Method for determining influence range of advance bearing pressure of working face based on drilling cutting amount

Abstract

The invention discloses a method for determining the influence range of advance bearing pressure of a working face based on the drilling cutting quantity, and the method comprises the steps: drilling a plurality of drill holes in a coal body in front of the working face according to a preset interval, and obtaining the drilling cutting quantity data of each drill hole in the drilling process; the drilling cutting quantity data of each drill hole and the corresponding stress meter data are sent to a pre-trained deep learning model, so that the deep learning model predicts the current advanced bearing pressure influence range of the working face according to the drilling cutting quantity data; and predicting the influence range of the advance bearing pressure of the working face according to the deep learning model to optimize the supporting design of the working face, and supporting the working face according to the optimized supporting design. According to the method, the monitoring accuracy can be improved, the supporting design is optimized, the stability and reliability of a supporting structure are improved, waste of supporting materials is reduced, the production cost is reduced, potential safety hazards are reduced, the safety of a working face is improved, and the production efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of rock burst prevention, control, monitoring and early warning, and in particular to a method for determining the influence range of the advance support pressure of a working face based on the amount of drill cuttings. Background Art

[0002] During coal mining, advanced abutment pressure refers to the stress concentration in the coal mass ahead of the working face caused by mining activities. This stress concentration directly affects the stability of the coal mass, and thus the support design and safe production of the working face. Traditional support design methods are often based on empirical formulas and field observations. These methods lack accurate prediction of the impact range of advanced abutment pressure, leading to inappropriate support design and negatively impacting working face safety and production efficiency.

[0003] Traditional drill cuttings monitoring methods rely on manual timed measurements or simple sensor monitoring, which can lead to data lag and insufficient accuracy. Manual coal dust collection is inefficient, reducing the reliability of monitoring results. Furthermore, real-time monitoring is not possible, hindering timely reflection of changes in coal stress. Furthermore, traditional support design methods, due to inaccurate predictions, can increase safety hazards at the working face, reduce production efficiency, and impose high labor intensity on operators. Summary of the Invention

[0004] The present invention aims to solve, at least to a certain extent, one of the technical problems in the related art. To this end, the first purpose of the present invention is to propose a method for determining the influence range of the advance support pressure of the working face based on the amount of drill cuttings, accurately predict the influence range of the advance support pressure through a deep learning model, provide a scientific basis for support design, ensure the rationality and safety of the support design, thereby improving monitoring accuracy, optimizing support design, improving the stability and reliability of the support structure, reducing the waste of support materials, reducing production costs, reducing safety hazards, improving the safety of the working face, and improving production efficiency.

[0005] To achieve the above-mentioned purpose, an embodiment of the first aspect of the present invention proposes a method for determining the influence range of the advance support pressure of the working face based on the amount of drill cuttings, the method comprising: drilling a plurality of boreholes in the coal body in front of the working face at a preset interval, and obtaining the drill cuttings amount data of each borehole during the drilling process; sending the drill cuttings amount data of each borehole and the corresponding stress gauge data to a pre-trained deep learning model, so that the deep learning model predicts the current influence range of the advance support pressure of the working face based on the drill cuttings amount data; optimizing the support design of the working face based on the influence range of the advance support pressure of the working face predicted by the deep learning model, and supporting the working face according to the optimized support design.

[0006] In addition, the method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to the above embodiment of the present invention may also have the following additional technical features:

[0007] According to one embodiment of the present invention, the deep learning model predicts the influence range of the working face advance support pressure based on the drill cuttings amount data, including: extracting characteristic parameters of the drill cuttings amount data, the characteristic parameters including the peak value, increment and change rate of the drill cuttings amount; substituting the peak value, increment and change rate of the drill cuttings amount into a mathematical model between the drill cuttings amount and the influence range of the advance support pressure to predict the current influence range of the advance support pressure.

[0008] According to one embodiment of the present invention, the mathematical model between the amount of drill cuttings and the influence range of the advance support pressure is a multiple linear regression model, which is expressed as follows:

[0009] L=α0+α1P+α2ΔQ+α3R+ε1

[0010] Wherein, L is the influence range of the leading support pressure, α0 is the intercept, α1, α2, and α3 are regression coefficients, ε1 is the error, P is the peak value of the cuttings amount, ΔQ is the increment, and R is the rate of change.

