A steeply inclined thin coal seam unmanned mining method

Abstract

The present application provides a kind of steeply inclined thin coal seam unmanned mining method, belong to coal mining technical field, the present application constructs three-dimensional geological model by geological radar and drilling data, adopts layered stoping method to divide coal seam into 0.8 to 1.2 meter thick horizontal layer, and utilizes multi-feature clustering function to optimize layering. Adopt honeycomb hexagonal grid to divide working face, improve path planning accuracy;Optimal mining path is designed by considering factors such as coal seam thickness, dip angle and coal quality;Multi-dimensional weighted graph structure is constructed to form mining path navigation map;Develop one-way zoned stoping strategy to improve mining efficiency;Real-time path dynamic planning is realized, and the path is updated according to real-time data such as gas content;Coal mining robot adjusts mining parameters accordingly, effectively solves the technical problems of low precision of unmanned mining path planning under complex geological conditions of steeply inclined thin coal seam and difficulty in adapting to coal seam changes.

Description

Technical Field

[0001] This invention belongs to the field of coal mining technology, and more specifically, relates to a method for unmanned coal mining of steeply inclined thin coal seams. Background Technology

[0002] Steeply inclined thin coal seam mining is an important branch of coal mining. Traditional mining techniques mainly rely on manual labor or semi-mechanized equipment to deploy working faces along the strike of the coal seam. Conventional mining methods include inclined stratification, horizontal stratification, and working face deployment along the strike, with manual operation of mining machinery to complete cutting and mining operations. With the development of intelligent coal mining, mining robots have begun to be used in some mining areas with better coal seam conditions, performing basic mining tasks through preset paths.

[0003] However, traditional steeply inclined thin coal seam mining technology has many shortcomings in practical applications. First, coal mining path planning usually uses a simple rectangular grid to divide the working face, which is difficult to accurately adapt to irregular coal seam boundaries. Second, path algorithms mostly use fixed patterns and cannot be dynamically optimized based on multi-dimensional characteristics such as coal seam thickness, dip angle, and coal quality. Furthermore, existing technologies lack a response mechanism to real-time geological changes during coal mining, making it difficult for mining operations to adapt to sudden changes in coal seams.

[0004] Especially in steeply inclined thin coal seams with an inclination greater than 45 degrees, due to complex and variable geological conditions, uneven coal seam thickness, and poor roof stability, the existing coal mining path planning technology has low accuracy and is difficult to dynamically adjust according to real-time changes in the coal seam, resulting in problems such as low mining efficiency, high safety risks, and insufficient resource recovery rate. There is an urgent need for an unmanned coal mining technology solution that can accurately plan the coal mining path and dynamically adapt to changes in the coal seam. Summary of the Invention

[0005] In view of this, the present invention provides a method for unmanned coal mining of steeply inclined thin coal seams, which can solve the technical problems of low accuracy of unmanned coal mining path planning and difficulty in adapting to changes in coal seams under complex geological conditions in the prior art.

[0006] This invention is implemented as follows: It provides a method for unmanned coal mining in steeply inclined thin coal seams, comprising: constructing a three-dimensional geological model of the thin coal seam to determine the coal seam dip angle distribution matrix, coal seam thickness data matrix, and fault distribution matrix; dividing the thin coal seam into horizontal layers and establishing a three-dimensional coal mining grid model; using a honeycomb hexagonal grid to divide the working face and form a basic network for calculating the coal mining path; designing the main travel path of the coal mining robot based on the Bellman-Ford algorithm, and calculating the optimal coal mining path by comprehensively considering the coal seam thickness data matrix, coal seam dip angle distribution matrix, and coal quality distribution matrix; constructing a coal mining path navigation map to form a multi-dimensional weighted graph structure; developing a unidirectional zoned mining strategy; and realizing dynamic path planning and coal mining parameter adjustment for the coal mining robot.

[0007] Specifically, the step of dividing the thin coal seam into horizontal layers and establishing a three-dimensional coal mining grid model involves designing working faces along the strike, using a layered mining method to divide the steeply inclined thin coal seam into multiple horizontal layers according to its thickness, with each layer having a thickness of 0.8 to 1.2 meters, using a multi-feature clustering function to optimize the coal seam layering, and establishing a three-dimensional coal mining grid model.

[0008] The step of using a honeycomb hexagonal grid to divide the working face into a basic network for coal mining path calculation specifically involves dividing the working face into multiple regular hexagonal units with a side length of 0.6 meters to form a basic network for coal mining path calculation, thereby improving the accuracy of path planning. The honeycomb hexagonal grid is a spatial representation method that divides the space of the coal mining working face into a set of regular hexagonal units with a side length of 0.6 meters. Each hexagonal unit shares a boundary with its six adjacent units. Compared with the traditional rectangular grid, the hexagonal grid has a constant distance between adjacent units in any direction, reducing the cumulative error of directional deviation and improving the smoothness of the path.

[0009] The specific steps of developing a unidirectional zone mining strategy involve dividing the working face into multiple mining zones and setting a unidirectional travel path for each mining zone to avoid the coal mining robot repeatedly passing through the same area and improve coal mining efficiency.

[0010] Specifically, the step of realizing real-time path dynamic planning involves dividing the boundary point set based on the layer thickness parameter matrix and the horizontal layer of the working face to achieve real-time path dynamic planning. The coal mining robot collects the gas content distribution matrix according to the real-time detection data of the coal seam, periodically updates the local area grid weights, and recalculates the optimal path to adapt to changes in the coal seam.

[0011] Specifically, the coal mining robot adjusts its mining parameters to achieve precise cutting and safe and efficient mining by adjusting the mining parameters based on the coal seam thickness data matrix, coal seam dip angle distribution matrix, layer thickness parameter matrix, and boundary point set of the working face horizontal layer. This allows the robot to control the drum height and cutting depth, thereby achieving precise cutting and safe and efficient mining of steeply inclined thin coal seams.

[0012] Specifically, the layered mining method involves dividing the steeply inclined thin coal seam into multiple parallel layers with a thickness of 0.8 to 1.2 meters in the horizontal direction. The coal mining robot completes the mining operation independently in each parallel layer. After completing the mining of one parallel layer, the coal mining robot enters the next parallel layer. This method is suitable for steeply inclined coal seams with an inclination angle greater than 45 degrees. By controlling the range of roof movement through layering, the risk of roof collapse is reduced, and the coal recovery rate is improved.

[0013] The Bellman-Ford algorithm is a graph algorithm for finding the shortest path from a single source. It calculates the shortest path from the starting point to all vertices by performing multiple rounds of relaxation operations on all edges in the graph. In the weighted graph structure designed in coal mining path planning, the edge weights integrate the coal seam thickness coefficient, coal seam dip angle coefficient, and coal quality coefficient. By minimizing the comprehensive cost, it achieves the optimal utilization of resources in the coal mining path.

[0014] The multi-feature clustering function is used to scientifically divide the coal mining face based on multiple physical characteristics of the coal seam. The input includes a coal seam thickness data matrix, a coal seam dip angle distribution matrix, a coal quality distribution matrix, a fault distribution matrix, and a gas content distribution matrix. The output is a layer thickness parameter matrix and a set of boundary points for dividing the horizontal layers of the working face.

[0015] Specifically, the unidirectional zone mining strategy is a coal mining operation method designed for the characteristics of steeply inclined thin coal seams. A complete working face is divided into multiple mining areas of similar size according to the variation characteristics of the coal seam thickness data matrix and the coal seam dip angle distribution matrix. A fixed mining direction is set in each mining area. The coal mining robot completes the mining in a single direction in one mining area before entering the next mining area, avoiding the coal mining robot from moving back and forth in the working face and reducing non-productive movement time.

[0016] This invention achieves precise coal mining path planning in complex, steeply inclined thin coal seam environments by constructing a three-dimensional geological model of thin coal seams, employing a layered mining method and dividing the working face into honeycomb hexagonal grids, and combining the Bellman-Ford algorithm to design the path for the coal mining robot. The method optimizes the coal seam layer by layer using a multi-feature clustering function, comprehensively considering multi-dimensional data such as coal seam thickness, dip angle, and coal quality. It constructs a multi-dimensional weighted graph structure with weight factors including coal seam thickness coefficient, coal seam dip angle coefficient, and coal quality coefficient, overcoming the shortcomings of traditional techniques in handling irregular boundaries and multi-feature fusion. In particular, through real-time dynamic path planning and adaptive adjustment of mining parameters, it solves the problem of difficulty in coping with sudden changes in coal seams during mining, enabling the coal mining robot to update its path planning based on real-time detection data and adapt to changes in the coal seam.

