Coal mine water drainage quantity prediction and drilling field optimization design method based on neural network

Abstract

The application discloses a coal mine water drainage quantity prediction and drilling field optimization design method based on a neural network, introduces a space occurrence characteristic quantification model and a convolutional neural network, deeply analyzes stratum characteristics such as sandstone and mudstone proportion and lithology categories, combines a bidirectional long short-term memory network to dynamically learn historical water drainage quantity data and real-time monitoring information, and significantly improves prediction accuracy. Meanwhile, an adaptive optimization module of the drilling field is constructed based on a reinforcement learning algorithm, and drilling hole spacing, depth and layout mode are automatically generated according to a prediction result. The application divides sandstone and mudstone categories and space occurrence characteristics into a neural network input layer, solves the problem of insufficient modeling of lithologic heterogeneity in a traditional method, realizes real-time optimization of drilling hole parameters through dynamic coupling of prediction and design links, and significantly improves universality and flexibility of the technology. The application provides an efficient, accurate and economic technical means for coal mine water disaster prevention and control.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mining, in particular to a coal mine water drainage quantity prediction and drilling site optimization design method based on a neural network. BACKGROUND

[0002] Coal occupies a pivotal position in China's energy structure and is a key energy source for ensuring national energy security and stable economic development. For a long time, coal has been the main energy source in China, providing indispensable power support for industrial production, power supply and many other fields. With the sustained development of the economy, the demand for coal resources is increasing, which has led to the expansion of the scale and depth of coal mining. However, as the depth of mining continues to increase, the geological conditions of coal mining become increasingly complex, and water damage problems become increasingly prominent, becoming an important factor restricting the safety and efficient production of coal mines.

[0003] In the process of coal mining, underground water, goaf water, surface water and other water sources are extremely prone to water damage accidents under the combined action of complex geological structures, rock properties and other factors. These water damage not only causes serious damage to the production equipment of the coal mine, but also threatens the safety of the miners, causing huge economic losses and social impact to the coal mining enterprises. In order to effectively prevent and control the occurrence of water damage accidents and ensure the safety of coal mining, water drainage work is particularly important. Through water drainage, the water pressure and water level of the aquifer can be reduced, and the possibility of water damage can be reduced, creating a safe working environment for coal mining.

[0004] In the process of coal mining, the drainage of the aquifer in the coal seam roof and floor is one of the core measures to prevent and control water damage. The purpose of water drainage is to reduce the water pressure of the aquifer through borehole drainage to prevent water inrush accidents. However, the successful implementation of water drainage engineering depends on accurate water quantity prediction and scientific drilling site design. Traditional water drainage quantity prediction methods mainly rely on empirical formulas (such as the large well method, analytical method) or static geological models, which are difficult to adapt to complex and variable geological conditions. For example, in sand and mudstone alternating strata, the permeability of sandstone and mudstone differs significantly, and the traditional method cannot accurately quantify the influence of sand and mudstone proportion and its category division (such as the permeability difference between brittle sandstone and plastic mudstone) on water drainage quantity, resulting in large prediction errors. In addition, existing drilling site design mostly uses fixed drilling spacing and layout mode, lacking dynamic analysis of the spatial occurrence characteristics of the coal seam roof and floor (such as heterogeneity, fracture development degree), leading to unreasonable drilling layout, low drainage efficiency or serious resource waste.

[0005] Currently, the main problems of drilling site design are as follows:

[0006] Insufficient geological parameter modeling: Traditional methods lack quantitative analysis of sand-shale ratio and its classification, making it difficult to accurately reflect the impact of lithologic heterogeneity on drainage path.

[0007] Poor design flexibility: Existing solutions cannot dynamically adjust drilling parameters based on real-time geological data such as sand-shale interbed thickness and lithologic interface changes, resulting in suboptimal drainage effect.

[0008] Lack of multi-factor coupling: Traditional methods do not fully consider the synergistic effect of sand-shale classification and hydrogeological conditions, lacking precise modeling of drainage path under complex geological conditions.

[0009] Low intelligence: Existing technologies rely heavily on human experience and lack data-driven intelligent optimization methods, making it difficult to adapt to drainage needs under complex geological conditions.

