Fully mechanized coal mining face dynamic cutting simulation method based on black box-grey box-white box coal seam model

Abstract

The invention discloses a fully mechanized coal mining face dynamic cutting simulation method based on a black box-grey box-white box coal seam model. Constructing a virtual geological simulation module, and generating a virtual geological model data file; on the basis of virtual geologic model data, a coal rock interface is constructed through simulated drilling sampling and a Kriging interpolation algorithm, an exploration error coefficient is introduced, and black box, grey box and white box coal seam error models are constructed; importing a three-dimensional model and an error model of the fully-mechanized coal mining equipment into the simulation platform, and constructing an equipment-coal seam coupling model; calling real-time pose change data of a coal mining machine roller, and adaptively generating a cutting top plate and a cutting bottom plate; and utilizing a pre-trained intelligent model to predict the cutting force in real time, dynamically updating the state of the virtual exploration coal seam error model based on a prediction result and a real-time cutting track, and autonomously regulating and controlling the operation parameters of the coal mining machine. According to the method, a three-level error model is constructed, a dynamic model system of the geological cognition process can be truly reflected, and deep fusion of exploration errors and cutting simulation is achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent coal mining technology, and more specifically, to a dynamic cutting simulation method for fully mechanized mining faces based on a black-box-grey-box-white-box coal seam model. Background Technology

[0002] Intelligent coal mining is a core technological support for the transformation and upgrading of the coal industry. With the deepening of the new generation of technological revolution and industrial transformation, intelligent technologies, represented by digital twins, are accelerating the intelligentization process of coal mines. As the core area of ​​coal mine production, the safe and efficient operation of the fully mechanized mining face highly depends on the dynamic adaptation between geological conditions and mining equipment. In the digital twin system of the fully mechanized mining face, the construction of traditional geological models mainly relies on geological exploration and sensing technologies to obtain relevant information; at the same time, the virtual model of fully mechanized mining equipment based on the geological model is also gradually evolving from static display to dynamic interaction and physical simulation, continuously improving the system's intelligence level and decision support capabilities. These advancements have expanded the application boundaries of digital twin technology in the collaborative simulation of coal seams and equipment.

[0003] In the prior art, invention patent CN120407885A discloses a method for fusing sensing data from fully mechanized mining face equipment with geological exploration data. Specifically, the method involves: Step 1, performing coordinate translation transformation on the currently established three-dimensional coal seam geological model to establish a matrix of geological data starting from the origin coordinates; Step 2, acquiring sensing data from the fully mechanized mining face equipment; Step 3, establishing a real-time two-dimensional mining model of the coal mining machine; Step 4, calculating the current advance speed z of the coal mining machine; Step 5, forming the mined coal seam model data based on the current advance speed z of the coal mining machine and the model from the fully mechanized mining face equipment sensing data; Step 6, fusing the mined coal seam model data and the translated geological model using a wavelet transform Kalman filter algorithm to form complete geological model data. This invention solves the problems of high cost and limited number of boreholes in existing technologies that rely on directional drilling and borehole geophysical exploration to establish models.

[0004] In the prior art, invention patent CN116291434A discloses a high-precision geological model coal cutting navigation method and device. The method includes: during the cutting time of the current cutter of the coal mining machine, based on existing geological data prior to the start of cutting by the current cutter, using a three-dimensional implicit geological modeling algorithm to generate a high-precision geological model of the coal mining face. The high-precision geological model includes information on the coal-rock boundary location of the unmined area of ​​the coal mining face; determining the cutting path of the next cutter based on the high-precision geological model; and transmitting the cutting path of the next cut to the fully mechanized mining equipment control system, so that the control system can control the coal mining machine to perform the next cut according to the cutting path. Therefore, throughout the entire coal mining cutting process, a high-precision geological model can be dynamically generated during each cutter, and the cutting path of the next cut can be determined based on the high-precision geological model, serving as the basis for the coal mining machine to carry out cutting.

[0005] Among existing patents, invention patent CN113887111A discloses a virtual integrated testing method for geology, coal seams, and equipment in fully mechanized mining faces. This method includes the following models: a virtual geological model, a coal seam exploration and modeling simulation model, a coal seam model dynamic correction simulation model, a theoretical workspace simulation model, and a coal mining machine pose inversion workspace simulation model. Through quantitative comparative analysis between these models, the methods for coal seam exploration and modeling, dynamic correction of coal seam models, coal mining machine cutting techniques or methods, and coal mining machine pose inversion workspace modeling can be tested, providing an effective solution to the systematic testing problems of fully mechanized mining faces.