[0011] According to another embodiment of the present invention, the mathematical model between the amount of drill cuttings and the influence range of the advanced support pressure is a multivariate nonlinear regression model, which is expressed as follows:

[0012] L=β0+β1P+β2P 2 +β3ΔQ+β4R+ε2

[0013] Wherein, L is the influence range of the leading support pressure, β0 is the intercept, β1, β2, β3, and β4 are regression coefficients, ε2 is the error, P is the peak value of the cuttings amount, ΔQ is the increment, and R is the rate of change.

[0014] According to one embodiment of the present invention, the training process of the deep learning model includes: collecting the drill cuttings data and corresponding stress gauge data of each drill hole in the working face, and preprocessing the drill cuttings data and stress gauge data; constructing the deep learning model, the deep learning model includes an input layer, a position code, an encoder, a decoder and an output layer; wherein the encoder includes a multi-head self-attention mechanism, a feedforward neural network and a residual connection and normalization, and the decoder includes a multi-head self-attention mechanism, an encoder-decoder attention mechanism, a feedforward neural network and a residual connection and normalization; selecting the mean square error as the loss function, and using an optimizer to minimize the loss function to train and optimize the deep learning model, and storing the trained deep learning model; applying the trained deep learning model to an actual coal mining project, monitoring the drill cuttings data of each drill hole in the current working face in real time, predicting the current advance support pressure influence range, and verifying the prediction results of the deep learning model through field monitoring data and numerical simulation results, wherein the verification indicators include R 2 Score,Root Mean Square Error.

[0015] According to one embodiment of the present invention, preprocessing of drill cuttings data and strain gauge data includes: cleaning the drill cuttings data and strain gauge data, removing missing values and outliers, and ensuring the integrity and accuracy of the data; standardizing the cleaned drill cuttings data and strain gauge data, and normalizing the data to improve the convergence speed and prediction accuracy of the model.

[0016] According to one embodiment of the present invention, training and optimizing the deep learning model includes: training the deep learning model using a training set, updating the model parameters through forward propagation, loss calculation and back propagation; and regularly evaluating the model performance on the validation set to prevent overfitting; the training process adopts an early stopping mechanism, and when the validation set loss no longer decreases within 5 consecutive epochs, the training is stopped to ensure the generalization ability of the deep learning model; and optimizing the performance of the deep learning model by adjusting hyperparameters, wherein the hyperparameters include learning rate, batch size, and hidden layer dimension.

[0017] According to one embodiment of the present invention, the support design of the working face is optimized based on the influence range of the advance support pressure of the working face predicted by the deep learning model, including: when the predicted influence range of the advance support pressure is greater than the preset influence range threshold, increasing the support strength of the working face; when the predicted influence range of the advance support pressure is not greater than the preset influence range threshold, reducing the support strength of the working face to save costs.

[0018] According to one embodiment of the present invention, the method further includes: sending early warning information for early warning when the deep learning model predicts an abnormal change in the advance support pressure.

[0019] According to one embodiment of the present invention, the abnormal changes in the advance support pressure include: the increase in the peak value of the drill cuttings volume within a preset time is greater than the preset peak increase threshold; or, the increase in the drill cuttings volume continues to increase within a preset time; or, the increase in the rate of change of the drill cuttings volume within a preset time is greater than the preset rate of change increase threshold.

[0020] Compared with the prior art, the method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings in the embodiment of the present invention has the following beneficial effects:

[0021] 1. Improve monitoring accuracy: Through high-precision sensors and automation devices, real-time and accurate monitoring of drill cuttings volume can be achieved, ensuring the reliability and timeliness of monitoring data.

[0022] 2. Accurately predict the impact range of advance support pressure: Through deep learning models, the impact range of advance support pressure can be accurately predicted, providing a scientific basis for support design and ensuring the rationality and safety of support design.

[0023] 3. Optimize support design: Based on the prediction results, optimize the support design to improve the stability and reliability of the support structure, reduce the waste of support materials, and reduce production costs.

[0024] 4. Improve working face safety: Through real-time monitoring and prediction, timely measures can be taken to reduce safety hazards, improve the safety of the working face, and ensure the smooth progress of coal mining.

[0025] 5. Improve production efficiency: reduce manual operations, reduce labor intensity, improve work efficiency and production efficiency, and ensure the efficiency and economy of coal mining.