[0017] This invention effectively solves the technical problems of low accuracy in unmanned coal mining path planning and difficulty in adapting to changes in coal seams under complex geological conditions of steeply inclined thin coal seams. It improves coal mining efficiency and resource recovery rate, reduces the risk of roof collapse, and provides technical support for intelligent unmanned mining of steeply inclined thin coal seams. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention.

[0019] Figure 2 This is a schematic diagram of the overall structure of the unmanned coal mining system for steeply inclined thin coal seams in Example 2.

[0020] Figure 3 This is a schematic diagram of the coal mining robot structure in Example 2.

[0021] Figure 4 This is a schematic diagram of the honeycomb hexagonal grid path planning in Example 2. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0023] like Figure 1 The diagram shown is a flowchart of a method for unmanned coal mining of steeply inclined thin coal seams provided by the present invention. This method includes the following steps:

[0024] S01. Construct a three-dimensional geological model of thin coal seam using ground-penetrating radar and drilling data, and determine the coal seam dip angle distribution matrix, coal seam thickness data matrix and fault distribution matrix to provide basic data for coal mining robot path planning;

[0025] S02. The working face is arranged along the strike and the layered mining method is adopted. The steeply inclined thin coal seam is divided into multiple horizontal layers according to its thickness, with each layer having a thickness of 0.8 to 1.2 meters. The coal seam is optimized by multi-feature clustering function and a three-dimensional coal mining grid model is established.

[0026] S03. A honeycomb hexagonal grid is used to divide the working face into multiple regular hexagonal units with a side length of 0.6 meters, forming a basic network for coal mining path calculation and improving the accuracy of path planning.

[0027] S04. Design the main travel path of the coal mining robot based on the Bellman-Ford algorithm. Taking into account the coal seam thickness data matrix, coal seam dip angle distribution matrix and coal quality distribution matrix, calculate the optimal coal mining path and generate the path node coordinate sequence.

[0028] S05. Construct a coal mining path navigation map, establish weighted connections based on the honeycomb hexagonal grid nodes, and the weight factors include coal seam thickness coefficient, coal seam dip angle coefficient and coal quality coefficient to form a multi-dimensional weighted graph structure.

[0029] S06. Develop a one-way zone mining strategy, divide the working face into multiple mining zones, set a one-way travel path for each mining zone, avoid the coal mining robot repeatedly passing through the same area, and improve coal mining efficiency.

[0030] S07. Based on the layer thickness parameter matrix and the boundary point set of the working face horizontal layer, real-time path dynamic planning is realized. The coal mining robot collects the gas content distribution matrix according to the real-time coal seam detection data, periodically updates the local area grid weight, recalculates the optimal path, and adapts to coal seam changes.

[0031] S08. The coal mining robot adjusts the coal mining parameters and controls the drum height and cutting depth based on the coal seam thickness data matrix, coal seam dip angle distribution matrix, layer thickness parameter matrix and the boundary point set of the working face horizontal layer division, so as to achieve precise cutting and safe and efficient mining of steeply inclined thin coal seams.

[0032] The layered mining method involves dividing a steeply inclined thin coal seam into multiple parallel layers with a thickness of 0.8 to 1.2 meters in the horizontal direction. The coal mining robot completes the mining operation independently in each parallel layer. After completing the mining of one parallel layer, the coal mining robot enters the next parallel layer. It is suitable for steeply inclined coal seams with an inclination angle greater than 45 degrees. By controlling the range of roof movement through layering, the risk of roof collapse is reduced and the coal recovery rate is improved.

[0033] Among them, the honeycomb hexagonal grid is a spatial representation method that divides the coal mining face space into a set of regular hexagonal cells with a side length of 0.6 meters. Each hexagonal cell shares a boundary with its six neighboring cells. Compared with the traditional rectangular grid, the hexagonal grid has a constant distance between adjacent cells in any direction, which reduces the cumulative error of directional deviation, improves the smoothness of the path, and the hexagonal grid shows higher adaptability when dealing with irregular boundaries.

[0034] The Bellman-Ford algorithm is a graph algorithm for finding the shortest path from a single source. It calculates the shortest path from the starting point to all vertices by performing multiple rounds of relaxation operations on all edges in the graph. In the weighted graph structure designed for coal mining path planning, the edge weights integrate the coal seam thickness coefficient, coal seam dip angle coefficient, and coal quality coefficient. By minimizing the comprehensive cost, it achieves the optimal utilization of resources in the coal mining path. Furthermore, the Bellman-Ford algorithm can handle negative weight edges and is suitable for considering the changes in coal mining value caused by differences in coal quality.

[0035] Among them, the unidirectional zone mining strategy is a coal mining operation method designed for the characteristics of steeply inclined thin coal seams. It divides a complete working face into multiple mining areas of similar size according to the variation characteristics of the coal seam thickness data matrix and the coal seam dip angle distribution matrix. A fixed mining direction is set in each mining area. The coal mining robot completes the mining in a single direction in one mining area before entering the next mining area, avoiding the coal mining robot from moving back and forth in the working face and reducing non-productive movement time. At the same time, by optimizing the boundary connection design of the mining area, the continuous advancement of the coal face is ensured.

[0036] In step S02, the multi-feature clustering function is used to scientifically divide the coal mining face based on multiple physical properties of the coal seam. The inputs include a coal seam thickness data matrix, a coal seam dip angle distribution matrix, a coal quality distribution matrix, a fault distribution matrix, and a gas content distribution matrix. The outputs are a layer thickness parameter matrix and a set of horizontal layer division boundary points for the working face. The coal seam thickness data matrix, the coal seam dip angle distribution matrix, and the fault distribution matrix are derived from the ground-penetrating radar and drilling data obtained in step S01. The coal quality distribution matrix is ​​derived from coal seam coal quality testing data. The gas content distribution matrix is ​​derived from coal seam gas content monitoring data. The layer thickness parameter matrix and the set of horizontal layer division boundary points for the working face are used to guide the real-time path dynamic planning in step S07. The multi-feature clustering function specifically includes the following steps:

[0037] Step 1: Construct a coal seam feature vector matrix for the coal mining face space. Each element of the coal seam feature vector matrix includes the coal seam thickness value, coal seam dip angle value, coal quality grade value, fault distance value, and gas content value of the sampling point. The coal seam thickness value comes from the coal seam thickness data matrix, the coal seam dip angle value comes from the coal seam dip angle distribution matrix, the coal quality grade value comes from the coal quality distribution matrix and represents a comprehensive evaluation of the coal's calorific value and ash content, the fault distance value comes from the fault distribution matrix and represents the Euclidean distance from the sampling point to the nearest fault, and the gas content value comes from the gas content distribution matrix.

[0038] Step 2: For the constructed coal seam feature vector matrix, the principal component analysis method is used to reduce the dimensionality, extract the feature combinations that mainly affect coal mining safety and efficiency, reduce the computational complexity, and obtain the dimensionality-reduced feature data.

[0039] Step 3: Use the K-means clustering algorithm to perform cluster analysis on the dimensionality-reduced feature data. The initial cluster centers are selected using the maximum distance principle to ensure that the feature differences between different categories are maximized and the feature differences within the same category are minimized, thus obtaining the clustering results.

[0040] Step 4: Based on the clustering results and the spatial location information of the coal seam, determine the precise thickness and boundary of each horizontal layer, generate the layer thickness parameter matrix and the set of boundary points for dividing the horizontal layers of the working face, and provide a basis for subsequent honeycomb hexagonal grid division.

[0041] Among them, the coal seam thickness data matrix is ​​a two-dimensional or three-dimensional data structure that represents the thickness of coal seams at different locations in the coal mining face. It is obtained through ground-penetrating radar detection and drilling sampling, and is used to quantitatively describe the thickness variation of coal seams in space, providing a basis for coal mining robot path planning and cutting depth control.

