Method for intelligently determining size of section coal pillar based on while-drilling parameters

Abstract

The invention provides a method for intelligently determining the size of a section coal pillar based on parameters while drilling, and relates to the technical field of coal mining, which comprises the following steps of: carrying out measurement while drilling experiment and recording data under different confining pressures and different elastic-plastic state conditions; a stress meter is installed on site for stress monitoring, and stress data in the coal pillar before and after working face stoping are obtained; preprocessing the obtained experimental data and stress test data, and dividing the data into a training set and a test set; establishing a section coal pillar size determination intelligent model, and training the model by using the training set data; testing the model by using test set data, namely adjusting parameters; a drilling machine is used for drilling the coal pillar, and the intelligent model meeting the requirement is used for determining the size of the section coal pillar; and while-drilling parameters and coal pillar data fed back in the using process are collected, and the intelligent model and the discrimination model are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mining, and in particular to a method for intelligently determining the size of a section coal pillar based on drilling parameters. Background Art

[0002] Leaving coal pillars to protect section tunnels is the main tunnel protection method adopted by many coal mines. The reasonable size of the section coal pillars plays an extremely important role in the stability of the coal pillars themselves and the tunnels, as well as the improvement of coal recovery rate. If the size of the section coal pillars is too large, it will lead to waste of coal resources, while if the size of the section coal pillars is too small, it may cause damage to the coal pillars themselves and adjacent tunnels, posing a threat to the safety of workers, causing damage to equipment in the tunnels and working faces, and causing unnecessary losses to the coal mine economy. Therefore, the rationality of determining the size of the section coal pillars is very important.

[0003] Different coal mines have different geological conditions, and the surrounding rock properties of different working faces in the same coal mine also vary. Therefore, the size of the segmental coal pillars required for working faces under different geological conditions and stress conditions should also be different. Currently, there are many methods for determining the size of segmental coal pillars, but the main methods can be roughly summarized into four: First, through on-site measurement, the stress distribution state in the coal body under mining action is measured, and the specific mining roadway layout and reasonable segmental coal pillar size are determined in combination with relevant mine pressure theory; second, using elastic-plastic theory, the formula for the size of the segmental coal pillar when the segmental coal pillar remains in a stable state is derived, and a reasonable calculation formula for the segmental coal pillar size is obtained. Different coefficients are assigned according to different geological conditions to determine the reasonable segmental coal pillar size; third, using computer software for numerical simulation, relevant geological information and parameters are imported into the simulation software for calculation simulation, and the reasonable size of the segmental coal pillar is determined based on the simulation results; fourth, using traditional on-site production experience, the segmental coal pillar size under similar on-site conditions is selected as a reference for the segmental coal pillar.

[0004] Currently commonly used methods for determining segmental coal pillar size all have certain limitations. For example, on-site measurement is complex, consuming significant manpower and resources and increasing coal mine production costs. Calculation formulas often make idealized assumptions, resulting in low accuracy in determining coal pillar size. Computer simulation model building and parameter selection are complex, and results rely on personal interpretation. Field production experience is often limited by regional constraints, often leading to conservative designs when determining coal pillar size, resulting in a waste of coal resources. Therefore, currently commonly used methods for determining segmental coal pillar size struggle to achieve both convenience and accuracy.

[0005] In view of this, the development of a method for determining the size of the segmented coal pillar that combines convenience and accuracy is of great significance to the current establishment of coal pillars in tunnels. As a cutting-edge technical means, measurement while drilling technology can capture and monitor a series of key drilling parameters such as drilling speed, rotation speed, torque and drilling pressure in real time during the drilling operation, and then accurately analyze and predict the lithology and stress state based on these parameters. After the working face is mined, the stress state in the coal pillar will change significantly. When the drill rig passes through the coal pillar, these stress changes will be indirectly reflected through the drilling parameters generated by drilling, and the stress changes in the coal pillar can determine the size of the segmented coal pillar that needs to be left. Based on this principle, measurement while drilling technology can play an important role in determining the size of the segmented coal pillar. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for intelligently determining the size of a section coal pillar based on drilling parameters. The method first determines the relationship between the drilling parameters and the stress within the coal pillar through indoor experiments, and then uses real-time stress monitoring on site to judge the stress changes before and after the coal pillar is mined. Combined with the theoretical method for determining the size of the coal pillar, an artificial intelligence algorithm is used to correlate the three, and finally the coal pillar size is intelligently determined.