[0026] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Flowchart of a method for determining the influence range of the advanced support pressure of a working surface based on the amount of drill cuttings according to an embodiment of the present invention;

[0028] Figure 2 A top view of a drilling arrangement according to one embodiment of the present invention;

[0029] Figure 3 is a front view of a drilling arrangement according to one embodiment of the present invention;

[0030] Figure 4 2 is an architecture diagram of a Transformer network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0032] The following describes a method for determining the influence range of the advanced support pressure of a working surface based on the amount of drill cuttings, as proposed in an embodiment of the present invention, with reference to the accompanying drawings.

[0033] like Figure 1 As shown, the method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings implemented by the present invention may include the following steps:

[0034] S1, drill multiple boreholes in the coal body in front of the working face according to the preset spacing, and obtain the drill cuttings data of each borehole during the drilling process. The preset spacing can be calibrated according to the actual situation of the working face, for example, Figure 2 and Figure 3 As shown, the preset spacing is 10-30m.

[0035] Specifically, a drill bit and drill rod are used to drill holes in the coal body. The drill bit is made of high-strength alloy material to ensure the accuracy and depth of the drilling. The coal dust in the drill hole is collected by a powder collection telescopic cylinder and a powder collection fixed cylinder. The telescopic cylinder can automatically adjust its length according to the drilling depth to ensure the complete collection of the coal dust. The drilling rig provides the power for drilling, and the air compressor is used to remove the coal dust in the drill hole to ensure smooth drilling. The coal dust collection monitoring system includes high-precision sensors and automation devices for real-time monitoring of changes in the amount of drill cuttings. The sensor uses a high-precision weighing sensor to ensure the accuracy of the data. The control system controls the operation of the entire monitoring system, including data acquisition, transmission and storage. The control system uses an industrial-grade computer to ensure the efficiency and reliability of data processing.

[0036] S2, sending the drill cuttings data of each drill hole and the corresponding strain gauge data to a pre-trained deep learning model, so that the deep learning model can predict the current leading support pressure influence range of the working face based on the drill cuttings data.

[0037] S3, optimize the support design of the working face based on the influence range of the working face advance support pressure predicted by the deep learning model, and support the working face according to the optimized support design.

[0038] First, multiple boreholes are drilled at preset intervals in the coal mass ahead of the working face. During the drilling process, drill cuttings data for each borehole is collected. Drill cuttings refers to the amount of coal dust removed from the coal mass during drilling. Changes in drill cuttings volume are closely related to the stress state of the coal mass. When the coal mass is subjected to high stress, drill cuttings volume may change, so drill cuttings volume can indirectly reflect the stress state of the coal mass. Simultaneously, strain gauge data corresponding to the drill hole locations is collected. Strain gauges are instruments used to measure internal stress in the coal mass, providing more direct stress information. The strain gauges are embedded in the coal mass corresponding to the boreholes, and data is read via data lines. The drill cuttings data and strain gauge data are then fed into a pre-trained deep learning model. A deep learning model (Transformer network) is an artificial intelligence model trained on large amounts of data that can learn complex relationships and patterns within the data. By learning the relationship between drill cuttings volume and the range of influence of the lead support pressure, the deep learning model can predict the current range of influence of the lead support pressure at the working face based on the input drill cuttings data.

[0039] Lead abutment pressure refers to the redistribution of stress within the coal body ahead of the working face caused by coal mining during the mining process. The range of this stress increase is the range of influence of the lead abutment pressure. Accurately determining this range is crucial for optimizing the support design of the working face. Support design refers to the support structures and measures designed to ensure the safety and stability of the working face. The range of influence of the lead abutment pressure predicted by the deep learning model can be optimized to make the support design of the working face more reasonable, ensuring the safety of the working face while improving mining efficiency. Finally, the working face is supported according to the optimized support design to ensure the safety and stability of the working face during the mining process.

[0040] According to one embodiment of the present invention, a deep learning model predicts the influence range of the advance support pressure of the working face based on the drill cuttings amount data, including: extracting characteristic parameters of the drill cuttings amount data, the characteristic parameters including the peak value, increment and change rate of the drill cuttings amount; substituting the peak value, increment and change rate of the drill cuttings amount into a mathematical model between the drill cuttings amount and the influence range of the advance support pressure to predict the current influence range of the advance support pressure.