[0010] Existing technologies have not addressed the deep integration of sand-shale classification, spatial occurrence characteristics, and dynamic hydrological monitoring data, and lack an intelligent method that can integrate drainage volume prediction and drilling layout optimization. Therefore, there is an urgent need for a technical solution that can overcome the above deficiencies and has high precision and flexibility to meet the needs of coal mine water disaster prevention under complex geological conditions. SUMMARY

[0011] To address the above technical deficiencies, the present invention provides a coal mine drainage volume prediction and drilling field optimization design method based on neural networks, which can solve the deficiencies of traditional drilling field design methods in terms of geological parameter limitations, design rigidity, and lack of multi-factor coupling. This method includes sand-shale classification and spatial occurrence characteristics in the neural network input layer, solving the problem of insufficient modeling of lithologic heterogeneity in traditional methods. By dynamically coupling the prediction and design stages, real-time optimization of drilling parameters is achieved, significantly improving the universality and flexibility of the technology.

[0012] To solve the above technical problems, the present invention adopts the following technical solutions:

[0013] The present invention provides a coal mine drainage volume prediction and drilling field optimization design method based on neural networks, which specifically includes the following steps:

[0014] (1) Data collection: Collect multi-dimensional data of the mining area, including geological exploration data, spatial occurrence characteristic data of coal seam roof and floor, sand-shale ratio and lithologic classification data, historical drainage volume data, and real-time hydrological monitoring information;

[0015] (2) Data preprocessing: Preprocess the collected data, including data cleaning, denoising, and standardization to ensure data accuracy and consistency;

[0016] (3) Feature quantization: Introduce a spatial occurrence feature quantization model to convert the heterogeneity and fracture development degree of the coal seam roof and floor into quantifiable indicators;

[0017] (4) Model construction: Incorporate the sandstone and mudstone classification and the indicators after spatial occurrence feature quantization into the input layer of the neural network to construct a hybrid neural network model containing a convolutional neural network and a bidirectional long short-term memory network. The convolutional neural network is used to analyze the stratigraphic features, and the hybrid neural network model with a bidirectional long short-term memory network is used to learn the time series dynamic characteristics of historical and real-time water drainage data;

[0018] (5) Model training: Train the hybrid neural network model based on the preprocessed data, and optimize the model parameters to minimize the prediction error;

[0019] (6) Model evaluation and optimization: Use the cross-validation method to evaluate and optimize the trained hybrid neural network model to ensure that the model has good generalization ability and stability;

[0020] (7) Drilling field optimization design: Use reinforcement learning algorithm to construct a drilling field adaptive optimization module, and automatically generate drilling spacing, depth and layout mode according to the water drainage prediction results output by the neural network;

[0021] (8) Dynamic coupling: Dynamically couple water drainage prediction and drilling field optimization design, and adjust drilling parameters according to real-time geological conditions.

[0022] Preferably, in the data collection stage of step (1), the geological exploration data covers detailed information of different depths and different regions, and is updated regularly every day to reflect the dynamic changes of the geological conditions in the mining area.

[0023] Preferably, in step (1), the spatial occurrence feature data includes sandstone and mudstone alternating layer thickness, lithological interface change and thin mudstone layer merging threshold value, wherein the thin mudstone layer merging threshold value is a set thickness value, and when the thickness of the thin mudstone layer is lower than the threshold value, the thin mudstone layer is merged with other thick rock layers. The thickness of the mudstone layer is less than 1m, and the ratio of sandstone and mudstone is greater than 5:1, so the mudstone is classified into sandstone.

[0024] Preferably, in step (5), the model training adopts a dynamic learning rate adjustment strategy to automatically adjust the learning rate according to the training progress to accelerate the convergence.

[0025] Preferably, in step (7), the reinforcement learning algorithm adopts a particle swarm optimization algorithm to search for the optimal drilling parameter combination in a high-dimensional solution space through a particle swarm, and the objective function is to maximize the drainage efficiency and minimize the drilling cost.

[0026] Preferably, in real-time adjustment of the drilling parameters in step (8), a monitoring data feedback mechanism is established to trigger the adjustment process of the drilling field design parameters when significant changes in the geological conditions are detected.

[0027] Preferably, the particle position updating formula in the particle swarm optimization algorithm is:

[0028]

[0029] wherein, is the current position of the i-th particle in the t-th generation, is the current position of the particle in the t+1-th generation;

[0030] The particle velocity updating formula is:

[0031]

[0032] wherein, w is the inertia weight, controlling the influence of the previous velocity of the particle, c1, c2 are acceleration constants, and r1, r2 are random numbers, is the historical best position of the particle itself, g (t) is the best position in the whole particle group; through the mutual cooperation of the particle group, a global optimal solution or an approximate optimal solution is found.