[0007] A method for intelligently determining the size of a section coal pillar based on drilling parameters, comprising: Step 1: Conduct measurement while drilling experiments under different confining pressure conditions and record the data; Step 2: Install a stress gauge on site to monitor stress and obtain stress test data inside the coal pillar before and after mining at the working face; Step 3: Preprocess the acquired experimental data and stress test data and divide the data into training set and test set; Step 4: Establish a coal pillar size prediction model for the section coal pillar, and use the training set data to train the coal pillar size prediction model; Step 5: Use the test set data to test the trained intelligent prediction model for the section coal pillar size, adjust the parameters, and obtain the intelligent determination model for the section coal pillar size; Step 6: Use a drilling rig to drill the section coal pillar, and use the section coal pillar size intelligent determination model to predict the section coal pillar size; Step 7: Collect the drilling parameters and coal pillar size data fed back during use, and optimize the intelligent determination model of the section coal pillar size.

[0008] In a preferred embodiment, in step 4, the process of constructing the intelligent model for determining the size of the coal pillar in the section is as follows: S41. Using the DQN reinforcement learning model as the base model, the drilling parameters obtained from the drilling measurement experiment are used as input parameters, and the stress state of the specimen is used as the output parameter. Model A of the relationship between the drill bit pressure, torque, and the stress state of the specimen during drilling is constructed, and model A is trained. S42, performing data fitting on the stress data x of the coal pillar before mining and the stress data y of the coal pillar after mining to obtain a fitting formula; S43, the stress data of the drilling test specimen is used as the post-mining stress data y1 and is substituted into the fitting formula to obtain the corresponding pre-mining stress data x1 under the state of the specimen; S44, bringing the pre-mining stress data x1 into model A to obtain theoretical pre-mining while-drilling parameters under laboratory parameter conditions; S45. Using the DQN reinforcement learning model as the base model again, the theoretical pre-mining while-drilling parameters obtained in step S44 are used as input parameters, and the post-mining stress is used as the output parameter. A relationship model B between the pre-mining while-drilling parameters and the post-mining stress is constructed, and model B is trained. S46. Record the stress value and drilling length d0 at the first stabilization of stress during the drilling experiment, and the stress value and drilling length d1 at the second stabilization of stress during the drilling experiment. The calculation formula for the coal pillar size M is: ; in, ; The formula Built into model B, when model B outputs, its output stress y1 command is mapped to the output drilling length d at that stress, and input into the calculation formula of the coal pillar size M built into model B, and the final output is the section coal pillar size M that needs to be left; The training process in S41 is repeated again to train the intelligent prediction model for the section coal pillar size, and after completion, the trained intelligent prediction model for the section coal pillar size is output.

[0009] In a preferred embodiment, step S41 includes: S411. Determine the state space : ; in, is the speed at time t; is the torque at time t; is the width of the coal pillar; is the number of damaged specimens; S412: Determine the action space; S413. Design reward function ; ; in, is the original data; For the corrected data; S414, -greedy action choose: ; in, For action; is the exploration rate; S415, calculate the target Q value: ; Where: is the target Q value; For immediate rewards; is the current state; is the discount factor; The state of the next moment; Output of the target network; is the target network parameter; To maximize operations; All possible actions for the next state.

[0010] S416, gradient descent update: ; in, is a parameter; is the learning rate; is the gradient operator; is the loss function.