[0041] Specifically, the peak value of the cuttings volume can reflect the stress concentration in the coal body at a specific depth. The peak value usually occurs at the edge of the influence range of the lead support pressure, because the coal body stress is concentrated here, which tends to produce more cuttings. The incremental change in the cuttings volume can indicate the gradual accumulation of coal body stress. The incremental calculation formula for the cuttings volume is as follows:

[0042] ΔQ i =Q i -Q i-1

[0043] Where ΔQ i is the increment of cuttings at the i-th time point, Q i is the amount of cuttings at the i-th time point, Q i-1 is the amount of drill cuttings at the previous point in time.

[0044] The change rate of drill cuttings can capture the dynamic changes of coal stress. The calculation formula of the change rate of drill cuttings is as follows:

[0045]

[0046] Among them, R i is the rate of change of the amount of cuttings at the i-th time point.

[0047] Specifically, the peak value, increment and change rate of the drill cuttings amount are substituted into the mathematical model between the drill cuttings amount and the influence range of the advance support pressure. The machine learning model predicts the current influence range of the advance support pressure based on the peak value, increment and change rate of the drill cuttings amount and the mathematical model.

[0048] According to one embodiment of the present invention, the mathematical model between the amount of drill cuttings and the influence range of the advance support pressure is a multiple linear regression model, which is expressed as follows:

[0049] L=α0+α1P+α2ΔQ+α3R+ε1

[0050] Where L is the influence range of the leading support pressure, α0 is the intercept, α1, α2, and α3 are regression coefficients, ε1 is the error, P is the peak value of the cuttings, ΔQ is the increment, and R is the rate of change.

[0051] According to another embodiment of the present invention, the mathematical model between the amount of drill cuttings and the influence range of the advance support pressure is a multivariate nonlinear regression model, which is expressed as follows:

[0052] L=β0+β1P+β2P 2 +β3ΔQ+β4R+ε2

[0053] Where L is the influence range of the leading support pressure, β0 is the intercept, β1, β2, β3, and β4 are regression coefficients, ε2 is the error, P is the peak value of the cuttings, ΔQ is the increment, and R is the rate of change.

[0054] According to one embodiment of the present invention, the training process of the deep learning model includes the following steps:

[0055] S21, collecting the drill cuttings data and the corresponding strain gauge data of each drill hole on the working surface, and pre-processing the drill cuttings data and the strain gauge data.

[0056] According to one embodiment of the present invention, preprocessing of drill cuttings data and strain gauge data includes: cleaning the drill cuttings data and strain gauge data, removing missing values and outliers, and ensuring the integrity and accuracy of the data; standardizing the cleaned drill cuttings data and strain gauge data, and normalizing the data to improve the convergence speed and prediction accuracy of the model; converting the standardized drill cuttings data and strain gauge data into time series data, and dividing the time series data into a training set, a validation set, and a test set.

[0057] Specifically, cleaning and standardizing the monitored drill cuttings data can improve data reliability and model convergence speed. Converting drill cuttings and strain gauge data into time series data facilitates deep learning model processing. The training set, validation set, and test set typically have a ratio of 70%, 15%, and 15%.

[0058] S22, construct a deep learning model, the deep learning model includes an input layer, a position encoding, an encoder, a decoder and an output layer; wherein the encoder includes a multi-head self-attention mechanism, a feedforward neural network and residual connection and normalization, and the decoder includes a multi-head self-attention mechanism, an encoder-decoder attention mechanism, a feedforward neural network and residual connection and normalization.

[0059] Specifically, the deep learning model architecture of the embodiment of the present invention is as follows Figure 4 As shown in the figure, the input layer defines the dimensionality of the input data, including the characteristic dimensions of the drill cuttings volume and strain gauge data. The input data is normalized to ensure model stability and convergence speed. Positional encoding adds positional encoding to the input data, enabling the model to understand the order of the sequence. Positional encoding uses sine and cosine functions to ensure the model's sensitivity to sequence position.

[0060] Encoder: Multi-head self-attention mechanism: Through multiple attention heads, the model can simultaneously focus on information at different positions in the input sequence. The dimension of each attention head is 64, ensuring the model's parallel computing capability and information extraction capability. Feedforward neural network: Performs nonlinear transformation on the features at each position to extract higher-level features. The feedforward network uses two fully connected layers with ReLU as the activation function to ensure the model's nonlinear fitting capability. Residual connection and normalization: Residual connection and normalization layers are added after each sublayer (self-attention and feedforward network) to accelerate training and improve model stability. Normalization uses Layer Normalization to ensure model stability and convergence speed.