[0042] Among them, the coal seam dip angle distribution matrix is ​​a two-dimensional or three-dimensional data structure that represents the dip angle of coal seams at different locations in the coal mining face. It is obtained through geological exploration and dip angle measurement instruments and is used to quantitatively describe the changes in the dip degree of coal seams in space, providing a basis for the attitude control and stability assurance of coal mining robots.

[0043] The coal quality distribution matrix is ​​a two-dimensional or three-dimensional data structure that represents the coal quality at different locations in the coal mining face. It is obtained through coal sample analysis and includes coal quality indicators such as calorific value, ash content, and sulfur content. It is used to quantitatively describe the quality changes of the coal seam in space and provide a basis for decision-making on prioritizing the mining of high-quality coal seams.

[0044] Among them, the fault distribution matrix is ​​a two-dimensional or three-dimensional data structure that represents the location and scale of geological faults in a coal mining face. It is obtained through geological exploration and seismic wave detection and is used to quantitatively describe the fault distribution of coal seams in space, providing safety assurance for coal mining robots to avoid dangerous areas.

[0045] Among them, the gas content distribution matrix is ​​a two-dimensional or three-dimensional data structure that represents the gas concentration at different locations in the coal mining face. It is obtained through gas detectors and gas sampling analysis and is used to quantitatively describe the changes in gas content in the coal seam in space, providing a basis for gas safety monitoring and explosion-proof measures for coal mining robots.

[0046] The layer thickness parameter matrix is ​​a two-dimensional data structure that represents the thickness values ​​of each layer after multi-feature clustering optimization. It contains the starting and ending depth information of each layer in the working face and is used to guide the division of the working range of the coal mining robot in the vertical direction, ensuring the scientific and safe nature of layered coal mining.

[0047] Specifically, the set of boundary points for dividing the horizontal layers of the working face is a spatial point set representing the contour of each horizontal layer boundary after multi-feature clustering optimization. It contains a sequence of three-dimensional coordinate points of each layer boundary and is used to accurately describe the layer boundaries of irregular coal seams, guide the coal mining robot to accurately identify the layer boundary positions, and avoid mining beyond the boundaries.

[0048] The specific implementation methods of the above steps are described in detail below.

[0049] The specific implementation of step S01 involves using a high-resolution ground-penetrating radar to scan the coal seam area, acquiring electromagnetic wave reflection data of the coal seam. The scanning frequency is set to 500MHz–2GHz, the radar detection depth is controlled within the range of 50–100m, and the scan line spacing is maintained at 1–2m. Simultaneously, boreholes are deployed with a spacing of 20–30m, and core samples are collected and analyzed to extract information on coal seam thickness, dip angle, and fault location. The ground-penetrating radar reflection data and borehole core data are input into 3D geological modeling software, and the Kriging interpolation algorithm is applied to spatially interpolate the coal seam parameters between sampling points, generating a continuous 3D coal seam model. Based on the 3D model, the coal seam dip angle distribution matrix is ​​extracted. The matrix resolution is 0.5m × 0.5m, recording the coal seam dip angle value at each grid point of the working face. Areas with dip angles greater than 45 degrees are marked as steeply dipping areas. A coal seam thickness data matrix is ​​extracted, with the same resolution as the dip angle distribution matrix, recording the coal seam thickness value at each grid point of the working face. Areas with a thickness less than 2m are marked as thin coal seam areas. Fault locations and scales are identified, generating a fault distribution matrix. This matrix records the distance from each grid point to the nearest fault, with areas less than 5m away marked as fault-prone areas. Finally, the coal seam dip angle distribution matrix, coal seam thickness data matrix, and fault distribution matrix are integrated into a unified geological database, providing fundamental data support for subsequent path planning of the coal mining robot. The purpose of this step is to construct a high-precision three-dimensional geological model through accurate geological data acquisition and processing, providing reliable geological parameters for unmanned coal mining operations.

[0050] The specific implementation of step S02 is based on the three-dimensional geological model obtained in S01. The working face is arranged along the strike direction of the coal seam, and its length is determined according to the coal seam stability, typically controlled within the range of 50–100m. A layered mining method is adopted, dividing the coal seam vertically into multiple horizontal layers, with each layer's thickness set within the range of 0.8–1.2m. The selection of the layer thickness threshold is based on the maximum cutting height of the mining robot and the roof stability requirements. A multi-feature clustering function combining K-means clustering and fuzzy C-means clustering is used to process the coal seam feature data. Input parameters include the coal seam thickness data matrix and the coal seam dip angle matrix. The distribution matrix, coal quality matrix, fault distribution matrix, and gas content distribution matrix are used to iteratively calculate and cluster similar feature regions into one class. The iteration terminates when the change in the cluster center position is less than 0.05m and the intra-class variance is less than a preset threshold of 0.1. Based on the clustering results, horizontal stratification is optimized to ensure relatively uniform coal seam characteristics within each stratum, generating a stratification thickness parameter matrix containing the start and end depths of each stratum. A three-dimensional coal mining grid model is then established using the stratified data, with a grid resolution of 0.2m × 0.2m × 0.2m. The model includes coal seam geometry, physical parameters, and stratification boundary information. The purpose of this step is to scientifically divide coal mining strata, providing a spatial partitioning framework for subsequent path planning, and improving mining efficiency and safety through multi-feature clustering optimization of the stratification scheme.

[0051] The specific implementation of step S03 involves dividing each horizontal working face into layers using a honeycomb hexagonal grid based on the established three-dimensional coal mining grid model. This transforms the working face space into a planar network composed of regular hexagonal units with a side length of 0.6m. The selection of the hexagonal side length considers the working radius and positioning accuracy of the coal mining robot. When the positioning accuracy of the coal mining robot is higher than 0.1m, the hexagonal side length can be set to 0.6m. The hexagonal grid is generated using the Thiessen polygon method. First, the set of boundary points of the working face is determined. Then, grid nodes are evenly distributed inside the working face, with the node spacing maintained at about 1m. Thiessen polygons are constructed based on these nodes, and the polygons are approximated as regular hexagons. For irregular hexagonal units at the working face boundary, an adaptive grid adjustment algorithm is used for optimization to ensure that the area change of the grid unit at the boundary does not exceed 20% of the area of ​​the standard hexagon. Each hexagonal grid node is assigned a unique identifier using a three-digit code. The first digit represents the layer number, and the last two digits represent the planar position number. A hexagonal grid adjacency table is established to record the identifiers of the six adjacent units of each hexagonal unit, providing topological relationship data for path planning. The purpose of this step is to establish a spatial grid system suitable for the navigation of coal mining robots. Compared with traditional rectangular grids, hexagonal grids provide a more uniform orientation distribution and smaller cumulative error, which is beneficial to improving the accuracy of path planning.

[0052] The specific implementation of step S04 is based on the honeycomb hexagonal grid constructed in S03, and the Bellman-Ford algorithm is applied to design the main travel path of the coal mining robot; a weighted directed graph is constructed, where the vertices are the set of hexagonal grid nodes and the edges are the set of connecting edges between nodes; the edge weight of each edge in the graph is defined as a comprehensive cost function, which considers multiple factors: including the edge thickness cost, the edge dip angle cost, and the edge coal quality cost, which are calculated by weighting the three factors according to proportional coefficients, which are set to 0.4, 0.3, and 0.3 respectively, and can be dynamically adjusted according to actual mining needs; the thickness cost is determined by the edge's corresponding position. The ratio of the coal seam thickness to the maximum coal seam thickness at the working face is obtained through transformation; the dip angle cost is calculated by the ratio of the coal seam dip angle at the corresponding edge position to the maximum dip angle at the working face; the coal quality cost is obtained by the ratio of the coal quality score at the corresponding edge position to the highest coal quality score at the working face through transformation; the Bellman-Ford algorithm is applied to calculate the shortest path from the starting point to the ending point, and a relaxation operation is performed on all edges in each iteration, with the total number of iterations set to the number of grid nodes minus 1; the algorithm terminates when the shortest path is stable and there are no negative weight cycles; based on the calculation results, a sequence of node coordinates for the main travel path of the coal mining robot is generated, with coordinate accuracy controlled at the centimeter level. The purpose of this step is to optimize the coal mining path based on multiple coal seam parameters, solve the shortest path problem of a weighted graph using the Bellman-Ford algorithm, and provide the optimal travel route for the coal mining robot.