[0011] In a preferred embodiment, the step five includes: S51, importing the test set data into the trained section coal pillar size intelligent prediction model; S52, checking whether the output parameters processed by the intelligent prediction model for the section coal pillar size are consistent with the theoretical output parameters, and determining the errors; S53. If the error is too large and cannot meet the accuracy requirement, the parameters of the intelligent prediction model for the section coal pillar size are adjusted, and after completion, the test set data is used again for testing; S54. If the error meets the accuracy requirement, the intelligent determination model of the section coal pillar size is output.

[0012] In a preferred embodiment, the step six includes: S61, drilling is carried out at 50m ahead of the working face, and the first drilling hole is located at the side of the coal pillar 50m away from the opening of the cutting hole; S62. Drilling should be done at three points: upper, middle, and lower. The final result should be the average. S63, drilling interval is 50m, use drilling rig to drill the section coal pillar; S64. Record the coal seam mining depth h, mining thickness m, and the width of the coal pillar crushing zone a0 before mining, the width of the plastic zone a1, and the drilling length b0 when the stress is initially stable.

[0013] In a preferred embodiment, the step 1 includes: S11. Prepare specimens and determine lithologic parameters of the specimens; S12, grouping the specimens, applying different confining pressures to each group of specimens, and establishing an internal stress state; S13. Conduct measurement while drilling experiments on different test specimen groups and record the measurement while drilling parameters.

[0014] In a preferred embodiment, the second step includes: S21, determine the location of the measuring station and measuring points; S22. Install the stress device and debug it; S23. Collect coal pillar stress test data, compare and analyze the changes in coal pillar stress state at different depths and the relationship between the stress states of the coal pillar before and after mining.

[0015] In a preferred embodiment, the step seven includes: S71. Collecting drilling parameters generated during drilling and pre-mining stress collected by a stress device; S72. Perform data cleaning on drilling parameters and pre-mining stress; S73, importing the measured pre-mining while-drilling parameters and pre-mining stress into the intelligent prediction model for the section coal pillar size to obtain the predicted section coal pillar size; S74. Set up coal pillars according to the predicted size of the coal pillars in the section. After setting up, observe the deformation of the coal pillars and the surrounding roadways. Modify the intelligent prediction model for the coal pillar size in the section according to the deformation. S75. Use the revised intelligent prediction model for the size of the coal pillar to predict the size of the coal pillar, and repeat steps S74 and S75 to improve the accuracy.

[0016] In a preferred embodiment, a correction parameter is inserted into the intelligent prediction model of the section coal pillar size. The output results of the intelligent prediction model of the section coal pillar size are corrected. When the deformation of the coal pillar and its surrounding roadways is large, the correction parameters are adjusted. >1, when the deformation of the coal pillar and its surrounding tunnels is small or no deformation, let the correction parameter .

[0017] The present invention has the following beneficial technical effects: Compared with the traditional method of determining the size of segmented coal pillars, the method of determining the size of segmented coal pillars based on drilling parameters has the following advantages: the size of the segmented coal pillars to be retained can be obtained by drilling the side walls of the coal pillars before the working face is mined, and the on-site operation is relatively convenient; the drilling parameters are analyzed and processed by an intelligent model, which avoids the influence of personal subjectivity and enhances the accuracy of determining the size of segmented coal pillars; the drilling rig can correct the obtained size of the segmented coal pillars as the working face continues to advance, which can further improve the accuracy of determining the size of the segmented coal pillars. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a structural diagram of the measurement while drilling test machine; Figure 2 Schematic diagram of the specimen shape and size; Figure 3 Schematic diagram of the preloaded elastic-plastic state and stress distribution in the measurement while drilling experiment; Figure 4 This is a schematic diagram of the layout of on-site stress monitoring stations and measuring points; Figure 5 This is a schematic diagram of on-site drilling rig drilling and coal pillar stress zoning; Figure 6 This is a flow chart of a method for intelligently determining the size of a section coal pillar based on drilling parameters according to the present invention; Figure 7 The figure is a schematic diagram of a specific implementation process of a method for intelligently determining the size of a section coal pillar based on drilling parameters according to the present invention.