[0006] The aforementioned patents have made significant progress in geological modeling, equipment modeling, and coupled simulation, but they have also exposed several key problems.

[0007] (1) The geological models constructed by the above methods are all macroscopic models, ignoring the local heterogeneity of coal seams and lacking the expression of microscopic characteristics such as coal type, coal hardness, coal brittleness, and structural uniformity. They cannot provide sufficient data support for algorithms such as intelligent height adjustment and adaptive truncation.

[0008] (2) The geological models constructed by the above methods are all static models, lacking a dynamic update mechanism. They cannot correct coal seam parameters and roof and floor morphology in real time as mining progresses, making it difficult to support the real-time perception requirements of the working face. They are only suitable for offline simulation and macroscopic demonstration.

[0009] (3) Although the invention patent with publication number CN113887111A mentions “dynamic correction of coal seam model”, its core method is to directly correct the digital elevation matrix by introducing new survey data, thereby improving the accuracy of the model. It lacks a theoretical framework to describe the qualitative change in the understanding of coal seams and cannot describe the dynamic evolution of coal seams in three-dimensional space with spatial location and time (mining process).

[0010] (4) The above methods focus on geometric shape construction and do not integrate physical models such as cutting mechanics and coal and rock crushing, which makes it difficult for the simulation results to reflect the actual load and energy consumption characteristics of the equipment.

[0011] In summary, although existing research has made significant progress in geological modeling, equipment modeling, and their coupled simulation, several key problems remain in practical applications. First, the concepts of geological model and coal seam model are often used interchangeably, with unclear definitions. Existing research frequently uses "geological model" and "coal seam model" interchangeably when describing the operation mechanism of fully mechanized mining equipment, without clearly distinguishing their connotations, boundaries, and applicable scope. Second, current geological modeling largely relies on real exploration data from specific areas, and the constructed models can only support the simulation operation of fully mechanized mining equipment in a single geological scenario. Key geological characteristic parameters in the model, such as dip angle, undulation morphology, coal and rock hardness, and bedding structure, are often statically fixed, making it difficult to adapt to changes in complex geological conditions of multiple types and scales. Finally, existing research generally fails to consider the impact of inherent uncertainties and modeling errors in geological exploration on the coal cutting and working face advancement processes. This static, singular, and one-sided modeling approach severely restricts the generalization ability and prediction accuracy of equipment simulation modules under different geological environments. Summary of the Invention

[0012] The purpose of this invention is to overcome the shortcomings of the prior art and provide a dynamic cutting simulation method for fully mechanized mining faces based on a black-box-grey-box-white-box coal seam model that can dynamically and realistically simulate the cutting process of fully mechanized mining faces.

[0013] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model includes: (1) Constructing a virtual geological simulation module Define the structure and storage mechanism of simulation data containing location point cloud data and coal and rock attribute parameters, and construct a virtual coal and rock boundary surface based on the tensor product surface method; simulate macroscopic data including the length, strike, dip angle range and coal thickness of the coal seam, add microscopic data containing virtual coal and rock attribute parameters at the current geological location point, and generate a virtual geological model data file; (2) Constructing a virtual coal seam error module Based on virtual geological model data, coal-rock interfaces are constructed by simulating borehole sampling and Kriging interpolation algorithm. Exploration error coefficients are introduced to construct black-box, gray-box, and white-box coal seam error models, forming a virtual exploration coal seam error model that can simulate the geological survey results obtained by exploration equipment in a real environment. (3) Constructing a virtual truncation dynamic module Import the 3D model of the fully mechanized mining equipment and the virtual exploration coal seam error model into the simulation platform to construct the equipment-coal seam coupling model; during equipment operation, establish a three-level coordinate system based on the spatial origin; call the real-time pose change data of the coal mining machine drum in the 3D software to adaptively generate the cutting roof and floor plates; (4) Construct a dynamic truncation response module The cutting force is predicted in real time using a pre-trained intelligent model. The state of the virtual exploration coal seam error model is dynamically updated based on the prediction results and the real-time cutting trajectory, and the operating parameters of the coal mining machine are autonomously adjusted.

[0014] In a preferred embodiment, the coal and rock attribute parameters in the virtual geological simulation module include four key parameters: coal and rock type, coal and rock hardness, coal and rock brittleness, and structural uniformity; the standardized data structure associates spatial coordinates (X,Y,Z) with the coal and rock attribute parameters.

[0015] As a preferred implementation, in constructing the virtual geological simulation module, the tensor product surface method specifically involves: using Catmull-Rom curves to generate interpolation points for the horizontal and vertical curves, and superimposing the coordinates of the two sets of interpolation points to form a virtual geological coal-rock boundary surface.

