A mining stress field advanced prediction method, device and medium

Abstract

This invention relates to the field of mining technology and discloses a method, equipment, and medium for predicting mining-induced stress fields in advance. The method includes: constructing a three-dimensional numerical model based on mining area measurement data; constructing multi-condition stress field data for the mining area based on the three-dimensional numerical model; preprocessing the multi-condition stress field data to obtain model training data; training the pre-constructed neural network model using the model training data to obtain a mining-induced stress field prediction model; and using the mining-induced stress field prediction model to predict the advance support pressure at preset detection points to obtain mining-induced stress field prediction data. This invention can achieve a nonlinear mapping from mining conditions and geological parameters to the displacement field using a neural network, adaptively inverting the mining-induced stress field and predicting the distribution law of advance support pressure, thereby improving the efficiency and accuracy of mining-induced stress field prediction.

Description

Technical Field

[0001] This invention relates to the field of mining technology, specifically to a method, equipment, and medium for predicting mining-induced stress fields in advance. Background Technology

[0002] In coal mining, tunnel engineering, underground chamber construction, and other underground engineering fields, accurate prediction of mining-induced stress fields is a core technological prerequisite for ensuring project safety and efficient progress. The reliability of the prediction results directly determines the rationality of the working face support system design, the effectiveness of preventing dynamic disasters such as rock bursts, and the comprehensive benefits of resource extraction and engineering construction. Among these, the distribution pattern of advance support pressure, as the most critical stress concentration area in the mining-induced stress field, has a decisive impact on the stability control of the surrounding rock and the effectiveness of roadway support during the working face mining process.

[0003] Existing methods for predicting mining-induced stress fields mainly include numerical simulation, multiple regression analysis, and boundary load adjustment. However, linear regression methods struggle to accurately describe the nonlinear characteristics of mining-induced stress fields; numerical simulation methods rely on manual experience to adjust boundary conditions, resulting in low inversion efficiency and high subjectivity; and traditional neural networks are prone to getting trapped in local optima and have insufficient generalization ability. Therefore, current technologies cannot efficiently, accurately, and adaptively invert mining-induced stress fields and predict the distribution of advance support pressure, leading to poor efficiency and accuracy in predicting mining-induced stress fields. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to efficiently and accurately predict the stress field caused by mining.

[0005] The present invention solves the above-mentioned technical problems through the following technical means: This invention provides a method for predicting mining-induced stress fields in advance, characterized by comprising: Acquire mining field measurement point data of the mining area, and construct a three-dimensional numerical model of the mining area based on the mining field measurement point data; Based on the three-dimensional numerical model, multi-condition stress field data of the mining area are constructed, and the multi-condition stress field data are preprocessed to obtain model training data. The pre-constructed neural network model is trained using the model training data to obtain a mining stress field prediction model. The mining-induced stress field prediction model is used to predict the pre-support pressure of the preset test point data, thereby obtaining the mining-induced stress field prediction data of the test point data.

[0006] The present invention also provides a processing device, characterized in that it includes at least one processor and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor can execute the above-described method for predicting the ahead of the mining stress field by calling the program instructions.

[0007] The present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions that cause the computer to execute the above-described method for predicting the ahead of the mining stress field.

[0008] The advantages of this invention are: This invention constructs multi-condition stress field data for mining areas, and can comprehensively simulate complex mining-induced stress environments by using various load combinations such as uniform, triangular, trapezoidal, and shear displacement. At the same time, it achieves nonlinear mapping from mining conditions and geological parameters to displacement fields by optimizing neural networks based on genetic algorithms, avoiding manual intervention in traditional inversion methods, thereby effectively improving the efficiency and accuracy of advanced prediction of mining-induced stress fields. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of a method for predicting dynamic stress field in one embodiment of the present invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Reference Figure 1 The diagram shown is a flowchart illustrating a method for predicting mining-induced stress fields in advance, according to an embodiment of the present invention. In this embodiment, the method for predicting mining-induced stress fields in advance includes: S1. Obtain the mining area measurement point data, and construct a three-dimensional numerical model of the mining area based on the mining area measurement point data.