[0019] Figure numerals: 1-mdw borehole; 2-segmental coal pillar; 3-non-coal pillar sidewall; 4-segmental roadway; 5-roadway roof; 6-stress curve. DETAILED DESCRIPTION

[0020] To make the purpose, technical solutions, and advantages of this embodiment more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, other embodiments obtained by ordinary technicians in this field without making creative efforts are all within the scope of protection of this application.

[0021] In the drawings of the specific embodiments of the present invention, in order to better and more clearly describe the working principles of the various components in the system, the connection relationship of the various parts in the device is shown, which only clearly distinguishes the relative position relationship between the various components, and does not constitute a limitation on the signal transmission direction, connection sequence and structural position, size and shape of each part within the component or structure.

[0022] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places in this specification does not necessarily refer to the same embodiment, nor does it necessarily refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0023] The present invention will be further described below by way of examples and in conjunction with the accompanying drawings, but is not limited thereto. Based on the embodiments of the present invention: This embodiment provides a method for intelligently determining the size of a coal pillar segment based on drilling parameters. The specific steps are as follows: Step 1: Conduct measurement while drilling experiments under different confining pressure conditions and record relevant data.

[0024] The drilling experiment steps of the measurement while drilling under different confining pressures are as follows: S11. Prepare specimens and determine the lithologic parameters of the specimens.

[0025] like Figure 2 As shown, the specimen preparation materials are cement, river sand and water, and the specimen shape is a cube specimen with a size of 300mm×300mm×300mm; A portion of the cast specimens was cored and subjected to uniaxial compression tests to obtain the mechanical properties of the specimens, including uniaxial compressive strength and elastic modulus.

[0026] S12. Group the specimens and apply different confining pressures to each group of specimens to establish the internal stress state.

[0027] The specimens were divided into 5 groups, 3 in each group, to simulate the coal pillar morphology. Each group of specimens was numbered. The specific specimen combination is shown in Figure 3 , the thickness of the simulated coal seam mining is n = 300 mm, the width of the simulated section coal pillar is 3n = 900 mm, and the three specimens in the same group are placed closely horizontally to reduce the influence of the gaps between the specimens on the drilling experiment; Apply different confining pressures to each group of specimens for pre-compression. On the one hand, the specimens can be pressed tightly so that they will not move due to drilling. On the other hand, different stress states and elastic-plastic states can be presented inside to simulate the stress state of the coal pillar under different burial depths. The simulated coal pillar state after pressurization is shown in Figure 2. Figure 3 , where the width of the coal pillar's unilateral crushing zone is c0, the width of the unilateral plastic zone is c1, and the stress state of the coal pillar is as follows: Figure 3 The stress curve in 6 is shown.

[0028] S13. Conduct measurement while drilling experiments on different test specimen groups and record the measurement while drilling parameters.

[0029] use Figure 1The measurement-while-drilling (MWD) drilling rig shown here conducts MWD drilling experiments on different specimen groups, recording MWD parameters such as drilling speed, drilling length, drilling pressure, and torque, as well as specimen-related parameters. Preferably, the drilling experiment in step one is conducted on the basis that the coal pillar is already in a state of being laid. That is, the MWD parameters obtained during drilling are determined by the stress state of the coal pillar after mining. This obviously does not meet the requirement of predicting the coal pillar width before mining. To solve this problem, it is necessary to study the stress state of the coal pillar before and after mining on site and construct a stress relationship model of the coal pillar before and after mining. The specific method is shown in step two.

[0030] Preferably, Figure 1 The figure shows a schematic diagram of a measurement while drilling (MWD) rig. Its main structure includes a drill pipe, a propulsion device, and a pressure device. The sensing device attached to the drill pipe can obtain real-time MWD parameters during drilling. The propulsion device can control the propulsion force, propulsion speed, and rotation speed of the drill pipe. The pressure device is equipped with a non-uniformly distributed loading cylinder, which can load different parts of the specimen with different forces to achieve pre-compression at different positions of the specimen, so that it reaches different stresses and different elastic-plastic states.