[0033] Preferably, the influence radius of the water drainage amount is determined based on the following formula:

[0034]

[0035] wherein: Q is set as the underground water drainage amount, M0 is the groundwater recharge index, and the value of M0 is 0.3-1.0 L / s·km 2 , and M0 is 0.5 L / s·km 2 .

[0036] Preferably, an optimization objective function is constructed with the goals of maximizing the drainage efficiency and minimizing the drilling cost, and the drilling spacing, depth and layout mode are dynamically adjusted through the reinforcement learning algorithm.

[0037] The present application has the following advantages:

[0038] 1. The traditional method relies on empirical formula and static geological model, and it is difficult to deal with complex geological conditions, resulting in low prediction accuracy of water drainage amount. The present application introduces a spatial occurrence characteristic quantitative model and a convolutional neural network, which can deeply analyze the formation characteristics such as sandstone-mudstone ratio and lithology category, and combine a bidirectional long short-term memory network to dynamically learn historical and real-time data, capture the time sequence characteristics and dynamic change law of the data, thereby significantly improving the prediction accuracy of the water drainage amount.

[0039] 2、The existing drilling field design mostly adopts a fixed mode, and lacks dynamic response to changes in geological conditions. The drilling field adaptive optimization module is constructed based on the reinforcement learning algorithm, which can automatically generate drilling spacing, depth and layout mode according to the prediction result, and adjust the drilling parameters in real time in combination with real-time changes in geological conditions, so that the dynamic optimization of the drilling field design is realized, the drainage efficiency is improved, and resource waste is avoided.

[0040] 3、The traditional method is insufficient for modeling the lithologic heterogeneity, and it is difficult to accurately reflect the influence of sandstone and mudstone ratio and category division on the drainage water path. The sandstone and mudstone category division and spatial occurrence characteristics are included in the neural network input layer, which effectively solves this problem and can more accurately simulate the drainage water process under complex stratum conditions;

[0041] 4、The existing technology does not fully consider the deep integration of sandstone and mudstone category division, spatial occurrence characteristics and dynamic hydrological monitoring data, and the intelligent degree is low. Through data collection, preprocessing, model construction and training and other steps, the deep integration of multiple factors is realized, and the intelligent optimization method based on data driving is adopted, which improves the universality and flexibility of the technology, and provides an efficient, accurate and economic technical means for coal mine water disaster prevention. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0043] Figure 1 The flowchart of the coal mine drainage water volume prediction and drilling field optimization design method based on neural network provided by the embodiment of the present application is shown in the figure;

[0044] Figure 2 In the figure, a is the unit water inflow distribution diagram of the coal seam roof aquifer;

[0045] Figure 2 In the figure, b is the aquiclude thickness diagram between coal seams;

[0046] Figure 2 In the figure, c is the aquifer thickness diagram between coal seams;

[0047] Figure 2 In the figure, d is the spacing diagram between the bottom of 9 coal and the top of 11 coal;

[0048] Figure 2 In the figure, e is the aquifer-aquiclude ratio diagram between coal seams;

[0049] Figure 2Fig. 2 is a schematic diagram of average water pressure of a working face roof aquifer;

[0050] Figure 3 Fig. 6 is a coal mine water drainage amount prediction result graph provided by an embodiment of the present application;

[0051] Figure 4 Fig. 7 is a coal mine working face drilling field design result graph provided by an embodiment of the present application. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0053] Referring to Figure 1 The embodiment provides a coal mine water drainage amount prediction and drilling field optimization design method based on a neural network, and specifically comprises the following steps:

[0054] (1) Data collection: collect multi-dimensional data of a mining area, including geological exploration data, spatial occurrence characteristic data of a coal seam roof and floor, sandstone and mudstone ratio and lithology classification data, historical water drainage amount data and real-time hydrological monitoring information;

[0055] Referring to Figure 2 Fig. 2 is a schematic diagram of distribution of each parameter of a coal mine stratum and aquifer, and in the embodiment, geological and hydrological data of a target coal mine working face are comprehensively collected, the aquifer-seal layer ratio between coal seams is calculated by means of drilling core data and logging data, the unit water inflow distribution of a coal seam roof aquifer is determined according to hydrological test data such as pumping test, the interval between the 9th coal seam floor and the 11th coal seam roof is determined by three-dimensional geological modeling or drilling data, the average water pressure of the working face roof aquifer is monitored in real time by using a pressure sensor or obtained according to existing hydrological monitoring data, and information such as the aquifer thickness and the seal layer thickness between coal seams is extracted from drilling data and geological profile graphs.