[0053] The specific implementation of step S05 is based on the honeycomb hexagonal grid in S03 and the edge weight calculation method in S04 to construct a complete coal mining path navigation map. First, the vertex set of the navigation map is determined, with the center point of each hexagonal grid serving as a navigation vertex. Vertex attributes include three-dimensional spatial coordinates, the layer number to which it belongs, and coal seam characteristic parameters. Connections between vertices are established; for adjacent hexagonal cell center points, bidirectional connection edges are established, and each edge is assigned a weight value. The edge weight factor is calculated using a comprehensive evaluation method, including three main coefficients: a coal seam thickness coefficient, with a value ranging from 0.2 to 0.5, where the coal seam thickness is close to the optimal working value of the coal mining robot. At a height of 1m, the smaller the coefficient value; the coal seam dip angle coefficient ranges from 0.2 to 0.5, and the smaller the coefficient value is when the coal seam dip angle is close to the optimal working dip angle of the coal mining robot (30 degrees); the coal quality coefficient ranges from 0.1 to 0.4, and the smaller the coefficient value is when the coal quality grade is higher; the three coefficients satisfy the normalization condition and their sum is 1; the edge weight is finally calculated by the weighted sum of the three coefficients and the standardized distance values ​​of the corresponding features; a multi-dimensional weighted graph structure is constructed, including a vertex set, an edge set, and a weight set; an adjacency list is used to store the graph structure to improve graph search efficiency; the navigation graph is simplified by removing edges with weights exceeding the threshold of 3.0 to reduce computational complexity. The purpose of this step is to establish a complete navigation graph structure that considers the influence of multiple factors, providing accurate spatial navigation information for the coal mining robot and supporting path planning in complex environments.

[0054] The specific implementation of step S06 is based on the geological data obtained in S01 and the navigation map constructed in S05, developing a unidirectional zoned mining strategy. First, the working face is divided based on the coal seam dip angle distribution matrix. When the dip angle difference between adjacent areas exceeds 15 degrees, a zone boundary is considered. The zone division is refined based on the coal seam thickness data matrix. When the thickness difference between adjacent areas exceeds 0.5m, the zone boundary is adjusted. Taking into account the location of geological faults, the areas on both sides of the fault are forcibly divided into different mining areas. The working face is typically divided into 3-5 mining areas, with each mining area maintaining a similar area and an error controlled within 15%. The optimal mining direction is determined for each mining area, and principal component analysis is used to determine the main characteristics of the coal seam. The mining direction is set perpendicular to the main direction to minimize changes in coal seam characteristics during mining. The mining directions of adjacent mining areas should be kept consistent or the angle between them should be less than 45 degrees to reduce adjustment ranges during transitions. A transition zone with a width of 2-3 meters is designed at the boundary of each mining area, using a smooth curve to connect the mining paths of different areas, with a radius of curvature not less than 1.5 times the minimum turning radius of the mining robot. A mining area priority ranking is established based on the coal quality distribution matrix and gas content distribution matrix, prioritizing mining areas with high coal quality and low gas content. Optimal transition paths between mining areas are designed to ensure that the mining robot can safely and efficiently enter the next mining area after completing one. The purpose of this step is to avoid redundant movement of the mining robot, reduce non-productive time, and improve mining efficiency by rationally dividing mining areas and planning mining directions.

[0055] The specific implementation of step S07 is based on the layer thickness parameter matrix generated in S02 and the set of boundary points for dividing the horizontal layers of the working face, to develop a real-time path dynamic planning strategy; the coal mining robot is equipped with a forward detection radar with a detection distance of 3-5m and a scanning frequency of not less than 1Hz to collect information on the coal seam in front in real time; a gas sensor array is installed with a sampling frequency of not less than 0.5Hz and a measurement accuracy better than 0.01% to construct a local gas content distribution matrix; the sliding window method is used to update the local area grid data, with the window size set to 5m×5m and the window sliding step size of 1m, only updating grid points where the difference between the detection data and the original data exceeds a threshold, the thickness difference threshold is set to 0.2m, the dip angle difference threshold is set to 5 degrees, and the gas content difference threshold is set to 0.1%; Based on updated local grid data, the D-Star algorithm is used for dynamic path planning. The D-Star algorithm adds dynamic updating capabilities to the Bellman-Ford algorithm, triggering replanning when environmental changes exceed preset thresholds. Replanning trigger conditions include: a difference of more than 0.3m between the detected coal seam thickness and the original data, a difference of more than 10 degrees between the detected coal seam dip angle and the original data, and a detected gas content exceeding the safety threshold of 1%. The replanning process employs a local search strategy, recalculating the path only in the affected area, with the search range limited to a radius of 10m. A new optimal path segment is calculated to replace the corresponding part of the original path. Path smoothing is performed using cubic spline interpolation to control the path curvature change rate to be less than 0.2 / m², ensuring smooth movement of the coal mining robot. The purpose of this step is to achieve dynamic adjustment of the coal mining path, enabling the coal mining robot to respond promptly to coal seam changes based on real-time detection data, ensuring the safety and adaptability of coal mining operations.

[0056] The specific implementation of step S08 involves the coal mining robot adjusting its mining parameters based on multi-source coal seam data to achieve precise cutting and safe mining. The robot is equipped with a laser rangefinder and an inclination sensor to measure its relative position and attitude with the coal seam in real time, with a ranging accuracy better than 0.05m and an inclination measurement accuracy better than 0.5 degrees. The drum height is adjusted based on the coal seam thickness data matrix, with a setting range of 0.6–1.5m and an adjustment accuracy of 0.01m. When the measured coal seam thickness is less than 0.7m, the drum height is set to the coal seam thickness minus 0.1m to preserve the integrity of the roof and floor. The robot's working posture is adjusted based on the coal seam inclination distribution matrix, maintaining the drum axis parallel to the coal seam direction through a hydraulic support system, with an inclination adaptation range of 0–60 degrees. When the coal seam inclination is greater than 50 degrees, an additional stabilizing support system is activated to prevent the robot from tipping over. The cutting depth is controlled based on the layer thickness parameter matrix, with a cutting depth setting range of… The cutting depth is 0.3–0.8m. As the working face approaches the layer boundary, the cutting depth gradually decreases, with a minimum adjustable depth of 0.1m, ensuring that it does not exceed the layer boundary. A set of boundary points is used to divide the working face into horizontal layers to guide boundary processing. When the distance to the layer boundary is less than 1m, the cutting speed is reduced to 50% of the normal speed to improve boundary processing accuracy. A real-time adjustment strategy for coal mining parameters is established, using a fuzzy control algorithm. The input variables are coal seam thickness, dip angle, and distance from the boundary; the output variables are drum height, cutting depth, and cutting speed. The fuzzy rule base contains 30–50 rules covering various coal seam conditions. A safety monitoring system is set up. When the gas concentration exceeds 1.5%, the cutting speed is automatically reduced; when it exceeds 2%, coal mining operations are suspended. A roof monitoring system is established, using acoustic detection technology to monitor the roof condition. When the roof integrity index is below 0.8, the cutting depth is reduced; when it is below 0.6, coal mining in the current area is stopped. The purpose of this step is to achieve precise control of coal mining parameters, adapt to the complex conditions of steeply inclined thin coal seams, ensure coal mining quality and safety, and improve coal recovery rate.

[0057] Specifically, the principle of this invention is as follows: The core technical principle of this invention lies in the organic combination of geological modeling, spatial grid division, path algorithm optimization, and dynamic adjustment mechanism to construct a complete unmanned coal mining path planning and execution system for steeply inclined thin coal seams. First, a three-dimensional geological model of the thin coal seam is constructed using geological exploration data to provide basic data support for path planning; second, a multi-feature clustering function is introduced to scientifically stratify complex coal seams, and the stratification parameters are optimized according to the physical characteristics of the coal seam to make the stratification more consistent with the actual geological characteristics.