[0061] Decoder: Multi-head self-attention mechanism: The decoder first performs self-attention calculation on the input to ensure that the model can capture the temporal dependencies in the input sequence. Encoder-decoder attention mechanism: The decoder interacts with the output of the encoder through the attention mechanism to obtain the features extracted by the encoder. The encoder-decoder attention mechanism uses multi-head attention to ensure the feature fusion capability of the model. Feedforward neural network: Similar to the encoder, it performs nonlinear transformation on the features at each position to extract more advanced features. The feedforward network uses two fully connected layers with ReLU as the activation function to ensure the nonlinear fitting capability of the model. Residual connection and normalization: Residual connection and normalization layers are also added to ensure the stability and convergence speed of the model. Output layer: The output layer is designed to predict the influence range of the advance support pressure. The output layer uses a linear layer with an output dimension of 1 to ensure the prediction accuracy of the model.

[0062] The operation process of the deep learning model of the embodiment of the present invention from input to output is as follows: the input layer receives the drill cuttings and stress gauge data, and the data is normalized. The position encoding layer adds position encoding to the input data so that the model can perceive the order of the sequence. The multi-head self-attention mechanism in the encoder processes the input data and captures the relationship between different positions; the feedforward neural network performs nonlinear transformation on the features of each position, and the residual connection and normalization layer accelerate training and improve model stability. The multi-head self-attention mechanism in the decoder processes the decoder input and captures time dependencies; the encoder-decoder attention mechanism interacts with the encoder output to obtain the features extracted by the encoder; the feedforward neural network performs nonlinear transformation on the features of each position; the residual connection and normalization layer accelerate training and improve model stability. The linear layer of the output layer maps the output of the decoder to the predicted range of influence of the lead support pressure.

[0063] S23, select the mean square error as the loss function, and use the optimizer to minimize the loss function to train and optimize the deep learning model, and store the trained deep learning model.

[0064] Specifically, the mean square error (MSE) is selected as the loss function to measure the difference between the model prediction value and the true value. The MSE formula is:

[0065]

[0066] Among them, y i is the true value, is the predicted value, and n is the number of samples.

[0067] The Adam optimizer is used to minimize the loss function. Combining the advantages of AdaGrad and RMSProp, the Adam optimizer features an adaptive learning rate and momentum term, ensuring rapid model convergence and stable training. It's important to understand that the Adam optimizer efficiently minimizes the loss function by combining momentum and an adaptive learning rate. It calculates the first-order moment (mean) and second-order moment (variance) of the gradient and dynamically adjusts the learning rate, decreasing it when the gradient is large and increasing it when the gradient is small. This mechanism not only accelerates model convergence but also improves training stability. Furthermore, the Adam optimizer introduces bias correction to ensure more accurate estimates in the initial stage and improves numerical stability by adding small numerical terms. In the documentation, the Adam optimizer is used to train the Transformer network, gradually optimizing model parameters by minimizing the mean squared error (MSE) loss function, thereby improving prediction accuracy and generalization.

[0068] According to one embodiment of the present invention, training and optimizing a deep learning model includes: training the deep learning model using a training set, updating model parameters through forward propagation, loss calculation, and backpropagation; regularly evaluating model performance on a validation set to prevent overfitting; employing an early stopping mechanism during training, stopping training when the validation set loss no longer decreases within five consecutive epochs to ensure the generalization ability of the deep learning model; and optimizing the performance of the deep learning model by adjusting hyperparameters, including the learning rate, batch size, and hidden layer dimensions. In some embodiments, a learning rate of 0.001, a batch size of 32, and a hidden layer dimension of 512 are used to ensure the training effect and generalization ability of the model.

[0069] Trained models can be saved for later use and deployment. Models are saved using PyTorch's torch.save and torch.load methods to ensure model persistence and reusability.

[0070] S24, the trained deep learning model is applied to the actual coal mining project, the cuttings data of each drill hole in the current working face is monitored in real time, the influence range of the current advanced support pressure is predicted, and the prediction results of the deep learning model are verified by field monitoring data and numerical simulation results. The verification indicators include R 2 Score,Root Mean Square Error.