[0056] The geological exploration data covers detailed information of different depths and different regions, and is regularly updated to reflect dynamic changes of the geological conditions of the mining area.

[0057] The spatial occurrence characteristic data includes sandstone and mudstone alternating layer thickness, lithology interface change and thin mudstone layer merging threshold value, wherein the thin mudstone layer merging threshold value is a set thickness value, and when the thickness of the thin mudstone layer is lower than the threshold value, the thin mudstone layer is merged with other thick rock layers.

[0058] (2) Data preprocessing: The collected data is preprocessed, including data cleaning, denoising and standardization, to ensure the accuracy and consistency of the data; eliminate the dimensional difference, so that the data of different orders of magnitude have comparability;

[0059] (3) Feature quantization: Introduce a spatial occurrence feature quantization model to convert the heterogeneity and fracture development degree of the coal seam roof and floor into quantifiable indicators; divide the data into training set and test set according to a certain proportion, the training set is used for neural network training, and the test set is used for neural network verification.

[0060] (4) Model construction: The sandstone and mudstone classification and the indicators after spatial occurrence feature quantization are included in the input layer of the neural network, a hybrid neural network model containing convolutional neural network and bidirectional long short-term memory network is constructed, the convolutional neural network is used to analyze the stratum characteristics, and the hybrid neural network model with bidirectional long short-term memory network is used to learn the time sequence dynamic characteristics of historical and real-time water drainage data;

[0061] In this embodiment, a hybrid neural network model is constructed, and the collected geological and hydrological parameters such as aquifer-barrier layer ratio, unit water inflow distribution, coal seam spacing, aquifer average water pressure, aquifer thickness and barrier layer thickness are introduced as input variables into the input layer, the hidden layer adopts neural network combined with bidirectional long short-term memory network, and the nonlinear relationship and spatiotemporal characteristics in the input data are fully mined to improve the learning ability of the model, and the output layer outputs the prediction value of the water drainage amount of the target working face.

[0062] (5) Model training: Based on the preprocessed data, the hybrid neural network model is trained, and the model parameters are optimized to minimize the prediction error; the model training adopts a dynamic learning rate adjustment strategy, and the learning rate is automatically adjusted according to the training progress to accelerate convergence.

[0063] In this embodiment, the training set data is used to train the constructed neural network, the mean square error is used as the loss function, the model parameters are continuously optimized through the back propagation algorithm, and the training is continued until the model converges, ensuring that the model can accurately learn the rules in the data.

[0064] (6) Model evaluation and optimization: The cross-validation method is used to evaluate and optimize the trained hybrid neural network model to ensure that the model has good generalization ability and stability;

[0065] In this embodiment, after the model is trained, the geological and hydrological data of the target working face, such as the aquifer-barrier layer ratio, unit water inflow distribution, coal seam spacing, average water pressure of the aquifer, aquifer thickness and barrier layer thickness, are accurately input into the trained neural network model. The neural network performs complex calculations based on the input data to obtain the water drainage prediction value and generate a water drainage distribution map, and outputs the prediction results including the water drainage at different positions and its spatial distribution characteristics.

[0066] (7) Drilling field optimization design: an adaptive optimization module of the drilling field is constructed by using a reinforcement learning algorithm, and the drilling spacing, depth and layout mode are automatically generated according to the water drainage prediction results output by the neural network.

[0067] The reinforcement learning algorithm uses a particle swarm optimization algorithm to search for the optimal drilling parameter combination in a high-dimensional solution space through a particle swarm, and the objective function is to maximize the drainage efficiency and minimize the drilling cost.

[0068] (8) Dynamic coupling: dynamically coupling the water drainage prediction and the drilling field optimization design, the drilling parameters are adjusted according to the real-time geological conditions.

[0069] Referring to Figures 3-4 In this embodiment, based on the water drainage prediction results, the drilling field design is optimized by using a reinforcement learning algorithm, and drilling fields are continuously set at the edge of the effective influence radius to superimpose water drainage. A reasonable optimization objective function is constructed to maximize the drainage efficiency and minimize the drilling cost, providing a clear direction for subsequent optimization work. Through the reinforcement learning algorithm, the drilling spacing, depth and layout mode are dynamically adjusted, considering the cost and construction convenience. After multiple iterations, the optimal drilling scheme is generated, which is then applied to the target working face to verify the drainage effect and economy, ensuring the feasibility and effectiveness of the scheme.

