Coal mining intelligent control method and system based on paste filling

Abstract

The invention discloses a coal mining intelligent control method and system based on paste filling, and the method comprises the steps: obtaining the underground environment of a coal mine and the geological condition data of a filling area, building a paste filling three-dimensional digital model according to the geological conditions and goaf forms of different filling areas, and enabling the model to comprise the attribute information of a paste ratio, pumping parameters and the like; the method comprises the following steps: arranging a multi-parameter sensor at a key position of a paste conveying pipeline, collecting rheological characteristic parameters, including viscosity, density, sand content, particle distribution and the like, of paste in the pipeline in real time, and forming a paste rheological characteristic data set; and the optimized paste proportion and pumping parameters are input into a paste preparation system and a pumping control system, the preparation and conveying process of the paste is dynamically controlled in real time, and it is ensured that the performance of the paste meets the filling requirement.

Description

Technical Field

[0001] The invention belongs to the field of coal mining, and in particular relates to an intelligent control method and system for coal mining based on paste filling. Background Art

[0002] When coal mines adopt paste filling mining methods, it is necessary to monitor the rheological properties and filling effects of the paste in real time to ensure the filling quality and mining safety. However, due to the complex and changeable underground environment of coal mines, the long paste delivery pipeline and the harsh layout environment, it is difficult for traditional monitoring methods to comprehensively and accurately obtain the key parameters of the paste filling process. At the same time, even a slight change in the paste ratio may cause its rheological properties to change, thereby affecting the filling effect, and the existing filling process is difficult to dynamically adjust the paste ratio and pumping parameters according to real-time monitoring data. In addition, the geological conditions and goaf morphology of different filling areas are different, and it is necessary to formulate filling plans in a targeted manner, but there is currently a lack of an intelligent method to optimize the filling sequence and time window, resulting in low filling efficiency. Therefore, there is an urgent need for an intelligent control method based on real-time monitoring data that can dynamically adjust the paste ratio and pumping parameters, and optimize the filling sequence and time window to improve the quality and efficiency of paste filling and ensure the safety of coal mining. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention proposes an intelligent control method and system for coal mining based on paste filling, which realizes the intelligent control and optimization of the paste filling process, significantly improves the filling quality and efficiency, and provides effective guarantee for ensuring the safety of coal mining.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] An intelligent control method for coal mining based on paste filling comprises the following steps:

[0006] Obtain geological condition data of the underground environment and filling area of ​​the coal mine, and establish a three-dimensional digital model of paste filling according to the geological conditions and goaf morphology of different filling areas, wherein the three-dimensional digital model of paste filling contains the attribute information of paste proportion and pumping parameters;

[0007] Multi-parameter sensors are placed at key locations of the paste delivery pipeline to collect rheological characteristic parameters of the paste in the pipeline in real time. The rheological characteristic parameters include viscosity, density, sand content, and particle distribution, forming a paste rheological characteristic data set.

[0008] Based on the real-time collected data set of paste rheological properties, combined with the pre-established three-dimensional digital model of paste filling, the paste ratio and pumping parameters are dynamically optimized using machine learning algorithms to obtain the optimized paste ratio and pumping parameters;

[0009] Input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system to perform real-time dynamic control on the paste preparation and transportation processes, ensuring that the paste performance meets the filling requirements;

[0010] Install stress sensors and displacement sensors in the filling area to monitor the stress distribution and deformation conditions inside the filling body in real time, obtain the stability data of the filling body. At the same time, use ultrasonic detection technology to obtain the density distribution data inside the filling body;

[0011] Input the stability data of the filling body and the density distribution data into the paste filling effect evaluation model, and use deep learning algorithms to comprehensively evaluate the filling effect, judge whether the filling quality meets the requirements. If the filling quality is unqualified, re-optimize the paste ratio and pumping parameters;

[0012] Install a multi-point displacement monitoring system in the underground coal mine environment to monitor the displacement changes of the goaf roof and surrounding rock in real time. Combine the stability evaluation results of the filling body, and use reinforcement learning algorithms to optimize the paste filling sequence and time window, and formulate an intelligent filling plan to maximize the integrity and timeliness of the goaf filling and ensure the safety of coal mining.

[0013] Preferably, obtain the geological condition data of the underground coal mine environment and the filling area. For the geological conditions and goaf morphology of different filling areas, establish a three-dimensional digital model of paste filling, including:

[0014] Obtain the geological condition data of the underground coal mine environment and the filling area, and preprocess the geological condition data to obtain a standardized geological condition dataset;

[0015] For the standardized geological condition dataset, use clustering algorithms for classification. According to the clustering results, divide different filling areas to obtain the geological condition characteristics of each filling area;

[0016] Obtain the goaf morphology data corresponding to each filling area, and construct a three-dimensional geometric model of the goaf through three-dimensional modeling to obtain the spatial morphology information of the goaf;

[0017] According to the geological condition characteristics of the filling area and the spatial morphology information of the goaf, use optimization algorithms to calculate the paste ratio parameters to obtain the optimal paste ratio parameters for each filling area;

[0018] According to the optimal paste ratio parameters and the spatial position of the filling area, use fluid mechanics simulation algorithms to simulate the filling process of the paste in the goaf to obtain the best pumping parameters;

[0019] Integrate the geological condition characteristics of the filling area, the spatial morphology information of the goaf, the optimal paste ratio parameters, and the optimal pumping parameters into a three-dimensional digital model to construct a three-dimensional visualization model of paste filling.