[0071] Specifically, the trained deep learning model is applied to actual coal mining projects to monitor the amount of drill cuttings in the boreholes near the working face in real time and predict the impact range of the advance support pressure. The monitoring data is transmitted to the control system in real time via a wireless transmission module to ensure the real-time and reliability of the data. The prediction results of the model are verified by field monitoring data and numerical simulation results to ensure the accuracy and reliability of the model. The verification indicators include R 2 Score, root mean square error (RMSE), etc., to ensure the prediction accuracy and generalization ability of the model. 2 The formula for the score is as follows:

[0072]

[0073] The formula for the root mean square error (RMSE) is as follows:

[0074]

[0075] Among them, y i is the true value, is the predicted value, is the average of all true values, and n is the number of samples.

[0076] According to one embodiment of the present invention, the support design of a working face is optimized based on the influence range of the advanced support pressure predicted by a deep learning model. This includes increasing the support strength of the working face if the predicted influence range of the advanced support pressure exceeds a preset influence range threshold; and reducing the support strength of the working face to save costs if the predicted influence range of the advanced support pressure does not exceed the preset influence range threshold. The preset influence range threshold can be calibrated based on actual conditions.

[0077] Specifically, if the predicted advance support pressure has a large impact range, it indicates that the stress concentration of the coal body is high and the support strength needs to be increased. The following methods can be used to increase the support strength:

[0078] Use higher-strength support materials: for example, use high-strength anchor rods, cables, or brackets.

[0079] Increase the cross-sectional area of the support structure: for example, increase the diameter of the anchor rod or the cross-sectional area of the support.

[0080] Increase support density: Increase the density of support points in stress concentration areas to better disperse stress.

[0081] If the predicted influence range of the advanced support pressure is small, it indicates that the stress concentration of the coal body is low, and the support strength can be appropriately reduced to save costs. The following methods can be used to reduce the support strength of the working face:

[0082] Use lower strength support materials: for example, use normal strength anchors or brackets.

[0083] Reduce the cross-sectional area of the support structure: for example, reduce the diameter of the anchor rod or the cross-sectional area of the support.

[0084] Reduce support density: Reduce the density of support points in areas with lower stress.

[0085] According to one embodiment of the present invention, the above method further includes: sending early warning information for early warning when the deep learning model predicts an abnormal change in the advance support pressure.

[0086] Furthermore, according to one embodiment of the present invention, abnormal changes in the lead support pressure include: an increase in the peak value of the cuttings volume within a preset time period exceeding a preset peak increase threshold; or an increase in the cuttings volume continuously increasing within a preset time period; or an increase in the rate of change of the cuttings volume within a preset time period exceeding a preset rate of change increase threshold. The preset time period, the preset peak increase threshold, and the preset rate of change increase threshold can all be calibrated based on actual conditions.

[0087] Specifically, if the peak value of the drill cuttings volume increases significantly, it indicates that the stress concentration of the coal body at a specific depth is high, and there may be abnormal changes in the lead support pressure. If the incremental increase in the drill cuttings volume continues to increase, it indicates that the stress in the coal body is rapidly accumulating, and the lead support pressure may be rapidly expanding. If the rate of change of the drill cuttings volume increases significantly, it indicates that the dynamic changes in the coal body stress are drastic, and the lead support pressure may be rapidly spreading.

[0088] In summary, the method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to the embodiment of the present invention has the following beneficial effects:

[0089] 1. Improve the accuracy and real-time performance of drill cuttings monitoring

[0090] High-precision sensors: High-precision sensors and automated cuttings measurement devices enable real-time, accurate monitoring of cuttings data in multiple boreholes near the working face. This significantly improves the reliability and timeliness of monitoring data, providing a solid data foundation for subsequent determination of the impact range of advanced support pressure.

[0091] Data preprocessing: The monitored drill cuttings data is cleaned and standardized to further improve the data quality and model training effect.

[0092] 2. Accurately predict the impact range of advance support pressure

[0093] Application of a deep learning model: By using an improved Transformer network, we deeply explore the inherent connection between drill cuttings data and strain gauge data, establishing a nonlinear model that accurately reflects the relationship between the two. This makes the prediction of the influence range of the lead support pressure more accurate and reliable.

[0094] Dynamic assessment and early warning: Using a trained deep learning model, combined with real-time monitoring of drill cuttings data, it is possible to quickly and accurately assess the current impact range of the advance support pressure and provide early warning of possible abnormal changes in the advance support pressure, providing timely decision-making support for coal mine safety production.