[0058] In terms of spatial representation, a novel honeycomb hexagonal grid is adopted instead of the traditional rectangular grid. This structure maintains a constant distance between adjacent cells in any direction, reducing the cumulative error of directional deviations and improving path smoothness. It also exhibits greater adaptability when handling irregular boundaries. Regarding the path algorithm design, the Bellman-Ford algorithm can handle negative weighted edges and is suitable for considering the changes in coal mining value caused by differences in coal quality. It constructs a weight function by integrating multi-dimensional factors such as coal seam thickness, dip angle, and coal quality to calculate the globally optimal coal mining path.

[0059] To address coal seam variations, this invention employs a real-time path dynamic planning mechanism. The coal mining robot collects data such as gas content in real time, periodically updates the local area grid weights, and recalculates the optimal path. Simultaneously, a unidirectional zoned mining strategy is introduced to prevent the coal mining robot from moving back and forth within the working face, reducing non-productive movement time.

[0060] The organic combination of these technical principles enables the present invention to achieve precise planning and dynamic adjustment of coal mining paths under complex, steeply inclined, thin coal seam conditions, thereby solving the core technical problem of low path planning accuracy and difficulty in adapting to coal seam changes in the prior art, and has a solid foundation at both the theoretical and practical levels.

[0061] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0062] The specific implementation of step S01 involves using a high-resolution ground-penetrating radar to scan the coal seam area, acquiring electromagnetic wave reflection data of the coal seam. The scanning frequency is set to 500MHz–2GHz, the radar detection depth is controlled within the range of 50–100m, and the scan line spacing is maintained at 1–2m. Simultaneously, boreholes are deployed with a spacing of 20–30m, and core samples are collected and analyzed to extract information on coal seam thickness, dip angle, and fault location. The ground-penetrating radar reflection data and borehole core data are input into 3D geological modeling software, and the Kriging interpolation algorithm is applied to spatially interpolate the coal seam parameters between sampling points, generating a continuous 3D coal seam model. The Kriging interpolation algorithm is based on random field theory and variogram calculations, and its estimated value Z... * The formula for calculating (x0) at position x0 is:

[0063]

[0064] In the formula, Z * (x0) is the estimated value at position x0; Z(x) i ) represents the position x i Known observations at λ; i For the position x i The weighting coefficients; n is the number of observation points involved in the estimation.

[0065] Weighting coefficient λ i The following Kriging equations were obtained by solving them:

[0066]

[0067] In the formula, γ(x) i x j ) represents the position x i and x j The variogram values ​​between; μ is the Lagrange multiplier; γ(x) i x0) is the position x i The variogram value between the target estimated position x0 and the target estimated position x0.

[0068] Based on the 3D model, the coal seam dip angle distribution matrix A is extracted, with a resolution of 0.5m × 0.5m. The dip angle value of the coal seam at each grid point of the working face is recorded. Areas with dip angles greater than 45 degrees are marked as steeply dipping areas. The coal seam dip angle distribution matrix A is represented as follows:

[0069] A = [a ij ] m×n ;

[0070] In the formula, a ij The value represents the dip angle of the coal seam at coordinates (i, j) on the working face, in degrees; m and n represent the number of grids on the working face in the x and y directions, respectively.

[0071] Extract the coal seam thickness data matrix H, with the same resolution as the dip angle distribution matrix. Record the coal seam thickness value at each grid point of the working face. Areas with a thickness less than 2m are marked as thin coal seams. The coal seam thickness data matrix H is represented as follows:

[0072] H = [h] ij ] m×n ;

[0073] In the formula, h ij This represents the coal seam thickness at coordinates (i, j) on the working face, in meters; m and n are the same as the values ​​in the dip angle distribution matrix.

[0074] Identify the location and scale of faults, and generate a fault distribution matrix F. The matrix records the distance from each grid point to the nearest fault, with areas less than 5m away marked as fault danger zones. The fault distribution matrix f is represented as:

[0075] F = [f ij ] m×n ;

[0076] In the formula, f ijThis represents the distance from the working face coordinates (i, j) to the nearest fault, in meters; m and n are the same as the values ​​in the dip angle distribution matrix.

[0077] Finally, the coal seam dip angle distribution matrix A, coal seam thickness data matrix H, and fault distribution matrix F are integrated into a unified geological database, providing fundamental data support for subsequent path planning of the coal mining robot. The purpose of this step is to construct a high-precision three-dimensional geological model through accurate geological data acquisition and processing, providing reliable geological parameters for unmanned coal mining operations.

[0078] The specific implementation of step S02 is based on the three-dimensional geological model obtained in S01. The working face is arranged along the strike direction of the coal seam, and its length is determined according to the stability of the coal seam, typically controlled within the range of 50–100 m. A layered mining method is adopted, dividing the coal seam vertically into multiple horizontal layers, with each layer's thickness set within the range of 0.8–1.2 m. The selection of the layer thickness threshold is based on the maximum cutting height of the mining robot and the roof stability requirements. A multi-feature clustering function combining K-means clustering and fuzzy C-means clustering is used to process the coal seam feature data. The input parameters include the coal seam thickness data matrix H, the coal seam dip angle distribution matrix A, the coal quality distribution matrix Q, the fault distribution matrix F, and the gas content distribution matrix G. The calculation process of the multi-feature clustering function includes the following steps: First, a coal seam feature vector matrix V is constructed, with each feature vector containing 5 feature components:

[0079] V = {v} ij} m×n , where v ij =(h ij a ij q ij f ij g ij );

[0080] In the formula, v ij h represents the eigenvector at coordinates (i, j) on the working surface; ij a ij q ij f ij g ij These represent the coal seam thickness, dip angle, coal quality grade, fault distance, and gas content at that location, respectively.

[0081] Secondly, the dimensionality of the eigenvectors is reduced using principal component analysis, and the covariance matrix C is calculated.

[0082]

[0083] In the formula, This represents the mean of all eigenvectors.

[0084] Solve for the eigenvalues ​​and eigenvectors of the covariance matrix C. Select the k eigenvectors with the largest contribution rates to form the projection matrix P. The value of k is typically 2 or 3, ensuring that the cumulative contribution rate exceeds 85%. Perform a projection transformation on the original eigenvectors to obtain the dimensionality-reduced feature data V′.

[0085] V′={v′ ij} m×n , where v′ ij =P T v ij ;

[0086] K-means clustering algorithm is used to perform cluster analysis on the dimensionality-reduced feature data. The initial cluster centers are selected based on the maximum distance principle. The clustering process iteratively optimizes the objective function J.

[0087]

[0088] In the formula, K represents the number of clusters; C l μ represents the l-th cluster set; l Represents the l-th cluster center; ||v′ ij -μ l || represents the eigenvector v′ ij To the cluster center μ l The Euclidean distance.

[0089] The iteration terminates when the cluster center position changes by less than 0.05m and the intra-cluster variance is less than a preset threshold of 0.1. Based on the clustering results, the horizontal stratification is optimized to ensure relatively uniform coal seam characteristics within each stratum, generating a stratification thickness parameter matrix L, which contains the start and end depths of each stratum.

[0090] L = [l ij ] p×2 ;

[0091] In the formula, l ij This represents the depth value of the i-th layer at position j, where j=1 represents the starting depth and j=2 represents the ending depth, with the unit being meters; p represents the number of layers.

[0092] A three-dimensional coal mining grid model was established using layered data, with a grid resolution of 0.2m × 0.2m × 0.2m. The model includes coal seam geometry, physical parameters, and layer boundary information. The purpose of this step is to scientifically divide the coal mining layers, providing a spatial partitioning framework for subsequent path planning. By optimizing the layering scheme through multi-feature clustering, the efficiency and safety of coal mining can be improved.

[0093] The specific implementation of step S03 involves dividing each horizontally layered working face into a honeycomb hexagonal grid based on the established three-dimensional coal mining grid model. This transforms the working face space into a planar network composed of regular hexagonal units with a side length of 0.6m. The selection of the hexagonal side length considers the working radius and positioning accuracy of the coal mining robot. When the positioning accuracy of the coal mining robot is higher than 0.1m, the hexagonal side length can be set to 0.6m. The hexagonal grid is generated using the Thiessen polygon method. First, the set of boundary points B of the working face is determined.

[0094] B={(x i y i |i = 1, 2, ..., b};

[0095] In the formula, (x i y i ) represents the coordinates of the i-th point on the working surface boundary; b represents the total number of boundary points.

[0096] A set of grid nodes N is uniformly distributed inside the working face.