[0012] In this embodiment of the invention, the mine measurement point data in the mining area consists of the coordinates of the measurement points determined by the actual displacement position provided by a pre-set mine measurement, as well as important parameters obtained from the measurement.

[0013] Furthermore, the step of constructing a three-dimensional numerical model of the mining area based on the mining area measurement data includes: The coordinates of the measuring points and the corresponding measuring point parameters are determined based on the measuring point data from the mine. Based on the coordinates and parameters of the measuring points, the mine measuring point data is generalized and partitioned to obtain partitioned data. A three-dimensional numerical model of the mining area is constructed based on the partitioned data.

[0014] In detail, the area can be simulated and modeled using the modeling software FLAC3D, and relevant parameters can be assigned to the model. The simulation selects the area containing the coordinates of the measuring points. The measurement area of ​​the mining area is generalized and divided according to the mechanical properties of each rock layer and the interlayering pattern of lithology. Based on the inversion pattern after multiple model building, it is concluded that when building the model, the location of the measuring points must be ensured to be within the effective range of the model.

[0015] S2. Construct multi-condition stress field data for the mining area based on the three-dimensional numerical model, and perform data preprocessing on the multi-condition stress field data to obtain model training data.

[0016] Specifically, the geostress in deep wells in mining areas is complex and the underground space is almost completely enclosed. Therefore, by setting different boundary load conditions, a roof displacement field database containing various working conditions is constructed to simulate the roof deformation response under different mining and support conditions.

[0017] In this embodiment of the invention, constructing multi-condition stress field data of the mining area based on the three-dimensional numerical model includes: The three-dimensional numerical model is simulated using preset multi-condition data to obtain a condition simulation model; The stress is solved by the simulation model of the working conditions to obtain stress field data for multiple working conditions.

[0018] In detail, the preset multi-condition data sets various boundary stress forms for the boundary loads of the 3D numerical model; for example, 27 conditions can be set. Based on the distribution of measuring points, the conditions are reduced or increased in the x, y, and z directions. Using FLAC3D, uniform loads, triangular and trapezoidal loads along the x direction, and trapezoidal loads along the y direction are applied to the z-plane of the mining area model, respectively, along with normal stress. On the x-plane, uniform loads and trapezoidal loads along the y and z directions are applied, along with normal stress. On the y-plane, uniform loads and trapezoidal loads along the x and z directions are applied, along with normal stress. For the shearing structural motion in the xy, xz, and yz planes, shear stress is applied to each side with a shear boundary displacement of 600 mm. The applied displacement load is controlled by three forms of velocity boundary conditions: uniform, triangular, and trapezoidal, with a boundary velocity of 3e-4 and a running step of 2000 steps. This results in 27 conditions. Stress is solved using FLAC3D, ultimately yielding all data for all 27 conditions.

[0019] A three-dimensional numerical model of the mining area was established using numerical software such as FLAC3D. The three-dimensional model was established based on the actual geological conditions of the tunnels, and the rock strata mechanical parameters and boundary constraints were reasonably set to ensure that the area of ​​interest was within the effective range of the model.

[0020] Furthermore, the stress field data from the multi-condition simulation are extracted and combined with the measured data, and then normalized to obtain the model training data.

[0021] S3. Use the model training data to train the pre-constructed neural network model to obtain the mining stress field prediction model.

[0022] In this embodiment of the invention, the model training data can be divided into a training set and a test set proportionally. This is applied to MATLAB for building and training pre-built neural network models (BP neural network models). The dataset is divided into training and test samples in a 4:1 ratio. The pre-built neural network model has 7 hidden layer nodes, a population size of 5, and a learning rate of 0.01.