[0031] Step 2: Install a stress gauge on site to monitor stress and obtain stress test data inside the coal pillar before and after mining at the working face.

[0032] The specific steps of on-site stress monitoring are as follows: S21, determine the location of the measuring station and measuring points; like Figure 4 As shown, a stress gauge is installed by drilling holes in the side of the coal pillar 50m ahead of the working face. The stress gauge is divided into three measuring stations with a spacing of 50m. Each measuring station has 10 measuring points. The depth difference of each measuring point is set to 5m. The depth difference between measuring point 9 and measuring point 10 is 10m, and the spacing between measuring points is 2m.

[0033] S22. Install the stress device and debug it; Install the strain gauge at the selected measuring point and debug it using relevant software after installation.

[0034] S23. Collect coal pillar stress test data, compare and analyze the changes in coal pillar stress state at different depths and the relationship between the stress states of the coal pillar before and after mining.

[0035] Stress data at different depths of the coal pillar are collected before and after mining until all stress gauges at the measuring points enter the goaf for more than 100 m. The collected data are imported into relevant software to compare and analyze the changes in the stress state of the coal pillar at different depths and the relationship between the stress state of the coal pillar before and after mining.

[0036] Step 3: Preprocess the acquired experimental data and stress test data, and divide the data into training set and test set.

[0037] The specific steps of data preprocessing are as follows: S31. Data cleaning.

[0038] Delete duplicate data, fill missing data, delete data with too much missing data, and use the Z-score statistical method to preliminarily repair abnormal data.

[0039] S32. Calculate the mean : ;

[0040] Where: is the total number of data; For the Initial data value.

[0041] S33. Calculate standard deviation : ;

[0042] Where: is the data mean; is the total number of data; For the Initial data value.

[0043] S34. Calculate the Z-score of each data: ;

[0044] Where: is a standardized value; is the original data value.

[0045] S35. Data segmentation.

[0046] The data is divided into training data set and test data set, of which the training data set accounts for 70% and the test data set accounts for 30%.

[0047] S36. Add eigenvalues.

[0048] Adding new features such as product and difference makes the representative data in the data set stand out, making it easier to perform image fitting on the data later.

[0049] S37. Data enhancement.

[0050] Randomly insert and delete data to increase data diversity and prevent overfitting.

[0051] S38. Data batch processing.

[0052] The data is divided into groups of 64 to facilitate the input of the data into the model later.

[0053] Step 4: Establish a segment coal pillar size to determine the segment coal pillar size intelligent prediction model, and use the training set data to train the segment coal pillar size intelligent prediction model.

[0054] The process of constructing the intelligent model for determining the size of the section coal pillar is as follows: S41. Using the DQN reinforcement learning model as the base model, the drilling parameters obtained from the drilling measurement experiment are used as input parameters, and the specimen stress state is imported into the DQN reinforcement learning model as the output parameter. A model of the relationship between the drilling parameters such as drill bit pressure and torque and the specimen stress state during drilling is constructed and named Model A. The model is then trained. The specific process is as follows.

[0055] S411. Determine the state space : ; in, is the speed at time t; is the torque at time t; is the width of the coal pillar; is the number of damaged specimens; S412. Determine the action space : ; S413. Design reward function : ;

[0056] Where: is the original data; For the corrected data; S414, -greedy action choose ;

[0057] Where, is the exploration rate; S415, calculate Target Q-value ;

[0058] Where: is the target Q value; For immediate rewards; is the current state; is the discount factor; The state of the next moment; Output of the target network; is the target network parameter; To maximize operations; All possible actions for the next state.

[0059] S416, gradient descent update: ;

[0060] in, are the parameters of model A; is the learning rate; is the gradient operator; is the loss function.