[0016] As a preferred implementation, in the virtual exploration coal seam error module, "boreholes" are sampled at fixed intervals in the horizontal and vertical directions in the virtual geological model data to obtain simulated geological data at the point location, including coal and rock type, coal and rock hardness, coal and rock brittleness, and structural uniformity. Based on the simulated geological data in the vertical direction at the point, two coal and rock boundary points, including the roof and intermediate coal seam and the floor and intermediate coal seam, are determined. The information of the coal and rock boundary points at each "borehole" is recorded to obtain rough data on the coal and rock interface.

[0017] As a preferred embodiment, the three error levels—black box, gray box, and white box—are defined as follows in the virtual coal seam error module: The black-box coal seam error model corresponds to a coal seam region that is far from the fully mechanized mining face, relies on macroscopic geological laws for deduction, and has high uncertainty. The gray box coal seam error model corresponds to a coal seam area near the fully mechanized mining face that has uncertainty after being corrected using the sensing information of the fully mechanized mining equipment, but has a high degree of confidence. The white-box coal seam error model corresponds to a coal seam area that has been cut, has clear geological information, and is error-free.

[0018] As a preferred implementation, in the construction of the virtual cutting dynamic module, a Mesh collider component is added to the virtual exploration coal seam error model to achieve accurate collision between the equipment and the coal seam model and terrain fit; at the same time, the coordinate position of the coal seam model in the cutting direction in space is obtained, and the three-axis position and rotation attitude of the hydraulic support and scraper conveyor are matched with the coordinate position of the coal seam model to achieve close coupling between the equipment and the coal seam.

[0019] As a preferred implementation, in the construction of the virtual cutting dynamic module, the starting point of the coal seam is taken as the origin of the first-level coordinate system, and the direction of the three-axis coordinates is the same as the direction of the spatial coordinate system; when the three fully mechanized mining machines move, the center point of the middle trough of the first scraper conveyor is taken as the origin of the second-level coordinate system, and the direction of the three-axis coordinates is the same as the direction of the spatial coordinate system; the origin of the third-level coordinate system is the origin of each piece of equipment.

[0020] In a preferred implementation, the virtual cutting dynamic module utilizes real-time pose data of the coal mining machine drum combined with the drum radius to calculate the cutting trajectory points, and dynamically constructs a cutting top and bottom plate mesh using a Mesh component; the formula for calculating the cutting trajectory points is: ; in, X new 、Y new 、Z new These represent the X, Y, and Z coordinate values ​​of the bottom point or the highest point of the coal mining machine; X g-s 、 Y g-s 、Z g-s These represent the X, Y, and Z coordinate values ​​of the coal mining machine drum, respectively. r g This indicates the radius of the coal mining machine drum. It is "+" when the point is selected as the high point of the coal mining machine and "-" when it is the bottom point of the coal mining machine.

[0021] In a preferred embodiment, the pre-trained intelligent model in the dynamic cutting response module is an intelligent agent trained based on a deep reinforcement learning algorithm. It is trained offline using a large amount of experimental or high-fidelity physical simulation data and is used to establish a nonlinear mapping relationship between coal and rock properties, coal mining machine conditions and cutting force.

[0022] As a preferred implementation, in the dynamic cutting response module, when dynamically updating the state of the virtual exploration coal seam error model, the model state of the cut area is updated from "black box" or "gray box" to "gray box" or "white box" model with fully known geological information based on the real-time cutting trajectory of the coal mining machine. Using the geological information newly revealed during the cutting process, the coal-rock interface and attribute prediction of the "black box" and "gray box" areas ahead are assimilated and corrected, thereby gradually improving the prediction accuracy of the unmined area model.

[0023] Compared with existing technologies, the dynamic cutting simulation method for fully mechanized mining faces based on the black-box-grey-box-white-box coal seam model of this invention has the following beneficial effects: (1) A black-box-gray-white-box graded coal seam error model was constructed, clarifying the internal logic and dynamic evolution relationship of the three in the digital twin scenario, and constructing a dynamic model system that can truly reflect the geological cognition process, thus solving the problem of rigid and singular geological models in traditional simulation.

[0024] (2) It has achieved efficient and diversified generation of virtual geological models. Through a unified standardized data structure and unconditional simulation, a large number of "geological twins" that conform to the statistical laws of real geology can be quickly constructed, providing an unprecedented and highly challenging "virtual test field" for fully mechanized mining equipment.

[0025] (3) A deep integration of exploration error and cutting simulation was achieved. The three-level error model was coupled with the high-fidelity equipment model through the physics engine and Mesh dynamic generation technology, which accurately simulated the actual cutting shape of the coal mining machine drum and the changes in equipment load under uncertain geological conditions, so that the simulation results moved from "morphological similarity" to "mechanical reality".