[0020] Preferably, according to the dataset of the rheological properties of the paste collected in real time, combined with the pre-established three-dimensional digital model of paste filling, use machine learning algorithms to dynamically optimize the paste ratio and pumping parameters. The optimized paste ratio and pumping parameters obtained include:

[0021] Discretize the three-dimensional digital model of the paste filling area into a finite element mesh model;

[0022] Input the rheological property data of the paste collected in real time into the machine learning algorithm, and establish a mapping relationship between the rheological properties of the paste and the paste ratio and pumping parameters through the support vector machine algorithm;

[0023] Use the genetic algorithm to optimize the mapping relationship, and through crossover and mutation operations, search for the optimal combination of paste ratio and pumping parameters;

[0024] Input the optimized paste ratio and pumping parameters into the finite element mesh model, and simulate the flow and filling process of the paste in the filling area through the computational fluid dynamics equation;

[0025] If the simulation results meet the construction requirements, apply the optimized paste ratio and pumping parameters to the actual construction.

[0026] Preferably, input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system to perform real-time dynamic control on the paste preparation and transportation process to ensure that the paste performance meets the filling requirements, including:

[0027] According to the optimized paste ratio, determine the dosage and proportion of each component, and input the ratio parameters into the paste preparation system;

[0028] Obtain the optimized pumping parameters and input the pumping parameters into the pumping control system;

[0029] During the paste preparation process, adopt on-line monitoring technology to obtain the viscosity and density performance parameters of the paste in real time;

[0030] Compare the real-time obtained paste performance parameters with the preset target performance parameters. If the deviation exceeds the threshold, dynamically adjust the paste ratio and pumping parameters;

[0031] Through the machine learning algorithm, establish a mapping relationship model between the paste ratio and pumping parameters and the paste performance;

[0032] During the paste transportation process, according to the real-time monitored pipeline pressure and flow parameters, combined with the mapping relationship model, dynamically optimize the pumping parameters to ensure the stability of the transportation process;

[0033] Fill the paste to the specified position, and use visual detection technology to judge whether the filling effect meets the requirements. If not, readjust the paste ratio and pumping parameters until the filling requirements are met.

[0034] Preferably, input the filling body stability data and density distribution data into the paste filling effect evaluation model, and use the deep learning algorithm to comprehensively evaluate the filling effect to judge whether the filling quality meets the requirements. If the filling quality is unqualified, then re-optimize the paste ratio and pumping parameters, including:

[0035] Obtain the filling body stability data and density distribution data, and input them into the pre-constructed paste filling effect evaluation model;

[0036] Adopt the convolutional neural network algorithm to perform feature extraction and representation learning on the input filling body stability data and density distribution data, and obtain the comprehensive evaluation result of the filling effect;

[0037] Judge whether the filling quality meets the preset qualified standard according to the evaluation result. If it meets, output the qualified result; otherwise, trigger the optimization process of the paste ratio and pumping parameters;

[0038] In the optimization process, search for the optimal combination of paste ratio and pumping parameters through the genetic algorithm, and input the optimized parameters into the paste filling effect evaluation model for re-evaluation;

[0039] Apply the optimized paste ratio and pumping parameters to the actual filling operation, obtain the new filling body stability data and density distribution data; and input them into the paste filling effect evaluation model to re-evaluate the filling effect, forming a closed-loop optimization mechanism.

[0040] Preferably, deploy a multi-point displacement monitoring system in the underground coal mine environment to real-time monitor the displacement changes of the goaf roof and surrounding rock. Combine with the filling body stability evaluation result, and use the reinforcement learning algorithm to optimize the paste filling sequence and time window, and formulate an intelligent filling plan to maximize the integrity and timeliness of the goaf filling and ensure the safety of coal mining, including:

[0041] According to the characteristics of the coal mine environment, deploy multi-point displacement sensors at the key positions of the goaf roof and surrounding rock to form a real-time displacement monitoring network;

[0042] Obtain the displacement data collected by the sensors, and through data preprocessing and feature extraction, obtain the key feature parameters that can reflect the displacement change trend of the goaf roof and surrounding rock;

[0043] Input the displacement characteristic parameters into the pre - constructed filling body stability evaluation model, and judge the stability level of the current filling body through model reasoning and calculation;

[0044] If the stability level of the filling body is lower than the preset threshold, trigger the generation process of the intelligent filling plan;

[0045] Otherwise, keep the current filling plan unchanged and continue to monitor the displacement change;

[0046] Adopt the reinforcement learning algorithm, with the improvement of the filling body stability as the optimization goal, and the filling sequence and time window as decision variables. Through the interaction and exploration between the agent and the environment, gradually optimize the filling plan;

[0047] During the training process of reinforcement learning, guide the agent to learn through the reward function, while taking into account the filling integrity and timeliness, maximize the stability level of the filling body;

[0048] Send the optimized intelligent filling plan to the automated filling equipment, control the sequence and time of paste filling, achieve precise and rapid filling of the goaf, and ensure the safety and stability of the coal mining process.

[0049] The present invention also provides an intelligent control system for coal mining based on paste filling, including: a construction module, a collection module, an optimization module, a control module, an acquisition module, a discrimination module, and a monitoring module;

[0050] The construction module is used to obtain the geological condition data of the underground environment of the coal mine and the filling area, and establish a three - dimensional digital model of paste filling according to the geological conditions and goaf morphology of different filling areas. Among them, the three - dimensional digital model of paste filling includes paste ratio and pumping parameter attribute information;

[0051] The collection module is used to deploy multi - parameter sensors at key positions of the paste conveying pipeline, and collect the rheological characteristic parameters of the paste in the pipeline in real time. Among them, the rheological characteristic parameters include: viscosity, density, sand content, particle distribution, to form a paste rheological characteristic data set;

[0052] The optimization module is used to dynamically optimize the paste ratio and pumping parameters according to the real - time collected paste rheological characteristic data set, combined with the pre - established three - dimensional digital model of paste filling, and obtain the optimized paste ratio and pumping parameters by using machine learning algorithms;

[0053] The control module is used to input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system, and conduct real - time dynamic control on the paste preparation and conveying process to ensure that the paste performance meets the filling requirements;

[0054] The acquisition module is used to deploy stress sensors and displacement sensors in the filling area, monitor the stress distribution and deformation conditions inside the filling body in real time, obtain the stability data of the filling body, and at the same time, obtain the density distribution data inside the filling body through ultrasonic detection technology;

