the present application, the method further includes Figure 1 As shown in Figure 5 The step 106 can be implemented by the following steps 106a to 106c.

[0096] 106a, determining the accuracy of the trained support parameter prediction model based on the predicted support parameter and the actual support parameter.

[0097] 106b, obtaining the trained support parameter prediction model in the case that the accuracy is within the accuracy threshold range.

[0098] 106c, adjusting the parameters of the initial support parameter prediction model for continuous training until the accuracy is within the accuracy threshold range, and obtaining the trained support parameter prediction model.

[0099] It can be understood that, in the case that the type of the support parameter is numerical, the accuracy refers to the proportion of the samples whose output values of the model are the same as the true values in the test set; in the case that the type of the support parameter is factor, the goodness of fit R is used to evaluate the model 2 The size of the goodness of fit determines the degree of correlation, the greater the goodness of fit, the higher the degree of explanation of the independent variable to the dependent variable, the higher the percentage of the total variation caused by the independent variable, and the better the fitting degree of the model.

[0100] It can be understood that, in the case that the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted, including the selection of the kernel function and the setting of the hyperparameters.

[0101] It can be understood that, in the case that the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted and the training is continued, that is, the adjusted support parameter prediction model is trained through the training set, the test set is input into the trained support parameter prediction model, the predicted support parameter is obtained, the accuracy of the trained support parameter prediction model is determined based on the predicted support parameter and the actual support parameter, and the above process is repeated until the accuracy is within the accuracy threshold range, and the trained support parameter prediction model is obtained.

[0102] In the embodiments of the present application, the accuracy of the trained support parameter prediction model is determined based on the predicted support parameter and the actual support parameter; in the case that the accuracy is within the accuracy threshold range, the trained support parameter prediction model is obtained; in the case that the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted and the training is continued until the accuracy is within the accuracy threshold range, and the trained support parameter prediction model is obtained. In this way, the accuracy of the trained support parameter prediction model is ensured to be within the threshold range, that is, the accuracy of the model prediction is ensured, so that the accuracy of the support parameter determined by the model is higher and the support quality is more reliable.

[0103] The embodiments of the present application provide a support parameter determination method, as shown in Figure 6 The method comprises the following steps 201 and 202.

[0104] 201, obtaining the roadway parameters of a roadway to be supported.

[0105] 202, inputting the roadway parameters of the roadway to be supported into the trained support parameter prediction model to obtain the support parameters of the roadway to be supported.

[0106] The trained support parameter prediction model is obtained by the training method of the support parameter prediction model.

[0107] It should be noted that the relevant descriptions of steps 201 to 202 can refer to the relevant descriptions of the training method of the support parameter prediction model mentioned above, and will not be repeated here.

[0108] In this embodiment, the roadway parameters of the roadway to be supported are obtained, and these parameters are input into a trained support parameter prediction model to obtain the support parameters for the roadway. This reduces reliance on the experience of technical personnel (reducing the uncertainty caused by different personnel relying on experience), greatly simplifies the process of determining support parameters, and facilitates more reasonable, rapid, and sufficient stabilization support of the surrounding rock of coal mine roadways, thereby improving the design efficiency of the overall coal mine roadway support scheme.

[0109] Figure 7 This is a structural block diagram of a training device for a support parameter prediction model, as shown in an embodiment of this application. Figure 7 As shown, it includes: a construction module 701, an acquisition module 702, a partitioning module 703, a training model 704, an input module 705, and an optimization module 706. The construction module 701 is used to construct an initial support parameter prediction model; the acquisition module 702 is used to acquire multiple first sample data, each first sample data including: actual roadway parameters and actual support parameters; the partitioning module 703 is used to partition the multiple first sample data into a training set and a test set; the training model 704 is used to train the initial support parameter prediction model using the training set to obtain a trained support parameter prediction model; the input module 705 is used to input the test set into the trained support parameter prediction model to obtain predicted support parameters; and the optimization module 706 is used to optimize the trained support parameter prediction model based on the predicted support parameters to obtain a well-trained support parameter prediction model.

[0110] In some embodiments of this application, the construction module 701 is further configured to construct a roadway support database based on the acquired historical roadway support data before acquiring multiple first sample data; the acquisition module is specifically configured to acquire multiple first sample data from the roadway support database.