[0070] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A neural network-based coal mine drainage volume prediction and drilling field optimization design method, characterized in that, Specifically comprising the following steps: (1) Data collection: Collect multi-dimensional data of the mining area, including geological exploration data, spatial occurrence characteristic data of coal seam roof and floor, sandstone and mudstone ratio and lithology classification data, historical drainage water volume data and real-time hydrological monitoring information; (2) Data preprocessing: Preprocess the collected data, including data cleaning, denoising and standardization, to ensure the accuracy and consistency of the data; (3) Feature quantization: Introduce a spatial occurrence characteristic quantization model to convert the heterogeneity and fissure development degree of the coal seam roof and floor into quantifiable indicators; (4) Model construction: Construct a hybrid neural network model containing a convolutional neural network and a bidirectional long short-term memory network, wherein the convolutional neural network is used to analyze the stratum characteristics, and the bidirectional long short-term memory network is used to learn the time series dynamic characteristics of the historical and real-time drainage water volume data; introduce the collected aquifer-aquiclude ratio, unit water inflow distribution, coal seam spacing, aquifer average water pressure, aquifer thickness and aquiclude thickness, and geological and hydrological parameters as input variables into the input layer, use the convolutional neural network in the hidden layer combined with the bidirectional long short-term memory network, and output the target working face drainage water volume prediction value in the output layer; (5) Model training: Train the model based on the preprocessed data, and optimize the model parameters to minimize the prediction error; (6) Model evaluation and optimization: Evaluate and optimize the trained model using the cross-validation method to ensure that the model has good generalization ability and stability; (7) Drilling field optimization design: Use the particle swarm optimization algorithm to construct a drilling field adaptive optimization module, search for the optimal drilling parameter combination in the high-dimensional solution space through the particle swarm, and automatically generate the drilling spacing, depth and layout mode according to the neural network output drainage water volume prediction result; the objective function is to maximize the drainage efficiency and minimize the drilling cost; The particle position update formula in the particle swarm optimization algorithm is: wherein, is the current position of the i-th particle at the t-th generation, is the current position of the particle at the t+1-th generation; The particle velocity update formula is: where w is the inertia weight, controlling the influence of the previous velocity of the particle, c1, c2 are acceleration constants, and r1, r2 are random numbers, is the best position of the particle itself, g (t) is the best position of the entire population of particles; through the mutual cooperation of the population of particles, a global optimal solution or an approximate optimal solution is found; (8) Dynamic coupling: Dynamically couple the drainage water volume prediction and drilling field optimization design, and adjust the drilling parameters according to the real-time geological conditions; Specifically: based on the drainage water volume prediction result, use the particle swarm optimization algorithm to optimize the drilling field design, The following formula is used to determine the influence radius of the drainage water volume: In the formula: Q is the underground drainage water volume, and M0 is the groundwater recharge index.

2. The neural network-based coal mine drainage volume prediction and drilling field optimization design method according to claim 1, characterized in that, In the data collection stage of step (1), the geological exploration data covers detailed information of different depths and different regions, and is updated regularly to reflect the dynamic changes of the geological conditions in the mining area.

3. The neural network-based coal mine drainage volume prediction and drilling field optimization design method according to claim 1, characterized in that, In step (1), the spatial occurrence characteristic data includes sandstone and mudstone interbed thickness, lithology interface change and thin mudstone layer merging threshold value, wherein the thin mudstone layer merging threshold value is a set thickness value, and when the thickness of the thin mudstone layer is lower than the threshold value, the thin mudstone layer is merged with other thick rock layers.

4. The neural network-based coal mine drainage volume prediction and drilling field optimization design method according to claim 1, characterized in that, In step (5), the model training adopts a dynamic learning rate adjustment strategy to automatically adjust the learning rate to accelerate convergence according to the training progress.

5. The neural network-based coal mine drainage volume prediction and drilling field optimization design method according to claim 1, characterized in that, In step (8), when adjusting the drilling parameters in real time, a monitoring data feedback mechanism is established to trigger the adjustment process of the drilling field design parameters when significant changes in the geological conditions are detected.

6. The neural network-based coal mine drainage volume prediction and drilling field optimization design method according to claim 1, characterized in that, In step (8), the drill field is continuously set at the edge of the effective influence radius to superimpose water drainage; a reasonable optimization objective function is constructed with the goal of maximizing drainage efficiency and minimizing drilling cost, and the particle swarm optimization algorithm is used to dynamically adjust the drilling spacing, depth and layout mode.

Images