[0055] The discrimination module is used to input the stability data of the filling body and the density distribution data into the paste filling effect evaluation model, comprehensively evaluate the filling effect by using the deep learning algorithm, judge whether the filling quality meets the requirements, and if the filling quality is unqualified, re-optimize the paste ratio and pumping parameters;

[0056] The monitoring module is used to deploy a multi-point displacement monitoring system in the underground coal mine environment, monitor the displacement changes of the roof and surrounding rock of the goaf in real time, combine the evaluation results of the filling body stability, and use the reinforcement learning algorithm to optimize the paste filling sequence and time window, formulate an intelligent filling plan, ensure the integrity and timeliness of the goaf filling to the greatest extent, and ensure the safety of coal mine mining.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0058] The present invention discloses an intelligent control method and system for coal mine mining based on paste filling. Aiming at the problems that the paste performance is difficult to accurately control and the filling effect is difficult to evaluate in real time during the paste filling process of the coal mine goaf, by establishing a three-dimensional digital model of paste filling, multi-parameter sensors are deployed at key positions of the pipeline to collect paste rheological characteristic data in real time, and the paste ratio and pumping parameters are dynamically optimized by combining machine learning algorithms. At the same time, by deploying stress and displacement sensors and ultrasonic detection devices in the filling area, the stability and density data of the filling body are obtained, and the filling effect is evaluated by using the deep learning algorithm. If the filling quality is unqualified, the paste parameters are re-optimized. In addition, combining the roof displacement monitoring data of the goaf, the reinforcement learning algorithm is used to optimize the filling sequence and time window. The present invention realizes the intelligent control and optimization of the paste filling process, significantly improves the filling quality and efficiency, and provides an effective guarantee for ensuring the safety of coal mine mining. Description of the Drawings

[0059] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0060] Figure 1 It is a schematic flow chart of an intelligent control method for coal mine mining based on paste filling according to an embodiment of the present invention. Detailed Embodiments

[0061] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0062] To make the above objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0063] Embodiment 1

[0064] As Figure 1 shown, the present invention provides an intelligent control method for coal mine mining based on paste filling, including the following steps:

[0065] Obtain the geological condition data of the underground environment of the coal mine and the filling area, and establish a three-dimensional digital model of paste filling for different geological conditions and goaf shapes in the filling area. Among them, the three-dimensional digital model of paste filling includes paste ratio and pumping parameter attribute information;

[0066] Arrange multi-parameter sensors at key positions of the paste conveying pipeline to collect the rheological characteristic parameters of the paste in the pipeline in real time. Among them, the rheological characteristic parameters include: viscosity, density, sand content, and particle distribution, forming a paste rheological characteristic data set;

[0067] According to the real-time collected paste rheological characteristic data set, combined with the pre-established three-dimensional digital model of paste filling, use machine learning algorithms to dynamically optimize the paste ratio and pumping parameters, and obtain the optimized paste ratio and pumping parameters;

[0068] Input the optimized paste ratio and pumping parameters into the paste preparation system and pumping control system to perform real-time dynamic control on the preparation and transportation process of the paste, and ensure that the paste performance meets the filling requirements;

[0069] Arrange stress sensors and displacement sensors in the filling area to monitor the stress distribution and deformation of the filling body in real time, obtain the filling body stability data, and at the same time, obtain the density distribution data inside the filling body through ultrasonic detection technology;

[0070] Input the filling body stability data and density distribution data into the paste filling effect evaluation model, and use deep learning algorithms to comprehensively evaluate the filling effect to judge whether the filling quality meets the requirements. If the filling quality is unqualified, re-optimize the paste ratio and pumping parameters;

[0071] Deploy a multi-point displacement monitoring system in the underground environment of a coal mine to monitor the displacement changes of the roof and surrounding rocks in the goaf in real time. Combine the evaluation results of the filling body stability, use the reinforcement learning algorithm to optimize the paste filling sequence and time window, formulate an intelligent filling plan, and ensure the integrity and timeliness of the goaf filling to the greatest extent, so as to ensure the safety of coal mine mining.

[0072] In this embodiment, obtain the geological condition data of the underground environment of the coal mine and the filling area. For the geological conditions and goaf morphology of different filling areas, establish a three-dimensional digital model of paste filling, and the model contains attribute information such as paste ratio and pumping parameters, including:

[0073] Obtain the geological condition data of the underground environment of the coal mine and the filling area, and preprocess the geological condition data to obtain a standardized geological condition data set;

[0074] For the standardized geological condition data set, use the clustering algorithm for classification, divide different filling areas according to the clustering results, and obtain the geological condition characteristics of each filling area;

[0075] Obtain the goaf morphology data corresponding to each filling area, and construct a three-dimensional geometric model of the goaf through three-dimensional modeling to obtain the spatial morphology information of the goaf;

[0076] According to the geological condition characteristics of the filling area and the spatial morphology information of the goaf, use the optimization algorithm to calculate the paste ratio parameters to obtain the optimal paste ratio parameters for each filling area;

[0077] According to the optimal paste ratio parameters and the spatial position of the filling area, use the fluid mechanics simulation algorithm to simulate the filling process of the paste in the goaf to obtain the best pumping parameters;

[0078] Integrate the geological condition characteristics of the filling area, the spatial morphology information of the goaf, the optimal paste ratio parameters and the best pumping parameters into the three-dimensional digital model to construct a three-dimensional visualization model of paste filling.