[0111] In some embodiments of the present application, the device further comprises a determining module; the constructing module 701 is further configured to, before obtaining the plurality of first sample data from the tunnel support database, construct a plurality of random separation models by using a random forest algorithm, each support parameter corresponding to a random separation model; the obtaining module 702 is further configured to obtain a plurality of second sample data, each second sample data comprising: an actual tunnel parameter and an actual support parameter; the input module 705 is further configured to input the plurality of second sample data into the plurality of random separation models respectively to obtain a plurality of first rankings, each first ranking indicating an importance ranking of the plurality of tunnel parameters with respect to a support parameter; the determining module is configured to determine the target tunnel parameter based on the plurality of first rankings; and the obtaining module 702 is specifically configured to obtain the plurality of first sample data corresponding to the target tunnel parameter from the tunnel support database.

[0112] In some embodiments of the present application, the device further comprises a fusion module; the training module 704 is specifically configured to train the initial support parameter prediction model by using the training set to obtain a plurality of trained support parameter prediction sub-models, each support parameter corresponding to a trained support parameter prediction sub-model; and the fusion module is configured to fuse the plurality of trained support parameter prediction sub-models to obtain the trained support parameter prediction model.

[0113] In some embodiments of the present application, the type of support parameter comprises: a factor type and a numerical type; and the plurality of trained support parameter prediction sub-models comprise: a classification prediction model and a regression prediction model, the type of support parameter corresponding to the classification prediction model being the factor type, and the type of support parameter corresponding to the regression prediction model being the numerical type.

[0114] In some embodiments of the present application, the optimization module 706 is specifically configured to determine the accuracy of the trained support parameter prediction model based on the predicted support parameter and the actual support parameter; in a case where the accuracy is within an accuracy threshold range, obtain the trained support parameter prediction model; and in a case where the accuracy is not within the accuracy threshold range, adjust the parameters of the initial support parameter prediction model for further training until the accuracy is within the accuracy threshold range, and obtain the trained support parameter prediction model.

[0115] It should be noted that the training device of the support parameter prediction model described above can be an electronic device in the method embodiments of the present application, or a functional module and / or functional entity capable of realizing the functions of the device embodiments in the electronic device, and the embodiments of the present application do not make any limitation.

[0116] In the embodiments of the present application, each module can implement the training method of the support parameter prediction model provided in the method embodiments, and achieve the same technical effects. To avoid repetition, no further description is given here.

[0117] The beneficial effects of the various implementation methods in this embodiment can be found in the beneficial effects of the corresponding implementation methods in the training embodiment of the above support parameter prediction model. To avoid repetition, they will not be described again here.

[0118] Figure 8 This is a structural block diagram of a support parameter determination device shown in an embodiment of this application, such as... Figure 8 As shown, it includes: an acquisition module 801 and an input module 802; the acquisition module 801 is used to acquire the roadway parameters of the roadway to be supported; the input module 802 is used to input the roadway parameters of the roadway to be supported into a trained support parameter prediction model to obtain the support parameters of the roadway to be supported; wherein, the trained support parameter prediction model is trained by the training method of the support parameter prediction model described above.

[0119] It should be noted that the aforementioned support parameter determination device can be the electronic device in the above method embodiment of this application, or it can be a functional module and / or functional entity in the electronic device that can realize the function of the device embodiment. This application embodiment does not limit it.

[0120] In this embodiment, each module can implement the support parameter determination method provided in the above method embodiment and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0121] The beneficial effects of the various implementation methods in this embodiment can be found in the beneficial effects of the corresponding implementation methods in the above-mentioned support parameter determination method embodiment. To avoid repetition, they will not be repeated here.

[0122] like Figure 9 As shown in the figure, an electronic device is also provided in this application embodiment. The electronic device may include: a processor 901, a memory 902, and a program or instructions stored in the memory 902 and executable on the processor 901. When the program or instructions are executed by the processor 901, they can implement the various processes of the training method of the support parameter prediction model provided in the above method embodiment, or the various processes of the support parameter determination method provided in the above method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0123] This application provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the training method for the support parameter prediction model provided in the above-described method embodiments, or the various processes of the support parameter determination method provided in the above-described method embodiments, and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0124] The embodiment of the present application further provides a computer program product, wherein the computer program product comprises a computer program or instructions, and when the computer program product runs on a processor, the processor executes the computer program or instructions to realize each process of the training method of the support parameter prediction model provided by the method embodiment or each process of the support parameter determination method provided by the method embodiment, and the same technical effects can be achieved. To avoid repetition, details are not repeated here.