[0026] (4) Intelligent response and autonomous equipment control in the cutting process have been realized. A data-driven rapid prediction model for cutting force is constructed through deep reinforcement learning, and the real-time mechanical response is dynamically linked with coal seam properties and equipment operating conditions, forming a closed-loop intelligent control logic of "perception-prediction-control". This upgrades the simulation system from passive simulation to an intelligent agent with autonomous decision-making and optimization capabilities, providing an efficient platform for the verification of intelligent algorithms in real working faces. Attached Figure Description

[0027] Figure 1 This is a block diagram illustrating the construction method of the simulation method of the present invention; Figure 2 This refers to the virtual geological simulation modeling module described in this invention. Figure 3 This is the virtual exploration coal seam error modeling module described in this invention; Figure 4 This refers to the virtual truncation dynamic modeling module described in this invention; Figure 5 This refers to the dynamic truncation response module described in this invention. Detailed Implementation

[0028] The basic concept of this invention is to introduce a black-box-grey-box-white-box coal seam model to achieve a quantitative expression and dynamic evolution of the uncertainty of geological cognition, and to integrate physical models and intelligent algorithms to improve the realism of simulation and decision support capabilities.

[0029] Based on the above basic concepts, this invention provides a typical embodiment of a dynamic cutting simulation method for fully mechanized mining faces based on a black-box-grey-box-white-box coal seam model, such as... Figure 1 As shown, the system includes a virtual geological simulation module, a virtual coal seam exploration error module, a virtual cutting dynamic module, and a dynamic cutting response module. Corresponding to the above method, the fully mechanized longwall face dynamic cutting simulation system based on the black-box-grey-box-white-box coal seam model constructed in this embodiment includes a virtual geological simulation modeling module, a virtual coal seam exploration error modeling module, a virtual cutting dynamic modeling module, and a dynamic cutting response module.

[0030] Virtual geological simulation modeling module The virtual geological simulation modeling module, during the simulation of a fully mechanized mining face, defines the structure and storage mechanism of the simulation data and constructs the coal-rock interface using the tensor product surface method. Simultaneously, it simulates macroscopic data such as basic coal seam parameters including length, strike, dip angle range, and coal thickness, and adds microscopic data such as virtual coal-rock attribute parameters at the current geological location, achieving more accurate geological simulation.

[0031] like Figure 2 As shown, the specific process of constructing the virtual geological simulation modeling module is as follows: Step 101: Simulate Data Structures and Storage Mechanisms The simulated geological data includes location point cloud data and its coal and rock attribute parameters, such as coal and rock type, coal and rock hardness, coal and rock brittleness, and structural uniformity. Given the wide coverage and large amount of data of the fully mechanized mining face, this paper chooses the CSV file format for storage. Before standardizing the data structure, the types and units of various data are unified, and the reading and writing rules of the CSV file are pre-defined. Four key parameters, namely coal and rock type, coal and rock hardness, coal and rock brittleness, and structural uniformity, are selected and associated with their corresponding spatial locations X, Y, and Z to construct the simulated data structure system shown in Equation (1).

[0032] ;

[0033] Each formula consists of eight comma-separated numbers: the first number is the geological data sequence number; the second, third, and fourth numbers represent the starting coordinates (X, Y, Z) of the point in space, with a spatial sampling interval of 0.1 meters; the fifth number represents the coal-rock type, i.e., the average coal-rock mixing ratio in the area where the current point is located, with a value range of (0,1), where a larger value indicates a higher proportion of rock and vice versa; the sixth number represents the coal-rock hardness, using the average Protodyakonov hardness coefficient of the area; the seventh number represents the coal-rock brittleness, with a value range of (0,1), where a larger value indicates greater brittleness; and the eighth number represents the structural homogeneity, also with a value range of (0,1), where a larger value indicates a more homogeneous structure in the area.

[0034] Step 102: Construction of the Virtual Coal-Rock Boundary Surface In the construction of virtual geological models, accurately determining key parameters such as the strike, dip angle, and undulation angle of the geological formation, as well as identifying the coal-rock interfaces, can effectively ensure the macroscopic morphology of the model. The virtual geological model is divided into three layers: the roof layer, the intermediate coal seam, and the floor layer. Two coal-rock interfaces are located at the roof layer and the intermediate coal seam, and the intermediate coal seam and the floor layer, respectively.