[0097] N={(u j v j |j=1,2,...,n};

[0098] In the formula, (u j v j ) represents the coordinates of the j-th grid node inside the working surface; n represents the total number of internal grid nodes.

[0099] Thiessen polygons are constructed based on these nodes, approximating them as regular hexagons. The construction of Thiessen polygons is based on the principle of duality graphs; for each internal node (u... j v j ), its Thiessen polygon P j Defined as:

[0100]

[0101] In the formula, d((x,y),(u,v)) represents the Euclidean distance between two points (x,y) and (u,v) on the plane.

[0102] For irregular hexagonal elements at the working surface boundary, an adaptive mesh adjustment algorithm is used for optimization to ensure that the area change of the mesh element at the boundary does not exceed 20% of the area of ​​a standard hexagon. Each hexagonal mesh node is assigned a unique identifier (ID). ijk It uses a three-digit encoding method:

[0103] ID ijk = i×10000+j×100+k;

[0104] In the formula, i represents the layer number; j and k represent the row number and column number of the plane position, respectively.

[0105] Establish a hexagonal mesh adjacency table R, recording the identifiers of the six adjacent cells for each hexagonal cell:

[0106] Where r i ={ID i , {ID i1 ID i2 ID i6}};

[0107] In the formula, ID i The identifier representing the i-th hexagonal unit; ID i1 To ID i6 The identifier represents the six cells adjacent to the i-th hexagonal cell.

[0108] The purpose of this step is to establish a spatial grid system suitable for the navigation of coal mining robots. Compared with traditional rectangular grids, hexagonal grids provide a more uniform orientation distribution and smaller cumulative error, which is beneficial to improving the accuracy of path planning.

[0109] The specific implementation of step S04 is based on the honeycomb hexagonal mesh constructed in S03, and the Bellman-Ford algorithm is applied to design the main travel path of the coal mining robot; a weighted directed graph G = (V, E) is constructed, where V is the set of hexagonal mesh nodes and E is the set of connecting edges between nodes; for each edge e = (u, v) ∈ E in the graph, the edge weight w(e) is defined as a comprehensive cost function, which considers multiple factors:

[0110] w(e)=α×T(e)+β×S(e)+γ×Q(e);

[0111] In the formula, T(e) represents the thickness cost of edge e, S(e) represents the dip angle cost of edge e, and Q(e) represents the coal quality cost of edge e; α, β, and γ are weighting coefficients, set to 0.4, 0.3, and 0.3 respectively, which can be dynamically adjusted according to actual mining needs, and satisfy α+β+γ=1.

[0112] The formula for calculating the thickness cost T(e) is:

[0113]

[0114] In the formula, G(e) is the coal seam thickness at the position corresponding to edge e, in meters; H max This represents the maximum coal seam thickness at the working face, expressed in meters.

[0115] The formula for calculating the tilt cost S(e) is:

[0116]

[0117] In the formula, A(e) is the dip angle of the coal seam at the position corresponding to edge e, in degrees; A max The maximum inclination angle of the working face is expressed in degrees.

[0118] The formula for calculating the coal quality cost Q(e) is:

[0119]

[0120] In the formula, C(e) is the coal quality score at the position corresponding to edge e, which is dimensionless; C max The highest coal quality score for the working face is dimensionless.

[0121] The Bellman-Ford algorithm calculates the shortest path from the starting point *s* to all other nodes through a relaxation operation. Initialize the distance array *D*:

[0122] D[s] = 0, D[v] = ∞ for all v ∈ V, v ≠ s;

[0123] Perform |V|-1 iterations, relaxing all edges in each iteration:

[0124] D[v] = min(D[v], D[u] + w(u, v)) for all edges (u, v) ∈ E;

[0125] Check for the existence of negative weighted loops:

[0126] For all edges (u, v) ∈ E, if D[v] > D[u] + w(u, v), then there exists a negative weighted cycle.

[0127] The algorithm terminates when the shortest path is stable and there are no negative weight cycles; based on the calculation results, the main travel path node coordinate sequence P of the coal mining robot is generated:

[0128] P={(x1, y1), (x2, y2),..., (x n y n )};

[0129] In the formula, (x i y i ) represents the coordinates of the i-th node on the path, with coordinate precision controlled at the centimeter level.

[0130] The purpose of this step is to optimize the coal mining path based on multiple coal seam parameters, and to solve the shortest path problem of the weighted graph using the Bellman-Ford algorithm, so as to provide the optimal travel route for the coal mining robot.

[0131] The specific implementation of step S05 is based on the honeycomb hexagonal grid of S03 and the edge weight calculation method of S04 to construct a complete coal mining path navigation map. First, the vertex set V of the navigation map is determined, with the center point of each hexagonal grid serving as a navigation vertex. Vertex attributes include three-dimensional spatial coordinates, the layer number to which it belongs, and coal seam characteristic parameters. Connections between vertices are established; for adjacent hexagonal cell center points, bidirectional connection edges are established, and each edge is assigned a weight value. The edge weight factor is calculated using a comprehensive evaluation method, including three main coefficients: coal seam thickness coefficient W. h The value ranges from 0.2 to 0.5. The smaller the coefficient value, the closer the coal seam thickness is to the optimal working height of the coal mining robot (1m); the coal seam dip angle coefficient W... a The value ranges from 0.2 to 0.5. The smaller the coefficient value, the closer the coal seam dip angle is to the optimal working dip angle of the coal mining robot (30 degrees); the coal quality coefficient W... q The coefficient ranges from 0.1 to 0.4, and the value decreases as the coal quality grade increases; the three coefficients satisfy the normalization condition.

[0132] W h +W a +W q =1;

[0133] The final formula for calculating edge weights is:

[0134] W = W h ×D h +W a ×D a +W q ×D q ;

[0135] In the formula, D h D a D q These are the standardized distance values ​​for coal seam thickness, dip angle, and coal quality, respectively. The calculation formula is as follows:

[0136]

[0137] In the formula, h is the thickness of the coal seam at the current location, in meters; opt The optimal working height for a coal mining robot is typically 1 meter; h max and h min These represent the maximum and minimum coal seam thicknesses at the working face, in meters.

[0138]

[0139] In the formula, 'a' represents the dip angle of the coal seam at the current location, in degrees; opt The optimal working angle for a coal mining robot is typically 30 degrees; a max and a minThese are the maximum and minimum coal seam dip angles of the working face, respectively, in degrees.

[0140]

[0141] In the formula, q represents the coal quality grade value at the current location, which is dimensionless; q max and q min These are the highest and lowest coal quality grades for the working face, respectively, and are dimensionless.

[0142] A multi-dimensional weighted graph structure G = (V, E, W) is constructed, where V is the vertex set, E is the edge set, and W is the weight set. An adjacency list is used to store the graph structure to improve graph search efficiency. The navigation graph is simplified by removing edges with weights exceeding a threshold of 3.0 to reduce computational complexity. The purpose of this step is to establish a complete navigation graph structure that considers the influence of multiple factors, providing accurate spatial navigation information for coal mining robots and supporting path planning in complex environments.

[0143] The specific implementation of step S06 is based on the geological data obtained in S01 and the navigation map constructed in S05, developing a unidirectional zoned mining strategy. First, the working face is divided based on the coal seam dip angle distribution matrix A. When the dip angle difference between adjacent areas exceeds 15 degrees, setting zone boundaries is considered. The zone division is refined based on the coal seam thickness data matrix H. When the thickness difference between adjacent areas exceeds 0.5m, the zone boundaries are adjusted. Taking into account the location of geological faults, the areas on both sides of the fault are forcibly divided into different mining areas. The working face is typically divided into 3-5 mining areas, with each mining area maintaining a similar area and an error controlled within 15%. The optimal mining direction is determined for each mining area, and principal component analysis is used to determine the main characteristics of the coal seam. The mining direction is set perpendicular to the main direction to minimize changes in coal seam characteristics during mining. The mining directions of adjacent mining areas should be kept consistent or the angle between them should be less than 45 degrees to reduce adjustment ranges during transitions. A transition zone with a width of 2-3 meters is designed at the boundary of each mining area, using a smooth curve to connect the mining paths of different areas, with a radius of curvature not less than 1.5 times the minimum turning radius of the mining robot. A mining area priority ranking is established based on the coal quality distribution matrix Q and the gas content distribution matrix G, prioritizing mining areas with high coal quality and low gas content. Optimal transition paths between mining areas are designed to ensure the mining robot can safely and efficiently enter the next mining area after completing one. The purpose of this step is to avoid redundant movement of the mining robot, reduce non-productive time, and improve mining efficiency by rationally dividing mining areas and planning mining directions.