[0125] The embodiment of the present application further provides a chip, which comprises a processor and a communication interface, wherein the communication interface is coupled with the processor, and the processor is used to run a program or instructions to realize each process of the training method of the support parameter prediction model or each process of the support parameter determination method provided by the method embodiment, and the same technical effects can be achieved. To avoid repetition, details are not repeated here.

[0126] It should be understood that the chip mentioned in the embodiment of the present application can also be referred to as a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0127] In several embodiments provided in the present application, it should be understood that the disclosed system, device, server and method can be implemented in other manners. For example, the described device embodiments are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0128] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0129] In addition, each functional unit in the embodiments of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be in the form of hardware or software functional units.

[0130] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0131] The above description and the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for training a support parameter prediction model, characterized in that, The method comprises: constructing an initial support parameter prediction model; obtaining a plurality of first sample data, each first sample data comprising: actual roadway parameters and actual support parameters; dividing the plurality of first sample data into a training set and a test set; training the initial support parameter prediction model through the training set to obtain a plurality of trained support parameter prediction sub-models, each support parameter corresponding to a trained support parameter prediction sub-model; the type of the support parameter includes: factor type and numerical type; the plurality of trained support parameter prediction sub-models include: a classification prediction model and a regression prediction model, the type of the support parameter corresponding to the classification prediction model is factor type, and the type of the support parameter corresponding to the regression prediction model is numerical type; wherein, the target of training the initial support parameter prediction model includes: the kernel function used in the SVM algorithm and the hyperparameter, the hyperparameter includes: the penalty coefficient of the objective function and the coefficient of the kernel function, and the kernel function includes: the radial basis kernel function, the linear kernel function, the polynomial kernel function and the Sigmoid kernel function, the four kernel functions are sequentially substituted into the initial support parameter prediction model, the errors of the models are calculated respectively, and the kernel function with the minimum error is selected as the optimal kernel function of the initial support parameter prediction model; the penalty coefficient of the objective function and the coefficient of the kernel function can be determined by the genetic algorithm / grid search method / artificial ant colony algorithm, the penalty coefficient and the kernel function coefficient that make the model error minimum are selected as the optimal hyperparameter of the final initial support parameter prediction model, and the nonlinear fitting ability of the initial support parameter prediction model is better through the above optimization training; fuse the plurality of trained support parameter prediction sub-models to obtain a trained support parameter prediction model; input the test set into the trained support parameter prediction model to obtain a predicted support parameter; determine the accuracy of the trained support parameter prediction model based on the predicted support parameter and the actual support parameter, wherein, when the type of the support parameter is numerical type, the accuracy is the proportion of samples with the same predicted support parameter and actual support parameter, and when the type of the support parameter is factor type, the accuracy is the goodness-of-fit value; when the accuracy is within the accuracy threshold range, the trained support parameter prediction model is obtained; when the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted for continuous training until the accuracy is within the accuracy threshold range, and the trained support parameter prediction model is obtained; before obtaining the plurality of first sample data from the roadway support database, the method further comprises: A plurality of random separation models are constructed by a random forest algorithm, each corresponding to a support parameter; a plurality of second sample data are obtained, each including an actual roadway parameter and an actual support parameter; the plurality of second sample data are input into the plurality of random separation models respectively, to obtain a plurality of first rankings, each indicating an importance ranking of a plurality of roadway parameters for a support parameter; and a target roadway parameter is determined based on the plurality of first rankings. The plurality of first sample data are obtained from the roadway support database, including the plurality of first sample data corresponding to the target roadway parameter.

2. The method of claim 1, wherein, Before the plurality of first sample data are obtained, the method further includes: A roadway support database is constructed based on the obtained historical roadway support data. The plurality of first sample data are obtained, including: The plurality of first sample data are obtained from the roadway support database.