[0035] A complete surface is constructed using the tensor product surface method. The Catmull-Rom curve method is employed, with the number of lateral control points set and their positions adjusted based on simulated geological parameters to ensure the curve smoothly passes through all control points, thus constructing the Catmull-Rom curve. A linear interpolation algorithm is used to obtain lateral interpolation points at fixed lateral intervals within the lateral curve. A longitudinal Catmull-Rom curve is constructed using the same method, yielding two sets of independent curve interpolation points. Finally, the coordinates of these two curve interpolation points are superimposed to obtain the tensor product parametric surface, forming a virtual geological coal-rock interface.

[0036] Step 103: Assignment of simulated geological coal and rock property parameters A geological random field was generated using the exponential covariance model from geostatistics. A three-dimensional anisotropic random field was constructed using the open-source Python library GSTools. Range parameters were set to control the lateral, longitudinal, and height directions to reflect the simulated geological variations, and the nugget effect was introduced to represent sampling error and microscopic non-uniformity. Simultaneously, based on the virtual coal-rock boundary surface constructed in step 102, a random field following a Gaussian distribution was generated. Through linear mapping and physical constraints, the random field was transformed into a three-dimensional spatial distribution field simulating coal-rock type, hardness, brittleness, and structural homogeneity. The generated parameters containing simulated geological coal-rock attributes were sorted sequentially in the order of X, then Y, then Z, and the data was filled according to the simulated data structure described in step 101 to form a complete simulated geological data file.

[0037] Virtual Coal Seam Error Modeling Module The virtual coal seam error modeling module, based on the simulated geological data file generated in step 103, first simulates the "drilling" process to construct a rough virtual coal-rock interface, and then refines it using the Kriging interpolation algorithm to obtain a more detailed virtual coal-rock interface. Exploration error coefficients are introduced to construct exploration black-box, gray-box, and white-box coal seam models, forming a virtual coal seam error model capable of simulating real-world geological conditions based on exploration equipment.

[0038] like Figure 3 As shown, the specific process of constructing the virtual exploration coal seam error modeling module is as follows: Step 201: Construct a rough virtual coal-rock interface During the operation of the fully mechanized mining face, based on the three-dimensional spatial distribution field described in step 103, a virtual "borehole" sampling method is used. "Boreholes" are sampled at fixed intervals in the horizontal and vertical directions within the data file established in step 103 to obtain geological information at each point, including simulated geological data such as coal and rock type, coal and rock hardness, coal and rock brittleness, and structural homogeneity. Based on the simulated geological data in the vertical direction at this point, two coal-rock boundary points are determined, including those between the roof and the intermediate coal seam, and between the floor and the intermediate coal seam. The information of the coal-rock boundary points at each "borehole" is recorded to obtain rough data on the coal-rock interface.

[0039] Step 202: Reconstruction of the coal-rock interface based on Kriging interpolation Sequential Gauss-Schlikin interpolation was used to spatially reconstruct the virtual coal-rock interface. A variogram model was constructed to quantify the spatial correlation of geological variables, ensuring that the interpolation results met the requirements of unbiasedness and optimality at known borehole sampling points. The reconstruction process preserved the spatial structural characteristics of the original simulated geological body through covariance matrix optimization, while introducing an exploration accuracy-error transfer function to dynamically map the spatial error distribution of the coal-rock interface elevation under different borehole densities. Finally, based on the reconstructed continuous surface model, a high-precision 3D point cloud dataset was generated with a grid resolution of 0.1m × 0.1m.

[0040] Step 203: Construct exploration black-box, gray-box, and white-box coal seam error models. The aforementioned black-box, gray-box, and white-box coal seam error models correspond to different levels of coal seam error under different exploration accuracies: (1) Black Box Coal Seam Error Model: Based on the virtual coal-rock interface, this model describes the coal seam error in areas far from the fully mechanized mining face, limited by detection methods and cover density, and far from the cutting plane. This area relies on macroscopic geological laws for deduction, resulting in significant uncertainty and providing only trend guidance. The coal seam area far from the fully mechanized mining face is defined as the black box coal seam error model.

[0041] (2) Gray Box Coal Seam Error Model: Based on the black box coal seam error model, this model is obtained by correcting the black box coal seam error model using the sensing information from the fully mechanized mining equipment and sensing detection methods. The prediction in this area is based on existing geological trends and interpolation results, and while there is some uncertainty, it still possesses a high degree of confidence. The coal seam area closest to the fully mechanized mining face with smaller errors is defined as the gray box coal seam error model. (3) White Box Coal Seam Error Model: Due to the layout of the longwall mining face, the initial position of the equipment, and the exposed coal seam, a transparent geological model is obtained. The geological information of this area is clear, the measured data is sufficient, and the model accuracy is the highest, which can be regarded as "completely knowable". The coal seam area that has been cut and has no error is defined as the white box coal seam error model.