[0061] Preferably, the learning rate The value is 0.0005 to control the update step size of the parameters, the discount factor The value is set to 0.95 to balance current and future rewards, and the exploration rate decays The value is 0.5, the purpose is to gradually shift the model from exploration to utilization, and the network update frequency C The value is set to synchronize every 500 steps to train the model for stability.

[0062] S42, importing the coal pillar stress data obtained from monitoring before and after mining into relevant software, using the software to perform data fitting with the pre-mining stress data as x and the post-mining stress data as y to obtain a fitting formula; S43. Substitute the stress data of the drilling test specimen into the fitting formula as the post-mining stress data y1 to solve the corresponding pre-mining stress data x1 under the state of the specimen.

[0063] S44. Substituting the pre-mining stress data x1 into model A can obtain the theoretical pre-mining while-drilling parameters under laboratory-related parameter conditions.

[0064] S45. Again, the DQN reinforcement learning model is used as the basic model. The theoretical pre-mining while-drilling parameters are used as input parameters and the post-mining stress is used as the output parameter. A relationship model between the pre-mining while-drilling parameters and the post-mining stress is constructed and named Model B. The specific training process of Model B is consistent with the training process of Model A in S41.

[0065] S46, record the stress value and drilling length d0 at the first stable stress and the stress value and drilling length d1 at the second stable stress in the drilling experiment, and calculate the section coal pillar size according to the formula in , we can get the size M of the section coal pillar that needs to be left. n refers to Figure 3 The height of the specimen in the simulation is the height of the simulated coal seam mining.

[0066] The formula Built into model B, when model B outputs its output stress y1, it maps the command to output the drilling length d at that stress, and inputs this into the built-in coal pillar size calculation formula, so that the final output of model B is the required segment coal pillar size M. This model is the segment coal pillar size intelligent prediction model. The training process in S41 is repeated again to train the segment coal pillar size intelligent prediction model. After completion, the trained segment coal pillar size intelligent prediction model is output.

[0067] Step 5: Use the test set data to test the trained intelligent prediction model for the section coal pillar size, adjust the parameters, and obtain the intelligent determination model for the section coal pillar size. The steps are as follows: S51. Import the test set data into the trained intelligent prediction model for the size of the coal pillar in the section.

[0068] S52. Check whether the output parameters obtained after processing by the intelligent prediction model of the section coal pillar size are consistent with the theoretical output parameters and determine their errors.

[0069] S53. If the error is too large and cannot meet the accuracy requirement, the parameters of the intelligent prediction model for the section coal pillar size are adjusted. After completion, the test set data is used again for testing.

[0070] S54. If the error meets the accuracy requirement, the intelligent determination model of the section coal pillar size is output.

[0071] Step 6: Use a drilling rig to drill into the section coal pillar, and use the section coal pillar size intelligent determination model to predict the section coal pillar size.

[0072] The steps for section coal pillar drilling and its size determination are as follows: S61. Determine the drilling position Drilling is carried out 50m ahead of the working face, that is, the first drill hole is located at the side of the coal pillar 50m away from the cutting eye.

[0073] S62, drilling point determination like Figure 5 As shown in the measurement while drilling borehole 1, in order to ensure the reliability of the numerical prediction of the section coal pillar size, the drilling position should be divided into three points: upper, middle and lower, and the final result is the average.

[0074] S63. Use a drilling rig to drill into the section coal pillar.

[0075] The drilling interval is 50m, which facilitates real-time adjustment of the coal pillar size in each section according to different geological conditions.

[0076] S64, Data Collection like Figure 5, record relevant geological parameters such as coal seam mining depth h, mining thickness m, and drilling data such as coal pillar crushing zone width a0, plastic zone width a1, and drilling length b0 when stress is initially stable before mining, to facilitate further optimization of subsequent models.

[0077] Step 7: Collect the drilling parameters and coal pillar size data fed back during use, and optimize the intelligent determination model of the section coal pillar size.