[0079] Exemplarily, during coal mine exploitation, obtaining geological condition data of the underground environment is a crucial first step. For example, through geological exploration techniques such as drilling and seismic reflection methods, information about rock types, hardness, fracture development degree, etc. can be obtained. These data are preprocessed, such as removing outliers and normalizing, to form a standardized geological condition dataset. This step ensures the accuracy and reliability of subsequent analysis. Next, using clustering algorithms such as K-means or hierarchical clustering to classify the standardized data can effectively divide the filling areas with similar geological characteristics. For example, clustering areas with similar geological conditions together can help determine which areas may require similar filling materials and techniques. This classification is based on the similarity of geological data, thus improving the filling effect and safety. After determining the filling areas, it is necessary to obtain the goaf morphology data corresponding to each area. Through three-dimensional scanning techniques such as laser scanning or photogrammetry, a three-dimensional geometric model of the goaf can be constructed. This model provides detailed spatial morphology information of the goaf, such as volume, shape, and spatial position, which is crucial for subsequent filling work. According to the geological condition characteristics of the filling areas and the spatial morphology information of the goaf, optimization algorithms such as genetic algorithms or simulated annealing algorithms can be used to calculate the optimal paste mixing ratio parameters. These parameters include the water-cement ratio of the paste, types and proportions of additives, etc., to ensure that the fluidity and stability of the paste best adapt to specific geological and spatial conditions. With the optimal paste mixing ratio parameters, next, use fluid mechanics simulation algorithms such as finite element analysis to simulate the filling process of the paste in the goaf. This step can help determine the optimal pumping parameters, such as pumping pressure and speed, to ensure that the paste can evenly fill the entire goaf and avoid structural instability caused by uneven filling. Finally, integrate the geological condition characteristics of the filling areas, the spatial morphology information of the goaf, the optimal paste mixing ratio parameters, and the optimal pumping parameters into a three-dimensional digital model to construct a three-dimensional visualization model of paste filling. This model can not only intuitively display the filling process and results but also be used to analyze and evaluate the effect of the filling plan and optimize the filling plan. In this way, the optimized filling plan parameters can be applied to the actual paste filling operation to guide on-site construction and improve safety and economic benefits.

[0080] In this embodiment, multi-parameter sensors are arranged at key positions of the paste conveying pipeline to collect in real time the rheological characteristic parameters of the paste in the pipeline, including viscosity, density, sand content, particle distribution, etc., to form a paste rheological characteristic dataset.

[0081] According to the rheological characteristics of the paste in the pipeline, determine the specific position coordinates for arranging multi-parameter sensors at key positions. Use machine learning algorithms to analyze the data collected in real time by the multi-parameter sensors to obtain the real-time rheological characteristic parameters of the paste in the pipeline. If the real-time rheological characteristic parameters exceed the preset threshold range, it is judged that the pipeline transportation is abnormal and an early warning is triggered. Through big data analysis technology, explore the correlation law between historical rheological characteristic parameters and transportation status, and establish a prediction model. Use the prediction model to predict the pipeline transportation status in a future period of time. If the prediction result is abnormal, give an early warning in advance. Obtain particle distribution data, group the particles through a clustering algorithm, and judge whether there is abnormal aggregation of particles. Comprehensively analyze parameters such as viscosity, density, and sand content to evaluate the comprehensive transportation performance index of the paste in the pipeline, which is used to guide the optimization control of transportation.

[0082] In this embodiment, according to the dataset of the rheological characteristics of the paste collected in real time, combined with the pre-established three-dimensional digital model of paste filling, use machine learning algorithms to dynamically optimize the paste ratio and pumping parameters. The optimized paste ratio and pumping parameters obtained include:

[0083] Discretize the three-dimensional digital model of the paste filling area into a finite element mesh model;

[0084] Input the rheological characteristic data of the paste collected in real time into the machine learning algorithm, and establish the mapping relationship between the rheological characteristics of the paste and the paste ratio and pumping parameters through the support vector machine algorithm;

[0085] Use the genetic algorithm to optimize the mapping relationship, and through crossover and mutation operations, search for the optimal combination of paste ratio and pumping parameters;

[0086] Input the optimized paste ratio and pumping parameters into the finite element mesh model, and simulate the flow and filling process of the paste in the filling area through the computational fluid dynamics equation;

[0087] If the simulation result meets the construction requirements, apply the optimized paste ratio and pumping parameters to the actual construction.

[0088] Exemplarily, in modern engineering construction, the rheological properties of paste materials are crucial for ensuring filling quality and efficiency. By arranging sensors at key positions to collect parameters such as the viscosity, shear force, and shear rate of the paste in real time, a detailed dataset of the paste's rheological properties can be formed. For example, in a typical mine filling project, engineers will install these sensors at key positions such as the inlet, middle section, and outlet of the paste conveying pipeline. These data not only help monitor the flow state of the paste but also provide a basis for subsequent data analysis. Using these real-time data, a mapping relationship between the rheological properties of the paste, its formulation, and pumping parameters can be established through the support vector machine algorithm. For example, by analyzing the data, it is found that when the viscosity of the paste exceeds a certain threshold, it may be necessary to adjust the ratio of cement and water or change the pumping pressure to optimize the fluidity. The establishment of this mapping relationship relies on a large amount of historical data and complex mathematical models to ensure the accuracy and practicality of the prediction. Further, by optimizing these mapping relationships through the genetic algorithm, the optimal paste formulation and pumping parameters can be found. The genetic algorithm can effectively search for the optimal solution by simulating the process of natural selection, such as operations like crossover and mutation. In actual operation, this may mean continuously adjusting the type of cement, the pH value of water, and the type of additives to see which combination shows the best fluidity and stability in the simulation. Applying these optimized parameters to a three-dimensional digital model, the flow and filling process of the paste in the filling area can be simulated through computational fluid dynamics equations. This step is carried out by a high-performance computer, and the simulation results can intuitively show the distribution of the paste in the filling space, whether there are dead corners or insufficient filling. For example, in a simulation project, engineers found that the fluidity of the paste was insufficient in a certain corner and successfully solved the problem by adjusting the pumping rate and direction at the inlet. During the actual construction process, the continuously collected rheological property data of the paste is fed back into the machine learning algorithm to achieve dynamic optimization control of the paste formulation and pumping parameters. This dynamic adjustment mechanism makes the entire filling process more intelligent and efficient, able to quickly adjust the construction strategy according to real-time data to ensure construction quality and safety. Through the application of these technologies, the engineering team can achieve fine management of the paste filling process, greatly improving construction efficiency and material utilization rate, while also reducing construction risks. The application of this series of technologies demonstrates the innovative application of modern engineering technologies in traditional construction fields, providing new ideas and methods for future engineering construction.