[0042] Virtual Cutting Dynamic Modeling Module The aforementioned virtual cutting dynamic modeling module refers to the construction of fully mechanized mining equipment with high fidelity in simulation software.

[0043] like Figure 4 As shown, the specific process of constructing the virtual truncation dynamic modeling module is as follows: Step 301: Construction of the fully mechanized mining equipment-coal seam coupling model Based on the 3D model of the fully mechanized mining equipment, the format was converted and the final model was imported into Unity3D. In Unity3D, a physics engine component was added to the virtual equipment model, and C# code was written to realize the virtual operation of the fully mechanized mining equipment. Next, the virtual exploration coal seam error model constructed in step 203 was imported into Unity3D. A Mesh collider component was added to the virtual exploration coal seam error model to achieve accurate collision and terrain fit between the equipment and the coal seam model. Simultaneously, the spatial coordinates of the coal seam model along the cutting direction were obtained, and the three-axis positions (X / Y / Z) and rotational attitudes (Euler angles or quaternions) of the hydraulic support and scraper conveyor were matched with the coordinates of the coal seam model to achieve tight coupling between the equipment and the coal seam.

[0044] Step 302: Construction of the three-dimensional coordinate system During equipment operation, to ensure consistency between equipment data and coal seam data, a three-level coordinate system based on a spatial origin is established. The origin of the first-level coordinate system is the starting point of the coal seam, with the three-axis coordinate directions being the same as those of the spatial coordinate system. When the three fully mechanized mining machines are in motion, the origin of the second-level coordinate system is the center point of the middle trough of the first scraper conveyor, with the three-axis coordinate directions being the same as those of the spatial coordinate system. The origin of the third-level coordinate system is the origin of each piece of equipment.

[0045] Step 303: Adaptive generation of top and bottom plates. To simulate the actual shape of the roof and floor of the coal seam after cutting, the real-time pose change data of the coal mining machine drum is called in Unity3d. Considering the influence of the drum radius, the cutting trajectory of the roof and floor formed by the coal mining machine is calculated. The calculation formula is shown in Equation (2): ;

[0046] in, X new 、Y new 、Z new These represent the X, Y, and Z coordinate values ​​of the bottom point or the highest point of the coal mining machine; X g-s 、 Y g-s 、Z g-s These represent the X, Y, and Z coordinate values ​​of the coal mining machine drum, respectively. r g This indicates the radius of the coal mining machine drum. It is "+" when the point is selected as the high point of the coal mining machine and "-" when it is the bottom point of the coal mining machine.

[0047] To achieve adaptive generation of the cutting roof and floor plates by the coal mining machine drum, not only are the three-dimensional coordinates of the current position point required, but also the rotational attitude of the drum at the current position point needs to be acquired simultaneously. Since the drum generation of the cutting roof and floor plates does not need to consider the pitch angle, only the yaw and roll angles need to be considered. Based on the transmission chain relationship from the coal mining machine to the drum, the roll and yaw angles of the drum are consistent with those of the coal mining machine. On this basis, combined with the real-time cutting depth of the current coal mining machine, the cutting roof and floor plate mesh network is dynamically constructed using the Mesh component in Unity3d.

[0048] Dynamic truncation response module The dynamic cutting response module simulates the physical response of the interaction between the coal mining machine and the coal seam, and drives the equipment to achieve adaptive control. This module predicts the cutting force in real time using a data-driven method, and dynamically updates the model and optimizes the equipment's actions accordingly, thereby completing a simulation closed loop from perception and decision-making to execution.

[0049] like Figure 5 As shown, the specific process of constructing the dynamic truncation response module is as follows: Step 401: Intelligent Prediction of Cutting Force Based on extensive experimental or high-fidelity physical simulation data, a deep reinforcement learning agent is trained offline to master the complex nonlinear mapping relationship between coal and rock properties, coal mining machine operating conditions, and cutting force. In real-time simulation, the system inputs the coal and rock properties at the current drum and the coal mining machine operating parameters into the trained agent, enabling it to quickly predict the instantaneous cutting resistance trend. Simultaneously, the prediction results are calibrated using real-time collision detection information from the physics engine, ultimately outputting cutting force data that is both predictive and realistic, providing immediate basis for intelligent control.

[0050] Step 402: Dynamic Update of Coal Seam Model Based on the real-time cutting trajectory of the coal mining machine, the model status of the cut area is updated from a "black box" or "gray box" to a "gray box" or "white box" model with fully known geological information. In Unity3D, the visual effects of coal wall advancement and goaf formation are reproduced in real time by dynamically modifying or removing the mesh in the corresponding area. Simultaneously, using the newly revealed geological information during the cutting process, data assimilation and correction are performed on the coal-rock interface and attribute predictions in the preceding "black box" and "gray box" areas, gradually improving the prediction accuracy of the unmined area model.