3. A method of support parameter determination, characterized by, The method includes: A roadway parameter of a roadway to be supported is obtained; The roadway parameter of the roadway to be supported is input into a trained support parameter prediction model, to obtain a support parameter of the roadway to be supported; The trained support parameter prediction model is trained by the method of any one of claims 1-2.

4. A device for training a support parameter prediction model, characterized by, The device includes a construction module, an acquisition module, a division module, a trained model, a fusion module, an input module, and an optimization module. The construction module is configured to construct an initial support parameter prediction model. The acquisition module is configured to acquire a plurality of first sample data, each including an actual roadway parameter and an actual support parameter. The division module is configured to divide the plurality of first sample data into a training set and a test set. The training model is configured to train the initial support parameter prediction model by using the training set to obtain a plurality of trained support parameter prediction sub-models, one trained support parameter prediction sub-model corresponding to each support parameter; the types of the support parameters include factor type and numerical type; the plurality of trained support parameter prediction sub-models include a classification prediction model and a regression prediction model, the type of the support parameter corresponding to the classification prediction model is factor type, and the type of the support parameter corresponding to the regression prediction model is numerical type; wherein the target for training the initial support parameter prediction model includes a kernel function used in an SVM algorithm and a hyperparameter, the hyperparameter includes a penalty coefficient of an objective function and a coefficient of the kernel function, and the kernel function includes a radial basis kernel function, a linear kernel function, a polynomial kernel function, and a Sigmoid kernel function; the four kernel functions are sequentially substituted into the initial support parameter prediction model, the error of the model is calculated respectively, and the kernel function with the minimum error is selected as the optimal kernel function of the initial support parameter prediction model; the penalty coefficient of the objective function and the coefficient of the kernel function can be determined by using a genetic algorithm, a grid search method, and an artificial ant colony algorithm, the penalty coefficient and the coefficient of the kernel function that make the model error minimum are selected as the optimal hyperparameter of the final initial support parameter prediction model, and the nonlinear fitting capability of the initial support parameter prediction model is better through the above optimization training; The fusion module is configured to fuse the plurality of trained support parameter prediction sub-models to obtain a trained support parameter prediction model; The input module is configured to input the test set into the trained support parameter prediction model to obtain predicted support parameters; The optimization module is configured to determine the accuracy of the trained support parameter prediction model based on the predicted support parameters and the actual support parameters, wherein when the type of the support parameter is numerical type, the accuracy is a proportion of samples with the same predicted support parameters and actual support parameters, and when the type of the support parameter is factor type, the accuracy is a goodness-of-fit value; when the accuracy is within an accuracy threshold range, the trained support parameter prediction model is obtained; when the accuracy is not within the accuracy threshold range, the parameters of the initial support parameter prediction model are adjusted for continuous training until the accuracy is within the accuracy threshold range, and the trained support parameter prediction model is obtained; The construction module is further configured to, before obtaining the plurality of first sample data from the roadway support database, construct a plurality of random separation models by using a random forest algorithm, each support parameter corresponding to a random separation model; the obtaining module is further configured to obtain a plurality of second sample data, each second sample data including actual roadway parameters and actual support parameters; the input module is further configured to input the plurality of second sample data into the plurality of random separation models respectively to obtain a plurality of first rankings, each first ranking indicating an importance ranking of the plurality of roadway parameters with respect to a support parameter; the determining module is configured to determine target roadway parameters based on the plurality of first rankings; and the obtaining module is specifically configured to obtain the plurality of first sample data corresponding to the target roadway parameters from the roadway support database.

5. A support parameter determination apparatus characterized by comprising: The device comprises an obtaining module and an input module; The obtaining module is configured to obtain roadway parameters of a roadway to be supported; The input module is configured to input the roadway parameters of the roadway to be supported into the trained support parameter prediction model to obtain support parameters of the roadway to be supported; The trained support parameter prediction model is obtained by the method in any one of claims 1-2.

6. An electronic device, comprising: A processor, a memory, and a program or instructions stored on the memory and executable on the processor, the program or instructions being executed by the processor to implement the training method of the support parameter prediction model in any one of claims 1-2 or the support parameter determination method in claim 3.

Images