[0089] In this embodiment, the optimized paste formulation and pumping parameters are input into the paste preparation system and the pumping control system to perform real-time dynamic control on the preparation and transportation processes of the paste, ensuring that the paste performance meets the filling requirements, including:

[0090] According to the optimized paste formulation, determine the dosage and proportion of each component, and input the formulation parameters into the paste preparation system;

[0091] Obtain the optimized pumping parameters and input them into the pumping control system;

[0092] During the paste preparation process, adopt on-line monitoring technology to obtain the viscosity and density performance parameters of the paste in real time;

[0093] Compare the paste performance parameters obtained in real time with the preset target performance parameters. If the deviation exceeds the threshold, dynamically adjust the paste formulation and pumping parameters;

[0094] Establish a mapping relationship model between the paste formulation, pumping parameters and paste performance through machine learning algorithms;

[0095] During the paste transportation process, according to the pipeline pressure and flow parameters monitored in real time, combined with the mapping relationship model, dynamically optimize the pumping parameters to ensure the stability of the transportation process;

[0096] Fill the paste into the specified position, adopt visual inspection technology to judge whether the filling effect meets the requirements. If not, readjust the paste formulation and pumping parameters until the filling requirements are met.

[0097] Exemplarily, paste preparation is the starting link of the entire paste filling process. According to the optimized paste formulation, the precise dosage and proportion of each component need to be determined. For example, an optimized formulation may be: cement 40%, fly ash 30%, water 25%, additive 5%. Assuming that 100 tons of paste need to be prepared, then 40 tons of cement, 30 tons of fly ash, 25 tons of water, and 5 tons of additive are required. These formulation parameters will be input into the paste preparation system, and the system will automatically control the addition amount of each component. At the same time, the optimized pumping parameters, such as the pumping pressure is set to 3 MPa and the flow rate is set to 50 m 3 / h, will also be input into the pumping control system. During the paste preparation process, on-line monitoring technology is the key to ensuring the quality of the paste. Through sensors, the viscosity, density and other performance parameters of the paste can be obtained in real time. For example, the target viscosity is set to 20 Pa·s and the target density is set to 0 g / cm 3 . If the viscosity monitored in real time is 22 Pa·s and the density is 9 g / cm 3, and if the deviations of both of these two values from the target value exceed the preset thresholds (for example, viscosity ±10%, density ±5%), then it is necessary to dynamically adjust the paste ratio and pumping parameters. This is like a chef constantly adjusting the amount of seasonings during the cooking process according to the color, aroma, and taste of the dish. To achieve dynamic adjustment, it is necessary to establish a mapping relationship model between the paste ratio, pumping parameters, and paste properties. This can be achieved through machine learning algorithms, such as support vector machines or neural networks. A large amount of historical data can be collected, including different ratios, pumping parameters, and the corresponding paste property parameters, and then these data are used to train the model. The trained model can predict whether the current paste ratio and pumping parameters are appropriate based on the real-time monitored paste property parameters and give adjustment suggestions. The stability of the paste conveying process is crucial for the entire filling project. During the conveying process, it is necessary to monitor parameters such as pipeline pressure and flow rate in real time. For example, set the target pipeline pressure to 5 MPa and the target flow rate to 45 m 3 / h. If the real-time monitored pipeline pressure is too high, for example, it reaches 8 MPa, it indicates that the conveying resistance of the paste is too large. At this time, in combination with the mapping relationship model, the pumping pressure can be appropriately reduced or the proportion of water can be increased to reduce the viscosity, thereby reducing the conveying resistance and ensuring the stability of the conveying process. Finally, the paste needs to be filled into the specified position. To ensure the filling effect, visual detection technology is used to judge whether the filling effect meets the requirements. For example, an image of the filling area can be taken by a camera, and then image analysis technology is used to judge whether there is underfilling or overfilling. If it is found that the filling effect is not ideal, for example, there are voids in a certain area, then it is necessary to return to the paste preparation stage and readjust the paste ratio and pumping parameters until the filling requirements are met. This is like a sculptor constantly adjusting the shape and position of the clay during the creation process until the desired effect is achieved.

[0098] In this embodiment, stress sensors and displacement sensors are arranged in the filling area to monitor the stress distribution and deformation conditions inside the filling body in real time, and obtain the stability data of the filling body. At the same time, ultrasonic detection technology is used to obtain the density distribution data inside the filling body.

[0099] According to the characteristics of the filling area, stress sensors and displacement sensors are reasonably arranged to realize the real-time monitoring of the internal stress distribution and deformation of the filling body. Stress data at different positions inside the filling body are obtained through stress sensors to judge whether the stress distribution is uniform. If the stress distribution is not uniform, it indicates that there are potential stability problems inside the filling body. Deformation data at different positions inside the filling body are obtained through displacement sensors to judge whether the deformation condition is normal. If the deformation exceeds the preset threshold, it indicates that there are potential stability problems inside the filling body. The obtained stress distribution data and deformation data are fused and analyzed to comprehensively evaluate the stability inside the filling body and obtain the stability evaluation result. Ultrasonic detection technology is used to scan the inside of the filling body to obtain ultrasonic reflection signal data at different positions. According to the intensity and time difference of the reflection signals, the density distribution inside the filling body is calculated. The obtained density distribution data is compared and analyzed with the stability evaluation result to judge the correlation between the density distribution and stability and determine the key factors affecting the stability of the filling body. According to the analysis results, a stability prediction model is established using the support vector machine algorithm. By inputting the stress distribution, deformation condition, and density distribution data monitored in real time, the stability trend of the filling body in the next period of time is predicted, providing a decision-making basis for the safety management of the filling area.