[0023] Specifically, the step of training a pre-constructed neural network model using the model training data to obtain a mining-induced stress field prediction model includes: The neural network model is used to predict the initial mining stress field based on the model training data, and the initial prediction result of the mining stress field is obtained. The mean square error of the neural network model is calculated based on the initial prediction results of the mining stress field. Based on the mean square error, the parameters of the neural network model are optimized using a genetic algorithm to obtain a mining stress field prediction model.

[0024] Specifically, the topology of the BP neural network model consists of three layers: the input layer, the hidden layer, and the output layer.

[0025] Threshold of hidden layer: ; Weights from nodes in the hidden layer to nodes in the output layer:

[0026] The selection process involves using geometric programming to sort and select the best individual components.

[0027] The crossover process uses an arithmetic mean.

[0028] The mutation process employs non-uniform mutation.

[0029] The activation function for the hidden layer neurons is set to the tansig function, and its expression is:

[0030] The activation function of the output layer is set to purelin, and its expression is:

[0031] Mean square error function of BP neural network model

[0032] In this embodiment of the invention, the GA-BP neural network model is trained using a training set, with mean squared error as the loss function, until the loss function is less than a preset threshold, thus obtaining the mining stress field prediction model.

[0033] S4. Using the mining-induced stress field prediction model, predict the advanced support pressure of the preset test point data to obtain the mining-induced stress field prediction data of the test point data.

[0034] In this embodiment of the invention, after training is completed, the data of the points to be predicted are imported into MATLAB, which enables the prediction of mining stress field and further the prediction of advance support pressure.

[0035] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting mining-induced stress fields in advance, characterized in that, include: Acquire mining field measurement point data of the mining area, and construct a three-dimensional numerical model of the mining area based on the mining field measurement point data; Based on the three-dimensional numerical model, multi-condition stress field data of the mining area are constructed, and the multi-condition stress field data are preprocessed to obtain model training data. The pre-constructed neural network model is trained using the model training data to obtain a mining stress field prediction model. The mining-induced stress field prediction model is used to predict the pre-support pressure of the preset test point data, thereby obtaining the mining-induced stress field prediction data of the test point data.

2. The method for predicting mining-induced stress fields as described in claim 1, characterized in that, The construction of a three-dimensional numerical model of the mining area based on the mining area measurement data includes: The coordinates of the measuring points and the corresponding measuring point parameters are determined based on the measuring point data from the mine. Based on the coordinates and parameters of the measuring points, the mine measuring point data is generalized and partitioned to obtain partitioned data. A three-dimensional numerical model of the mining area is constructed based on the partitioned data.

3. The method for predicting mining-induced stress fields as described in claim 1, characterized in that, The construction of multi-condition stress field data for the mining area based on the three-dimensional numerical model includes: The three-dimensional numerical model is simulated using preset multi-condition data to obtain a condition simulation model; The stress is solved by the simulation model of the working conditions to obtain stress field data for multiple working conditions.

4. The method for predicting mining-induced stress fields as described in claim 1, characterized in that, The process of preprocessing the multi-condition stress field data to obtain model training data includes: The stress data under the multiple working conditions are normalized to obtain model training data.

5. The method for predicting mining-induced stress fields as described in claim 1, characterized in that, The step of training a pre-constructed neural network model using the model training data to obtain a mining-induced stress field prediction model includes: The neural network model is used to predict the initial mining stress field based on the model training data, and the initial prediction result of the mining stress field is obtained. The mean square error of the neural network model is calculated based on the initial prediction results of the mining stress field. Based on the mean square error, the parameters of the neural network model are optimized using a genetic algorithm to obtain a mining stress field prediction model.

6. A processing apparatus, characterized in that, The method includes at least one processor and at least one memory communicatively connected to the processor, wherein the memory stores program instructions executable by the processor, and the processor can execute the method as described in any one of claims 1 to 5 by invoking the program instructions.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause the computer to perform the method as described in any one of claims 1 to 5.

Images