[0051] Step 403: Equipment Autonomous Adaptive Control When the real-time cutting force exceeds the safety threshold, the system immediately triggers a speed reduction protection mechanism, reducing the traction speed of the coal mining machine according to a preset gradient. When the cutting force remains below the lower limit of the high-efficiency range, the system autonomously increases the traction speed to the optimal operating point. The drum height control loop is synchronously linked to the same cutting force signal: by analyzing the characteristics of sudden changes in cutting force and integrating the roof and floor lithology distribution data in the "white box" coal seam model, the system automatically calculates the distance of the drum offset from the optimal coal-rock interface and generates a height compensation command. The two control loops are linked through a collaborative optimization algorithm to output the final coal mining machine control command.

[0052] Step 404: Full Process Visualization and Performance Analysis In the Unity3D scene, "black box," "gray box," and "white box" areas are highlighted with different colors, and key data such as cutting force are displayed in real time graphically, making the simulation process intuitive and transparent. The system synchronously records all time-series data, such as pose and cutting force. After the simulation is completed, an analysis report is automatically generated, calculating key performance indicators such as cutting efficiency and model prediction accuracy, providing quantitative basis for optimizing the mining process.

[0053] The comprehensive application of the above four models can achieve the following applications: virtual modeling and simulation verification for digital twin systems of fully mechanized mining faces. Based on the virtual geological modeling module, virtual exploration coal seam error modeling module, and virtual cutting modeling module, combined with the Unity3d simulation platform and related modeling processes, the specific process is as follows: (1) Virtual geological simulation modeling: The coal-rock interface was constructed based on the Catmull-Rom curve and tensor product surface method, and a three-dimensional anisotropic coal-rock property random field was generated using GSTools. The CSV data containing coal-rock type, hardness and other attributes was imported into Unity3d to generate structured virtual coal-rock blocks.

[0054] (2) Virtual exploration coal seam error modeling: Virtual borehole sampling is achieved by simulating the exploration process to obtain coal-rock boundary data, and Kriging interpolation is used to reconstruct the coal-rock interface. Based on the working face advance distance, three levels of error regions, namely black box, gray box and white box, are divided to establish a dynamically updated exploration coal seam error model.

[0055] (3) Virtual cutting dynamic modeling: Import equipment models such as coal mining machine and hydraulic support into Unity3d and configure physical properties. Implement equipment motion control based on C# script, and dynamically generate cutting top and bottom plate mesh by real-time acquisition of drum trajectory points to complete the coupling simulation of equipment and coal seam.

[0056] (4) Dynamic cutting response simulation: Using a pre-trained deep reinforcement learning agent, the cutting force trend is predicted based on real-time working parameters. The traction speed and drum height of the coal mining machine are automatically adjusted based on the prediction results to achieve intelligent closed-loop control of "perception-prediction-control".

[0057] The scope of protection claimed by this invention is not limited to the specific embodiments described above. For those skilled in the art, this invention can have various modifications and alterations. Any modifications, improvements, and equivalent substitutions made within the concept and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model, characterized in that, include: (1) Constructing a virtual geological simulation module Define the structure and storage mechanism of simulation data containing location point cloud data and coal and rock attribute parameters, and construct a virtual coal and rock boundary surface based on the tensor product surface method; simulate macroscopic data including the length, strike, dip angle range and coal thickness of the coal seam, add microscopic data containing virtual coal and rock attribute parameters at the current geological location point, and generate a virtual geological model data file; (2) Constructing a virtual coal seam error module Based on virtual geological model data, coal-rock interfaces are constructed by simulating borehole sampling and Kriging interpolation algorithm. Exploration error coefficients are introduced to construct black-box, gray-box, and white-box coal seam error models, forming a virtual exploration coal seam error model that can simulate the geological survey results obtained by exploration equipment in a real environment. (3) Constructing a virtual truncation dynamic module Import the 3D model of the fully mechanized mining equipment and the virtual exploration coal seam error model into the simulation platform to construct the equipment-coal seam coupling model; during equipment operation, establish a three-level coordinate system based on the spatial origin; call the real-time pose change data of the coal mining machine drum in the 3D software to adaptively generate the cutting roof and floor plates; (4) Construct a dynamic truncation response module The cutting force is predicted in real time using a pre-trained intelligent model. The state of the virtual exploration coal seam error model is dynamically updated based on the prediction results and the real-time cutting trajectory, and the operating parameters of the coal mining machine are autonomously adjusted.

2. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 1, characterized in that: In the construction of the virtual geological simulation module, the coal and rock attribute parameters include four key parameters: coal and rock type, coal and rock hardness, coal and rock brittleness, and structural homogeneity; the standardized data structure associates spatial coordinates (X, Y, Z) with the coal and rock attribute parameters.

3. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 2, characterized in that: In constructing the virtual geological simulation module, the tensor product surface method specifically involves: using Catmull-Rom curves to generate interpolation points for both the horizontal and vertical directions, and then superimposing the coordinates of the two sets of interpolation points to form a virtual geological coal-rock boundary surface.

4. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 1 or 3, characterized in that: In constructing the virtual coal seam error module, "boreholes" are sampled at fixed intervals in the horizontal and vertical directions in the virtual geological model data to obtain simulated geological data at the location of the point, including coal and rock type, coal and rock hardness, coal and rock brittleness, and structural homogeneity. Based on the simulated geological data in the vertical direction of the point, two coal-rock boundary points, including the roof and intermediate coal seam and the floor and intermediate coal seam, are determined. The information of the coal-rock boundary points at each "borehole" is recorded to obtain rough data of the coal-rock interface.

5. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 4, characterized in that: In constructing the virtual coal seam error module, the three error levels—black box, gray box, and white box—are defined as follows: The black-box coal seam error model corresponds to a coal seam region that is far from the fully mechanized mining face, relies on macroscopic geological laws for deduction, and has high uncertainty. The gray box coal seam error model corresponds to a coal seam area near the fully mechanized mining face that has uncertainty after being corrected using the sensing information of the fully mechanized mining equipment, but has a high degree of confidence. The white-box coal seam error model corresponds to a coal seam area that has been cut, has clear geological information, and is error-free.

6. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 1 or 5, characterized in that: In constructing the virtual cutting dynamic module, a Mesh collider component is added to the virtual exploration coal seam error model to achieve accurate collision between the equipment and the coal seam model and terrain fit; at the same time, the spatial coordinate position of the coal seam model in the cutting direction is obtained, and the three-axis position and rotation attitude of the hydraulic support and scraper conveyor are matched with the coordinate position of the coal seam model to achieve tight coupling between the equipment and the coal seam.

7. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 6, characterized in that: In the construction of the virtual cutting dynamic module, the starting point of the coal seam is taken as the origin of the first-level coordinate system, and the direction of the three-axis coordinates is the same as the direction of the spatial coordinate system; when the three fully mechanized mining machines move, the center point of the middle trough of the first scraper conveyor is taken as the origin of the second-level coordinate system, and the direction of the three-axis coordinates is the same as the direction of the spatial coordinate system; the origin of the third-level coordinate system is the origin of each piece of equipment.

8. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 7, characterized in that: In constructing the virtual cutting dynamic module, the real-time pose data of the coal mining machine drum is called in combination with the drum radius to calculate the cutting trajectory points, and the Mesh component is used to dynamically construct the cutting top and bottom plate mesh; the formula for calculating the cutting trajectory points is: ； in, X new 、Y new 、Z new These represent the X, Y, and Z coordinate values ​​of the bottom point or the highest point of the coal mining machine; X g-s 、Y g-s 、Z g-s These represent the X, Y, and Z coordinate values ​​of the coal mining machine drum, respectively. r g This indicates the radius of the coal mining machine drum. It is "+" when the point is selected as the high point of the coal mining machine and "-" when it is the bottom point of the coal mining machine.

9. The dynamic cutting simulation method for fully mechanized mining faces based on a black-box-grey-box-white-box coal seam model according to claim 1 or 8, characterized in that: In the construction of the dynamic cutting response module, the pre-trained intelligent model is an intelligent agent trained based on a deep reinforcement learning algorithm. It is trained offline using a large amount of experimental or high-fidelity physical simulation data and is used to establish a nonlinear mapping relationship between coal and rock properties, coal mining machine conditions and cutting force.

10. The dynamic cutting simulation method for fully mechanized longwall mining faces based on a black-box-grey-box-white-box coal seam model according to claim 9, characterized in that: In the dynamic cutting response module, when dynamically updating the state of the virtual exploration coal seam error model, the model state of the cut area is updated from "black box" or "gray box" to "gray box" or "white box" model with fully known geological information based on the real-time cutting trajectory of the coal mining machine. Using the geological information newly revealed during the cutting process, the coal-rock interface and attribute prediction of the "black box" and "gray box" areas ahead are assimilated and corrected, gradually improving the prediction accuracy of the unmined area model.

Images