[0100] In this embodiment, the filling body stability data and the density distribution data are input into the paste filling effect evaluation model, and the deep learning algorithm is used to comprehensively evaluate the filling effect to judge whether the filling quality meets the requirements. If the filling quality is unqualified, the paste ratio and pumping parameters are re-optimized, including:

[0101] Obtain the filling body stability data and the density distribution data, and input them into the pre-constructed paste filling effect evaluation model;

[0102] The convolutional neural network algorithm is used to extract features and perform representation learning on the input filling body stability data and density distribution data to obtain the comprehensive evaluation result of the filling effect;

[0103] Judge whether the filling quality meets the preset qualified standard according to the evaluation result. If it meets, output the qualified result; otherwise, trigger the optimization process of the paste ratio and pumping parameters;

[0104] In the optimization process, the genetic algorithm is used to search for the optimal combination of paste ratio and pumping parameters, and the optimized parameters are input into the paste filling effect evaluation model for re-evaluation;

[0105] Apply the optimized paste ratio and pumping parameters to the actual filling operation to obtain new filling body stability data and density distribution data; and input them into the paste filling effect evaluation model to re-evaluate the filling effect, forming a closed-loop optimization mechanism.

[0106] In this embodiment, a multi-point displacement monitoring system is deployed in the underground coal mine environment to monitor the displacement changes of the goaf roof and surrounding rock in real time. Combining the evaluation results of the backfill stability, the paste filling sequence and time window are optimized using a reinforcement learning algorithm, and an intelligent filling plan is formulated to ensure the integrity and timeliness of the goaf filling to the greatest extent and ensure coal mine safety, including:

[0107] According to the characteristics of the coal mine environment, multi-point displacement sensors are deployed at key positions on the goaf roof and surrounding rock to form a real-time displacement monitoring network;

[0108] The displacement data collected by the sensors are obtained, and through data preprocessing and feature extraction, key feature parameters that can reflect the displacement change trend of the goaf roof and surrounding rock are obtained;

[0109] The displacement feature parameters are input into a pre-constructed backfill stability evaluation model, and through model inference and calculation, the stability level of the current backfill is judged;

[0110] If the stability level of the backfill is lower than the preset threshold, the generation process of the intelligent filling plan is triggered;

[0111] Otherwise, the current filling plan remains unchanged, and the displacement changes are continuously monitored;

[0112] A reinforcement learning algorithm is adopted, with the goal of improving the backfill stability as the optimization target and the filling sequence and time window as decision variables. Through the interaction and exploration between the agent and the environment, the filling plan is gradually optimized;

[0113] During the training process of reinforcement learning, the agent is guided to learn through a reward function, while taking into account the filling integrity and timeliness, the stability level of the backfill is maximally improved;

[0114] The optimized intelligent filling plan is sent to the automated filling equipment to control the filling sequence and time of the paste, realizing the precise and rapid filling of the goaf and ensuring the safety and stability of the coal mine mining process.

[0115] Exemplarily, in a coal mine environment, the stability of the roof and surrounding rock in the goaf is crucial for mine safety. To monitor the displacement changes in these critical areas in real time, multi-point displacement sensors can be deployed to form a real-time displacement monitoring network. For example, sensors are installed at different levels and key structural positions of the roof and surrounding rock. These sensors can collect displacement data in real time, such as recording data every minute, so as to capture possible minor displacement changes. The collected displacement data needs to undergo data preprocessing, such as removing noise and filling in missing values, to ensure the accuracy and integrity of the data. Then, through feature extraction techniques, such as using time series analysis methods, key feature parameters that can reflect the displacement trends of the roof and surrounding rock are extracted, such as displacement rate, acceleration, etc. These displacement feature parameters are then input into a pre-constructed filling body stability assessment model. This model may be based on machine learning techniques and can learn the relationship between displacement and filling body stability through historical data. Through inference calculation, the model evaluates the stability level of the current filling body. If the model determines that the stability of the filling body is lower than the preset safety threshold, this will automatically trigger the generation process of an intelligent filling plan. In the process of generating the intelligent filling plan, reinforcement learning algorithms can be adopted. This algorithm learns through the interaction between the agent and the environment and continuously optimizes the filling sequence and time window. For example, the agent tries different filling strategies in a simulated environment and evaluates the effect of each strategy through a reward function, which may consider factors such as filling speed, cost, and stability after filling. The optimized filling plan will guide the operation of automated filling equipment, controlling the filling sequence and time of the paste to achieve precise and rapid filling of the goaf. This method not only improves the filling efficiency but also greatly enhances the overall safety of the coal mine by precisely controlling the filling process. Through this series of steps, the coal mine can achieve real-time monitoring of the displacement of the roof and surrounding rock in the goaf, timely adjust the filling strategy, optimize the filling effect, and thus ensure the safety and stability of the coal mine mining process. This method of comprehensively applying various technologies demonstrates how modern mining can solve traditional safety and efficiency problems through technological innovation.