[0144] The specific implementation of step S07 is based on the layer thickness parameter matrix L generated in S02 and the set of boundary points B for dividing the horizontal layers of the working face, to develop a real-time path dynamic planning strategy; the coal mining robot is equipped with a forward detection radar with a detection range of 3-5m and a scanning frequency of not less than 1Hz to collect information about the coal seam ahead in real time; a gas sensor array is installed with a sampling frequency of not less than 0.5Hz and a measurement accuracy better than 0.01%, to construct a local gas content distribution matrix G′:

[0145] G′=[g′ ij ] m′×n′ ;

[0146] In the formula, g′ ij The value represents the gas content at coordinates (i, j) in the local area, expressed as a percentage; m′ and n′ represent the number of grids in the x and y directions of the local area, respectively.

[0147] A sliding window method is used to update local grid data, with a window size of 5m × 5m and a sliding step of 1m. Only grid points where the difference between the detected data and the original data exceeds a threshold are updated. The thresholds for thickness difference are set at 0.2m, dip angle difference at 5 degrees, and gas content difference at 0.1%. Based on the updated local grid data, the D-Star algorithm is used for dynamic path planning. The D-Star algorithm adds dynamic update capability to the Bellman-Ford algorithm, triggering replanning when environmental changes exceed preset thresholds. The replanning trigger conditions include: the difference between the detected coal seam thickness and the original data exceeds 0.3m, the difference between the detected coal seam dip angle and the original data exceeds 10 degrees, and the detected gas content exceeds the safety threshold of 1%. The replanning process adopts a local search strategy, recalculating the path only in the affected area, with the search range limited to a radius of 10m. The new optimal path segment replaces the corresponding part in the original path. Path smoothing is performed using cubic spline interpolation to control the path curvature change rate to be less than 0.2 / m. 2 This step ensures the smooth movement of the coal mining robot. The purpose of this step is to dynamically adjust the coal mining path, enabling the robot to respond promptly to changes in the coal seam based on real-time detection data, thus guaranteeing the safety and adaptability of the coal mining operation.

[0148] The specific implementation of step S08 involves the coal mining robot adjusting its mining parameters based on multi-source coal seam data to achieve precise cutting and safe mining. The coal mining robot is equipped with a laser rangefinder and an inclination sensor to measure its relative position and orientation to the coal seam in real time, with a ranging accuracy better than 0.05m and an inclination measurement accuracy better than 0.5 degrees. The drum height h is adjusted based on the coal seam thickness data matrix H. r The drum height setting range is 0.6–1.5 m, with an adjustment accuracy of 0.01 m. When the measured coal seam thickness h… meaWhen the height is less than 0.7m, the drum height should be set to the coal seam thickness minus 0.1m.

[0149]

[0150] In the formula, h r Set the roller height value in meters (h). mea The measured coal seam thickness is in meters.

[0151] Based on the coal seam dip angle distribution matrix A, the robot's working posture angle θ is adjusted. A hydraulic support system keeps the drum axis parallel to the coal seam direction. The dip angle adaptation range is 0–60 degrees. When the coal seam dip angle α… mea When the angle exceeds 50 degrees, activate the additional stabilization support system to prevent the robot from tipping over.

[0152]

[0153] In the formula, θ is the robot's working posture angle setting value, in degrees; a mea The measured coal seam dip angle is expressed in degrees; δ is the additional stability angle compensation value, which is usually set to 5 to 10 degrees.

[0154] Cutting depth d is controlled based on the layer thickness parameter matrix L. c The cutting depth is set within the range of 0.3 to 0.8 meters. When the working surface is d meters away from the layer boundary... b As the cutting approaches the layer boundary, the cutting depth gradually decreases, with a minimum adjustable depth of 0.1m, ensuring that the cutting does not exceed the layer boundary.

[0155]

[0156] In the formula, d c Set a value for the cutting depth, in meters; d max The maximum cutting depth is typically set to 0.8m; d b λ represents the distance from the layer boundary in meters; λ is an adjustment factor, usually set to 2.0.

[0157] The boundary treatment is guided by the set of boundary points B, which is divided into horizontal layers on the working face. When the distance to the layer boundary is less than 1m, the cutting speed v c Reduce speed to 50% of normal speed to improve boundary processing accuracy:

[0158]

[0159] In the formula, v c Set the cutting speed value in m / min; v max The maximum cutting speed is typically set to 3 m / min; d b This represents the distance from the layer boundary, in meters.

[0160] A real-time adjustment strategy for coal mining parameters is established, employing a fuzzy control algorithm. Input variables include coal seam thickness, dip angle, and distance from the boundary; output variables include drum height, cutting depth, and cutting speed. The fuzzy rule base contains 30-50 rules, covering various coal seam conditions. The fuzzy control algorithm is based on the following fuzzy inference process:

[0161] First, fuzzification is performed to convert the input variable x (coal seam thickness, coal seam dip angle, distance from boundary) into a fuzzy set:

[0162]

[0163] In the formula, μ A (x) represents the membership function value of the fuzzy set A corresponding to the input variable x; a, b, and c are the membership function parameters, which are determined according to the actual coal seam conditions.

[0164] Then, fuzzy inference is performed based on the fuzzy rule base. The fuzzy rules adopt the Mamdani inference model, and the rule R is... i The general form is:

[0165] If x1 is A1, x2 is A2, and x3 is A3, then y is B;

[0166] Where x1, x2, and x3 represent the coal seam thickness, coal seam dip angle, and distance from the boundary, respectively; A1, A2, and A3 represent the corresponding fuzzy sets; y represents the output variable (drum height, cutting depth, or cutting speed); and B represents the fuzzy set corresponding to the output variable.

[0167] Finally, deblurring is performed, and the accurate output value y is calculated using the centroid method. * :

[0168]

[0169] In the formula, y * This is the precise output value after deblurring; y i μ is the i-th discrete point of the output variable; B (y i ) represents the membership function value corresponding to this point; n is the number of discrete points.

[0170] A safety monitoring system is set up so that when the gas concentration g exceeds 1.5%, the cutting speed is automatically reduced, and when it exceeds 2%, the coal mining operation is suspended.

[0171]

[0172] In the formula, v c Set the cutting speed value in m / min; v maxThe maximum cutting speed is typically set to 3 m / min; v normal The normal cutting speed is calculated using the aforementioned fuzzy control algorithm; g represents the gas concentration, expressed as a percentage.

[0173] Establish a roof monitoring system and use acoustic detection technology to monitor the roof condition. When the roof integrity index I... r If the cutting depth is below 0.8, reduce the cutting depth; if it is below 0.6, stop mining in the current area.

[0174]

[0175] In the formula, d c Set a value for the cutting depth, in meters; d normal The normal cutting depth is calculated using the aforementioned fuzzy control algorithm; I r The roof integrity index is dimensionless and ranges from 0 to 1. It is calculated based on the characteristics of reflected waves obtained from acoustic detection.

[0176]

[0177] In the formula, f i The score for the i-th acoustic feature parameter, with a value ranging from 0 to 1; w i is the weight coefficient of the i-th feature parameter; k is the number of feature parameters, usually 3 to 5.

[0178] The purpose of this step is to achieve precise control of coal mining parameters, adapt to the complex conditions of steeply inclined thin coal seams, ensure coal mining quality and safety, and improve coal recovery rate.

[0179] To better understand and implement this invention, the following is a specific application scenario example 2: Workers implemented unmanned coal mining technology for steeply inclined thin coal seams in a mining area. The average dip angle of the coal seam in this area is 52 degrees, and the thickness ranges from 0.9 to 1.8 meters, classifying it as a typical steeply inclined thin coal seam. Due to the complex geological conditions in this mining area, traditional manual coal mining methods suffer from high safety risks and low recovery rates, necessitating the introduction of unmanned coal mining technology to improve safety and mining efficiency.