[0078] The pre-mining while-drilling parameters and pre-mining stress of the coal pillar in step 4 S43 and S44 are all obtained by calculation. Therefore, the intelligent determination model of the segmented coal pillar size output by the laboratory while-drilling drilling experiment will have some errors when predicting the segmented coal pillar size. Therefore, the intelligent determination model of the segmented coal pillar size can be further optimized. The specific optimization process is as follows.

[0079] S71. Collect the drilling parameters generated during the drilling process and the pre-mining stress collected by the stress device.

[0080] S72. Perform data cleaning on drilling parameters and pre-mining stress.

[0081] S73. Import the measured pre-mining while-drilling parameters and pre-mining stresses into the tested section coal pillar size intelligent prediction model to obtain the predicted section coal pillar size.

[0082] S74. Set up the coal pillar according to the predicted size. After setting up, observe the deformation of the coal pillar and the surrounding roadway. According to the deformation, modify the intelligent prediction model of the coal pillar size in the section. The modification method is to insert the correction parameter into the intelligent prediction model of the coal pillar size in the section. Correct the model output results. When the deformation of the coal pillar and its surrounding tunnels is large, correct the parameters. The value should be greater than 1. When the deformation of the coal pillar and its surrounding tunnels is small or no deformation, the correction parameter is taken. .

[0083] S75. Use the revised intelligent prediction model for the size of the coal pillar to predict the size of the coal pillar, and repeat steps S74 and S75 to further improve the accuracy of the intelligent prediction model for the size of the coal pillar.

[0084] Preferably, the model is further optimized using geological parameters and drilling data obtained from on-site construction in order to expand the model database so that it has the ability to cope with various geological conditions and disturbances, enhance the universality of the model and further improve the accuracy of the model.

[0085] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above. Persons skilled in the art may make various changes or modifications within the scope of the claims without affecting the essence of the present invention. The embodiments of this application and the features within these embodiments may be combined in any manner, unless there is a conflict.

Claims

1. A method for intelligently determining the size of a coal pillar segment based on drilling parameters, characterized in that: include: Step 1: Conduct measurement while drilling experiments under different confining pressure conditions and record the data; Step 2: Install a stress gauge on site to monitor stress and obtain stress test data inside the coal pillar before and after mining at the working face; Step 3: Preprocess the acquired experimental data and stress test data and divide the data into training set and test set; Step 4: Establish a coal pillar size prediction model for the section coal pillar, and use the training set data to train the coal pillar size prediction model; Step 5: Use the test set data to test the trained intelligent prediction model for the section coal pillar size, adjust the parameters, and obtain the intelligent determination model for the section coal pillar size; Step 6: Use a drilling rig to drill the section coal pillar, and use the section coal pillar size intelligent determination model to predict the section coal pillar size; Step 7: Collect the drilling parameters and coal pillar size data fed back during use, and optimize the intelligent determination model of the section coal pillar size.

2. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: In step 4, the process of constructing the intelligent model for determining the size of the coal pillar in the section is as follows: S41. Using the DQN reinforcement learning model as the base model, the drilling parameters obtained from the drilling measurement experiment are used as input parameters, and the stress state of the specimen is used as the output parameter. Model A of the relationship between the drill bit pressure, torque, and the stress state of the specimen during drilling is constructed, and model A is trained. S42, performing data fitting on the stress data x of the coal pillar before mining and the stress data y of the coal pillar after mining to obtain a fitting formula; S43, the stress data of the drilling test specimen is used as the post-mining stress data y1 and is substituted into the fitting formula to obtain the corresponding pre-mining stress data x1 under the state of the specimen; S44, bringing the pre-mining stress data x1 into model A to obtain theoretical pre-mining while-drilling parameters under laboratory parameter conditions; S45. Using the DQN reinforcement learning model as the base model again, the theoretical pre-mining while-drilling parameters obtained in step S44 are used as input parameters, and the post-mining stress is used as the output parameter. A relationship model B between the pre-mining while-drilling parameters and the post-mining stress is constructed, and model B is trained. S46. Record the stress value and drilling length d0 at the first stabilization of stress during the drilling experiment, and the stress value and drilling length d1 at the second stabilization of stress during the drilling experiment. The calculation formula for the coal pillar size M is: ； in, ; n is the height of simulated coal seam mining; The formula Built into model B, when model B outputs, its output stress y1 command is mapped to the output drilling length d at that stress, and input into the calculation formula of the coal pillar size M built into model B, and the final output is the section coal pillar size M that needs to be left; The training process in S41 is repeated again to train the intelligent prediction model for the section coal pillar size, and after completion, the trained intelligent prediction model for the section coal pillar size is output.

3. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 2, characterized in that: The step S41 includes: S411. Determine the state space : ; in, is the speed at time t; is the torque at time t; is the width of the coal pillar; is the number of damaged specimens; S412: Determine the action space; S413. Design reward function ; ； in, is the original data; For the corrected data; S414, -greedy action choose: ； in, For action; is the exploration rate; S415, calculate the target Q value: ； Where: is the target Q value; For immediate rewards; is the current state; is the discount factor; The state of the next moment; Output of the target network; is the target network parameter; To maximize operations; All possible actions for the next state; S416, gradient descent update: ； in, is a parameter; is the learning rate; is the gradient operator; is the loss function.

4. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: The step five includes: S51, importing the test set data into the trained section coal pillar size intelligent prediction model; S52, checking whether the output parameters processed by the intelligent prediction model for the section coal pillar size are consistent with the theoretical output parameters, and determining the errors; S53. If the error is too large and cannot meet the accuracy requirement, the parameters of the intelligent prediction model for the section coal pillar size are adjusted, and after completion, the test set data is used again for testing; S54. If the error meets the accuracy requirement, the intelligent determination model of the section coal pillar size is output.

5. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: The step six comprises: S61, drilling is carried out at 50m ahead of the working face, and the first drilling hole is located at the side of the coal pillar 50m away from the opening of the cutting hole; S62. Drilling should be done at three points: upper, middle, and lower. The final result should be the average. S63, drilling interval is 50m, use drilling rig to drill the section coal pillar; S64. Record the coal seam mining depth h, mining thickness m, and the width of the coal pillar crushing zone a0 before mining, the width of the plastic zone a1, and the drilling length b0 when the stress is initially stable.

6. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: The step one comprises: S11. Prepare specimens and determine lithologic parameters of the specimens; S12, grouping the specimens, applying different confining pressures to each group of specimens, and establishing an internal stress state; S13. Conduct measurement while drilling experiments on different test specimen groups and record the measurement while drilling parameters.

7. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: The second step includes: S21, determine the location of the measuring station and measuring points; S22. Install the stress device and debug it; S23. Collect coal pillar stress test data, compare and analyze the changes in coal pillar stress state at different depths and the relationship between the stress states of the coal pillar before and after mining.

8. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 1, characterized in that: The step seven comprises: S71. Collecting drilling parameters generated during drilling and pre-mining stress collected by a stress device; S72. Perform data cleaning on drilling parameters and pre-mining stress; S73, importing the measured pre-mining while-drilling parameters and pre-mining stress into the intelligent prediction model for the section coal pillar size to obtain the predicted section coal pillar size; S74. Set up coal pillars according to the predicted size of the coal pillars in the section. After setting up, observe the deformation of the coal pillars and the surrounding roadways. Modify the intelligent prediction model for the coal pillar size in the section according to the deformation. S75. Use the revised intelligent prediction model for the size of the coal pillar to predict the size of the coal pillar, and repeat steps S74 and S75 to improve the accuracy.

9. The method for intelligently determining the size of a coal pillar segment based on drilling parameters according to claim 8, characterized in that: Inserting correction parameters into the intelligent prediction model of section coal pillar size The output results of the intelligent prediction model of the section coal pillar size are corrected. When the deformation of the coal pillar and its surrounding roadways is large, the correction parameters are adjusted. >1, when the deformation of the coal pillar and its surrounding tunnels is small or no deformation, let the correction parameter .

Images