[0116] Embodiment 2

[0117] The present invention also discloses an intelligent control system for coal mine mining based on paste filling, including: a construction module, a collection module, an optimization module, a control module, an acquisition module, a discrimination module, and a monitoring module;

[0118] The construction module is used to obtain the geological condition data of the underground environment of the coal mine and the filling area, and establish a three-dimensional digital model of paste filling according to the geological conditions of different filling areas and the goaf morphology. Among them, the three-dimensional digital model of paste filling includes paste ratio and pumping parameter attribute information;

[0119] The acquisition module is used to deploy multi-parameter sensors at key positions of the paste conveying pipeline to collect the rheological characteristic parameters of the paste in the pipeline in real time. Among them, the rheological characteristic parameters include viscosity, density, sand content, and particle distribution, forming a paste rheological characteristic data set;

[0120] The optimization module is used to dynamically optimize the paste ratio and pumping parameters according to the paste rheological characteristic data set collected in real time, combined with the pre-established three-dimensional digital model of paste filling, using machine learning algorithms to obtain the optimized paste ratio and pumping parameters;

[0121] The control module is used to input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system to perform real-time dynamic control on the paste preparation and conveying processes to ensure that the paste performance meets the filling requirements;

[0122] The acquisition module is used to deploy stress sensors and displacement sensors in the filling area to monitor the stress distribution and deformation conditions inside the filling body in real time, obtain the filling body stability data. At the same time, the density distribution data inside the filling body is obtained through ultrasonic detection technology;

[0123] The discrimination module is used to input the filling body stability data and the density distribution data into the paste filling effect evaluation model, use deep learning algorithms to comprehensively evaluate the filling effect, and judge whether the filling quality meets the requirements. If the filling quality is unqualified, the paste ratio and pumping parameters are re-optimized;

[0124] The monitoring module is used to deploy a multi-point displacement monitoring system in the underground coal mine environment to monitor the displacement changes of the roof and surrounding rocks of the goaf in real time. Combining the filling body stability evaluation results, using reinforcement learning algorithms to optimize the paste filling sequence and time window, and formulating an intelligent filling plan to maximize the integrity and timeliness of goaf filling and ensure coal mine mining safety.

[0125] The above-described embodiments are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent control method for coal mine mining based on paste filling, characterized in that, The steps include: Obtain the geological condition data of the underground coal mine environment and the filling area. Establish a three-dimensional digital model of paste filling for different geological conditions and goaf shapes in the filling area. Among them, the three-dimensional digital model of paste filling includes the paste mixing ratio and pumping parameter attribute information; Install multi-parameter sensors at key positions in the paste conveying pipeline to collect the rheological characteristic parameters of the paste in the pipeline in real time. Among them, the rheological characteristic parameters include viscosity, density, sand content, and particle distribution, forming a paste rheological characteristic data set; According to the paste rheological characteristic data set collected in real time, combined with the pre-established three-dimensional digital model of paste filling, use machine learning algorithms to dynamically optimize the paste mixing ratio and pumping parameters, and obtain the optimized paste mixing ratio and pumping parameters; Input the optimized paste mixing ratio and pumping parameters into the paste preparation system and pumping control system to perform real-time dynamic control on the paste preparation and conveying processes to ensure that the paste performance meets the filling requirements; Install stress sensors and displacement sensors in the filling area to monitor the stress distribution and deformation conditions inside the filling body in real time, obtain the filling body stability data. At the same time, obtain the density distribution data inside the filling body through ultrasonic detection technology; Input the filling body stability data and density distribution data into the paste filling effect evaluation model, and use deep learning algorithms to comprehensively evaluate the filling effect to judge whether the filling quality meets the requirements. If the filling quality is unqualified, re-optimize the paste mixing ratio and pumping parameters; Install a multi-point displacement monitoring system in the underground coal mine environment to monitor the displacement changes of the goaf roof and surrounding rock in real time. Combine the filling body stability evaluation results, and use reinforcement learning algorithms to optimize the paste filling sequence and time window, and formulate an intelligent filling plan to ensure the integrity and timeliness of goaf filling to the greatest extent and ensure the safety of coal mining.

2. The intelligent control method for coal mine mining based on paste filling according to claim 1, wherein Obtain the geological condition data of the underground coal mine environment and the filling area. Establishing a three-dimensional digital model of paste filling for different geological conditions and goaf shapes in the filling area includes: Obtain the geological condition data of the underground coal mine environment and the filling area, and preprocess the geological condition data to obtain a standardized geological condition data set; For the standardized geological condition data set, use a clustering algorithm for classification, divide different filling areas according to the clustering results, and obtain the geological condition characteristics of each filling area; Obtain the goaf shape data corresponding to each filling area, and construct a three-dimensional geometric model of the goaf through three-dimensional modeling to obtain the spatial shape information of the goaf; According to the geological condition characteristics of the filling area and the spatial shape information of the goaf, use an optimization algorithm to calculate the paste mixing ratio parameters to obtain the optimal paste mixing ratio parameters for each filling area; According to the optimal paste mixing ratio parameters and the spatial position of the filling area, use a fluid mechanics simulation algorithm to simulate the filling process of the paste in the goaf to obtain the best pumping parameters; Integrate the geological condition characteristics of the filling area, the spatial shape information of the goaf, the optimal paste mixing ratio parameters, and the best pumping parameters into a three-dimensional digital model to construct a three-dimensional visualization model of paste filling.

3. The intelligent control method for coal mine mining based on paste filling according to claim 1, wherein Based on the dataset of the rheological properties of the paste collected in real time, combined with the pre-established three-dimensional digital model of paste filling, the machine learning algorithm is used to dynamically optimize the paste ratio and pumping parameters. The optimized paste ratio and pumping parameters obtained include: Discretize the three-dimensional digital model of the paste filling area into a finite element mesh model; Input the rheological property data of the paste collected in real time into the machine learning algorithm, and establish the mapping relationship between the rheological properties of the paste and the paste ratio and pumping parameters through the support vector machine algorithm; Use the genetic algorithm to optimize the mapping relationship, and search for the optimal combination of paste ratio and pumping parameters through crossover and mutation operations; Input the optimized paste ratio and pumping parameters into the finite element mesh model, and simulate the flow and filling process of the paste in the filling area through the computational fluid dynamics equation; If the simulation results meet the construction requirements, apply the optimized paste ratio and pumping parameters to the actual construction.