[0095] 3. Optimize support design

[0096] Scientific Basis: By accurately determining the influence range of the advanced support pressure, a scientific basis is provided for support design. This makes the design of the support structure more reasonable, effectively responding to stress changes in the dynamic pressure zone, and improving the stability and reliability of the support structure.

[0097] Practical application verification: Through a large number of field tests and data collection at different coal mining sites, the applicability and reliability of the model were verified, ensuring the optimization effect of support design.

[0098] 4. Improve the safety and production efficiency of the working surface

[0099] Reducing safety hazards: By real-time monitoring and accurately predicting the impact range of advance support pressure, timely and effective measures such as adjusting the support structure are taken to significantly reduce safety hazards on the working face and improve the safety of the working face.

[0100] Improved production efficiency: The application of intelligent drill cuttings monitoring systems and deep learning models reduces manual operations, reduces labor intensity, improves work efficiency and data reliability, and thus improves production efficiency.

[0101] 5. Realize intelligent and automated monitoring and analysis

[0102] Automated monitoring: The intelligent drill cuttings monitoring system realizes the automation of monitoring and analysis, reduces manual operations, and improves work efficiency and data reliability.

[0103] Real-time analysis: Deep learning models can process monitoring data in real time, provide real-time analysis results, and provide timely support for production decisions.

[0104] 6. Data-driven decision support

[0105] Multi-source data fusion: By integrating drill cuttings data and stress gauge data, more comprehensive information on coal stress status is provided, providing a more reliable basis for production decision-making.

[0106] Data visualization: Through data visualization technology, complex monitoring data and analysis results are presented to production managers in an intuitive way, allowing them to make decisions quickly.

[0107] 7. Improve economic efficiency

[0108] Reduce costs: By optimizing the support design, unnecessary support materials and labor costs are reduced, thereby reducing production costs.

[0109] Increased production: By improving the safety and production efficiency of the working face, the downtime caused by safety hazards and production accidents is reduced, and coal production is increased.

[0110] Through the obvious effects brought about by the above technology, the present invention has made significant progress in determining the influence range of the advance support pressure in coal mining, providing a strong guarantee for the safety and production efficiency of coal mining.

[0111] Practical application cases

[0112] Case 1: Determination of the influence range of advance support pressure in a coal mine

[0113] Background: During the mining process of a coal mine, traditional methods made it difficult to accurately determine the influence range of the advance support pressure, resulting in unreasonable support design and affecting the safe and efficient production of the working face.

[0114] Implementation: Multiple drill holes were placed in the coal body near the working face. High-precision sensors and automated cuttings monitoring devices were used to monitor changes in cuttings volume in real time. A deep learning model (Transformer network) was used to establish a mathematical model linking cuttings volume and the influence range of the lead support pressure.

[0115] Effect: Through model prediction, the influence range of the advance support pressure was accurately determined, the support design was optimized, and the safety and production efficiency of the working face were improved.

[0116] Case 2: Application of intelligent drilling cuttings monitoring system

[0117] Background: In a coal mine, the traditional method of monitoring drill cuttings volume has problems such as data lag and insufficient accuracy, which affects the accurate prediction of the advance support pressure.

[0118] Implementation: An intelligent cuttings monitoring system, combined with high-precision sensors and automated devices, enables real-time and accurate monitoring of cuttings. A deep learning model (Transformer network) is used to establish a mathematical model linking cuttings volume and the influence range of lead support pressure.

[0119] Effect: The accuracy and real-time performance of drill cuttings monitoring were significantly improved, the prediction of the influence range of the advance support pressure was optimized, and the safety and production efficiency of the working face were improved.

[0120] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0121] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0122] In the present invention, unless otherwise specified or limited, the terms "installed," "connected," "connect," "fixed," etc. should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediate medium; internal communication between two components, or interaction between two components, unless otherwise specified. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0123] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for determining the influence range of the advanced support pressure of a working surface based on the amount of drill cuttings, characterized in that: The method comprises: Drilling multiple boreholes at preset intervals in the coal body in front of the working face, and obtaining drill cuttings data for each borehole during the drilling process; The drill cuttings data and corresponding strain gauge data of each drill hole are fed into a pre-trained deep learning model, so that the deep learning model predicts the current leading support pressure influence range of the working face based on the drill cuttings data; The support design of the working face is optimized based on the influence range of the advance support pressure of the working face predicted by the deep learning model, and the working face is supported according to the optimized support design.

2. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 1, characterized in that: The deep learning model predicts the influence range of the advanced support pressure of the working face based on the cuttings volume data, including: Extracting characteristic parameters of the drill cuttings data, wherein the characteristic parameters include a peak value, an increment, and a change rate of the drill cuttings; The peak value, increment and change rate of the drill cuttings amount are substituted into a mathematical model between the drill cuttings amount and the influence range of the leading support pressure to predict the current influence range of the leading support pressure.

3. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 2, characterized in that: The mathematical model between the amount of drill cuttings and the influence range of the advance support pressure is a multiple linear regression model, and the expression is: L=α0+α1P+α2ΔQ+α3R+ε1 Wherein, L is the influence range of the leading support pressure, α0 is the intercept, α1, α2, and α3 are regression coefficients, ε1 is the error, P is the peak value of the cuttings amount, ΔQ is the increment, and R is the rate of change.

4. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 2, characterized in that: The mathematical model between the amount of drill cuttings and the influence range of the advanced support pressure is a multivariate nonlinear regression model, and the expression is: L=β0+β1P+β2P 2 +β3ΔQ+β4R+ε2 Wherein, L is the influence range of the leading support pressure, β0 is the intercept, β1, β2, β3, and β4 are regression coefficients, ε2 is the error, P is the peak value of the cuttings amount, ΔQ is the increment, and R is the rate of change.

5. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 1, characterized in that: The training process of the deep learning model includes: Collect the drill cuttings data and corresponding strain gauge data of each drill hole on the working surface, and pre-process the drill cuttings data and strain gauge data; Constructing the deep learning model, which includes an input layer, a positional encoding, an encoder, a decoder, and an output layer; wherein the encoder includes a multi-head self-attention mechanism, a feedforward neural network, and residual connections and normalization; and the decoder includes a multi-head self-attention mechanism, an encoder-decoder attention mechanism, a feedforward neural network, and residual connections and normalization; Selecting mean square error as a loss function and using an optimizer to minimize the loss function to train and optimize the deep learning model, and storing the trained deep learning model; The trained deep learning model is applied to actual coal mining projects to monitor the cuttings data of each drill hole in the current working face in real time, predict the current range of influence of the advance support pressure, and verify the prediction results of the deep learning model through field monitoring data and numerical simulation results. The verification indicators include R 2 Score,Root Mean Square Error.

6. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 5, characterized in that: Preprocessing of drill cuttings data and stress gauge data includes: Clean drill cuttings data and stress gauge data, remove missing values and outliers, and ensure data integrity and accuracy; Standardize the cleaned drill cuttings data and strain gauge data, and perform normalization on the data to improve the model's convergence speed and prediction accuracy; The standardized drill cuttings and strain gauge data are converted into time series data, and the time series data are divided into training set, validation set and test set.

7. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 6, characterized in that: Training and optimizing the deep learning model includes: Using the training set to train the deep learning model, updating the model parameters through forward propagation, loss calculation and backpropagation; and regularly evaluating the model performance on the validation set to prevent overfitting; The training process uses an early stopping mechanism. When the validation set loss stops decreasing within 5 consecutive epochs, the training is stopped to ensure the generalization ability of the deep learning model. The performance of the deep learning model is optimized by adjusting hyperparameters, including learning rate, batch size, and hidden layer dimension.

8. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 1, characterized in that: The support design of the working face is optimized by predicting the influence range of the working face advance support pressure based on the deep learning model, including: When the predicted influence range of the advance support pressure is greater than the preset influence range threshold, the support strength of the working face is increased; When the predicted influence range of the advance support pressure is not greater than the preset influence range threshold, the support strength of the working face is reduced to save costs.

9. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 1, characterized in that: The method further comprises: When the deep learning model predicts an abnormal change in the advance support pressure, an early warning message is sent to issue an early warning.

10. The method for determining the influence range of the advanced support pressure of the working surface based on the amount of drill cuttings according to claim 9, characterized in that: Abnormal changes in advance support pressure include: The increase in the peak value of the drill cuttings volume within the preset time is greater than the preset peak increase threshold; or The increment of the cuttings volume continues to increase within a preset time; or, An increase value of the change rate of the drill cuttings amount within a preset time is greater than a preset change rate increase threshold.

Images