4. The intelligent control method for coal mine mining based on paste filling according to claim 1, wherein Input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system, and conduct real-time dynamic control on the preparation and transportation process of the paste to ensure that the paste performance meets the filling requirements, including: Determine the dosage and proportion of each component according to the optimized paste ratio, and input the ratio parameters into the paste preparation system; Obtain the optimized pumping parameters and input the pumping parameters into the pumping control system; During the paste preparation process, adopt on-line monitoring technology to obtain the viscosity and density performance parameters of the paste in real time; Compare the real-time obtained paste performance parameters with the preset target performance parameters. If the deviation exceeds the threshold, dynamically adjust the paste ratio and pumping parameters; Establish a mapping relationship model between the paste ratio, pumping parameters and paste performance through the machine learning algorithm; During the paste transportation process, dynamically optimize the pumping parameters according to the real-time monitored pipeline pressure and flow parameters, combined with the mapping relationship model, to ensure the stability of the transportation process; Fill the paste to the specified position, and use visual inspection technology to judge whether the filling effect meets the requirements. If not, readjust the paste ratio and pumping parameters until the filling requirements are met.

5. The intelligent control method for coal mine mining based on paste filling according to claim 1, wherein, Input the filling body stability data and density distribution data into the paste filling effect evaluation model, and use the deep learning algorithm to comprehensively evaluate the filling effect to judge whether the filling quality meets the requirements. If the filling quality is unqualified, re-optimize the paste ratio and pumping parameters, including: Obtain the filling body stability data and density distribution data, and input them into the pre-constructed paste filling effect evaluation model; Adopt the convolutional neural network algorithm to extract features and perform representation learning on the input filling body stability data and density distribution data to obtain the comprehensive evaluation result of the filling effect; Judge whether the filling quality meets the preset qualified standard according to the evaluation result. If it meets, output the qualified result, otherwise trigger the optimization process of the paste ratio and pumping parameters; In the optimization process, search for the optimal combination of paste ratio and pumping parameters through the genetic algorithm, and input the optimized parameters into the paste filling effect evaluation model for re-evaluation; Apply the optimized paste ratio and pumping parameters to the actual filling operation to obtain new filling body stability data and density distribution data; and input them into the paste filling effect evaluation model to re-evaluate the filling effect and form a closed-loop optimization mechanism.

6. The intelligent control method for coal mine mining based on paste filling according to claim 1, characterized in that, Deploy a multi-point displacement monitoring system in the underground coal mine environment to monitor the displacement changes of the goaf roof and surrounding rock in real time. Combine the filling body stability evaluation results and use the reinforcement learning algorithm to optimize the paste filling sequence and time window, and formulate an intelligent filling plan to ensure the integrity and timeliness of goaf filling to the greatest extent and ensure coal mine safety, including: According to the characteristics of the coal mine environment, deploy multi-point displacement sensors at key positions on the goaf roof and surrounding rock to form a real-time displacement monitoring network; Obtain the displacement data collected by the sensors, and through data preprocessing and feature extraction, obtain key feature parameters that can reflect the displacement change trends of the goaf roof and surrounding rock; Input the displacement feature parameters into the pre-constructed filling body stability evaluation model, and through model inference calculation, judge the stability level of the current filling body; If the stability level of the filling body is lower than the preset threshold, trigger the generation process of the intelligent filling plan; Otherwise, keep the current filling plan unchanged and continue to monitor the displacement changes; Adopt the reinforcement learning algorithm, with the goal of improving the filling body stability, taking the filling sequence and time window as decision variables, and gradually optimize the filling plan through the interaction and exploration between the agent and the environment; During the training process of reinforcement learning, guide the agent to learn through the reward function, and while taking into account the filling integrity and timeliness, maximize the stability level of the filling body; Send the optimized intelligent filling plan to the automated filling equipment to control the filling sequence and time of the paste, and achieve precise and rapid filling of the goaf to ensure the safety and stability of the coal mine mining process.

7. An intelligent control system for coal mine exploitation based on paste filling, characterized in that, Including: A construction module, a collection module, an optimization module, a control module, an acquisition module, a discrimination module, and a monitoring module; The construction module is used to obtain the geological condition data of the underground coal mine environment and the filling area, and establish a three-dimensional digital model of paste filling for different geological conditions of the filling area and the goaf morphology. Among them, the three-dimensional digital model of paste filling contains paste ratio and pumping parameter attribute information; The collection module is used to deploy multi-parameter sensors at key positions of the paste conveying pipeline to collect the rheological characteristic parameters of the paste in the pipeline in real time. Among them, the rheological characteristic parameters include viscosity, density, sand content, and particle distribution, forming a paste rheological characteristic data set; The optimization module is used to dynamically optimize the paste ratio and pumping parameters according to the real-time collected paste rheological characteristic data set, combined with the pre-established three-dimensional digital model of paste filling, using machine learning algorithms, and obtain the optimized paste ratio and pumping parameters; The control module is used to input the optimized paste ratio and pumping parameters into the paste preparation system and the pumping control system to perform real-time dynamic control on the paste preparation and conveying process to ensure that the paste performance meets the filling requirements. The acquisition module is used to deploy stress sensors and displacement sensors in the filling area, monitor the stress distribution and deformation conditions inside the filling body in real time, obtain the stability data of the filling body, and at the same time, obtain the density distribution data inside the filling body through ultrasonic detection technology; The discrimination module is used to input the stability data of the filling body and the density distribution data into the paste filling effect evaluation model, comprehensively evaluate the filling effect by using deep learning algorithms, judge whether the filling quality meets the requirements, and if the filling quality is unqualified, re-optimize the paste ratio and pumping parameters; The monitoring module is used to deploy a multi-point displacement monitoring system in the underground coal mine environment, monitor the displacement changes of the roof and surrounding rocks of the goaf in real time, combine the evaluation results of the filling body stability, optimize the paste filling sequence and time window by using reinforcement learning algorithms, formulate an intelligent filling plan, maximize the integrity and timeliness of the goaf filling, and ensure the safety of coal mine mining.

Images