[0180] First, the staff used a high-resolution ground-penetrating radar with a frequency of 1.2 GHz to scan the target working face, with a scan line spacing of 1.5 m and a detection depth of 75 m. Simultaneously, 25 boreholes were drilled, spaced 25 m apart, and the core samples were analyzed in detail. A three-dimensional geological model of the coal seam was constructed using the Kriging interpolation algorithm. This model accurately describes the geometry and physical properties of the coal seam, achieving a resolution of 0.5 m × 0.5 m. Based on the model data, relevant feature matrices were extracted, as shown in Table 1.

[0181] Table 1 Basic parameters of coal seam characteristic matrix

[0182]

[0183]

[0184] Based on the acquired geological data, the staff used a stratified mining method to divide the coal seam into multiple horizontal layers. First, a multi-feature clustering function was used to analyze the coal seam characteristics; the feature vector had a dimension of 5 and included coal seam thickness, dip angle, coal quality grade, fault distance, and gas content. Dimensionality reduction was performed using principal component analysis, extracting two principal components with a cumulative contribution rate of 87.3%. The K-means clustering algorithm was then used to cluster the feature data, converging after 16 iterations, with the intra-class variance reduced to 0.08. Based on the clustering results, the coal seam was divided into 6 horizontal layers along its vertical thickness, with an average layer thickness of 1.05m. Figure 2 As shown in the diagram, the green dashed line represents the planned robot path. The structure of the coal mining robot used is as follows: Figure 3 As shown.

[0185] Table 2 Horizontal Stratification Parameters

[0186]

[0187] Subsequently, the staff divided each horizontal layer using a honeycomb hexagonal grid, with a grid side length set to 0.6m. The hexagonal grid was generated using the Thiessen polygon method, resulting in 2784 hexagonal units on the working surface. Figure 4 As shown in the diagram, each hexagonal cell is assigned a unique identifier using a three-digit code. The first digit represents the layer number (1–6), and the last two digits represent the planar position number (01–99). An adjacency table is constructed based on the hexagonal grid to record the connection relationships between each cell and its six adjacent cells, providing topological data support for path planning.

[0188] For path planning, the staff applied the Bellman-Ford algorithm to design the main travel path of the coal mining robot. First, a weighted directed graph was constructed with 2784 vertices and 16704 edges. For each edge, a weight value was calculated based on a comprehensive cost function. The weight coefficients for thickness cost, dip angle cost, and coal quality cost were set to 0.4, 0.3, and 0.3, respectively. The shortest path from the starting point to the ending point was calculated using the Bellman-Ford algorithm. After 2783 iterations, the algorithm converged, generating the coordinate sequence of the main travel path of the coal mining robot, containing 387 path nodes.

[0189] Table 3 Path Planning Parameters

[0190] Parameter name Parameter value unit illustrate The weighting coefficient α of the comprehensive cost function 0.4 Dimensionless Thickness cost weight The weighting coefficient β of the comprehensive cost function 0.3 Dimensionless Tilt cost weight The weighting coefficient γ of the comprehensive cost function 0.3 Dimensionless Coal quality cost weight Maximum coal seam thickness at the working face 1.8 m <![CDATA[H max Value Maximum dip angle of working face 58 Spend <![CDATA[A max Value Highest coal quality score at the working face 5 Dimensionless <![CDATA[C max Value Number of path nodes 387 indivual Total number of nodes in the optimal path Path length 232.2 m Optimal path total length

[0191] Based on a hexagonal grid and edge weight calculation method, the staff constructed a complete coal mining path navigation map. The edge weights in the navigation map were calculated using the coal seam thickness coefficient, dip angle coefficient, and coal quality coefficient, with values ​​ranging from 0.2 to 0.5, 0.2 to 0.5, and 0.1 to 0.4, respectively. The optimal working height for the coal mining robot was set at 1.0 m, and the optimal working dip angle at 30 degrees. The navigation map was simplified by removing edges with weights exceeding 2.8 to reduce computational complexity.

[0192] Subsequently, based on the acquired geological data and the constructed navigation map, the staff developed a unidirectional zoned mining strategy. The working face was divided into four mining zones, each with an area of ​​approximately 125m². 2 The area error was controlled within 10%. The optimal mining direction was determined for each mining area, and the width of the transition zone between mining areas was set to 2.5m. A smooth curve with a radius of curvature of 3.6m was used to connect the mining paths of different mining areas. Based on the coal quality distribution matrix and the gas content distribution matrix, mining operations were carried out in the mining areas according to the priority order in Table 2.

[0193] In actual coal mining operations, the coal mining robot is equipped with a forward-facing detection radar with a detection range of 4m and a scanning frequency of 1.5Hz. The gas sensor array has a sampling frequency of 0.8Hz and a measurement accuracy of 0.005%. A sliding window method is used to update the local area grid data, with a window size of 5m × 5m and a sliding step size of 1m. The thickness difference threshold is set to 0.2m, the dip angle difference threshold to 5 degrees, and the gas content difference threshold to 0.1%. When the detected environmental changes exceed the preset thresholds, the D-Star algorithm is triggered for dynamic path replanning, with the search range limited to a radius of 10m.

[0194] Table 4 Parameter Table for Coal Mining Robot

[0195] Parameter type Parameter value unit illustrate Roller height setting range 0.6～1.5 m Adjustable according to coal seam thickness Cutting depth setting range 0.3～0.8 m Adjustable according to layer boundaries Tilt angle adaptation range 0～60 Spend Supports high tilt angle operation Laser ranging accuracy 0.03 m 0.05m higher than required Inclination measurement accuracy 0.3 Spend 0.5 degrees higher than required Forward detection range 4 m Real-time coal seam detection Gas detection accuracy 0.005 ％ High-precision security monitoring

[0196] The coal mining robot adjusts mining parameters based on multi-source coal seam data to achieve precise cutting and safe mining. The drum height is automatically adjusted according to the coal seam thickness; when the coal seam thickness is less than 0.7m, the drum height is set to the coal seam thickness minus 0.1m. The robot's working posture is adjusted according to the coal seam dip angle; when the dip angle is greater than 50 degrees, an additional stabilizing support system is activated to prevent tipping. The cutting depth is dynamically adjusted based on the distance from the seam boundary; when the distance to the boundary is less than 1m, the cutting speed is reduced to 50% of the normal speed. The safety monitoring system monitors the gas concentration in real time; when the gas concentration exceeds 1.5%, the cutting speed is automatically reduced; when it exceeds 2%, coal mining operations are suspended.

[0197] Through the unmanned coal mining method described in this embodiment, workers have successfully achieved safe and efficient mining of steeply inclined thin coal seams. Compared with traditional coal mining methods, this invention has significant advantages. Traditional mining of steeply inclined thin coal seams mainly relies on manual top-coal caving or simple mechanized mining, which has problems such as high safety risks, high labor intensity, and low coal recovery rate. Workers need to operate in a high-angle environment, resulting in frequent safety accidents; roof management is difficult, with a high risk of collapse; coal mining efficiency is low, and resource waste is serious.

[0198] This invention's unmanned coal mining method completely solves these problems. By adopting a layered mining method and honeycomb hexagonal grid planning, the coal recovery rate is increased from 65% in traditional methods to over 85%; intelligent path planning and real-time dynamic adjustment improve mining efficiency by 2.3 times; coal mining robots replace manual labor, eliminating safety accidents; precise parameter control significantly improves coal quality, resulting in uniform coal particle size and reducing the generation of fine coal. After implementing this invention, the mining cost in this mine area decreased by 37%, the accident rate dropped to zero, and the overall economic benefits improved significantly.

[0199] It should be noted that the variables involved in this invention are explained in detail in Tables 5 and 6 below.

[0200] Table 5. Variable Explanation Table (Part 1)

[0201]

[0202]

[0203] Table 6. Variable Explanation Table (Part Two)

[0204]

[0205]

[0206] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for unmanned mining of steeply inclined thin coal seams, characterized in that, include: A three-dimensional geological model of thin coal seams was constructed to determine the coal seam dip angle distribution matrix, coal seam thickness data matrix, and fault distribution matrix. The thin coal seam is horizontally divided and a three-dimensional coal mining grid model is established; a honeycomb hexagonal grid is used to divide the working face to form the basic network for calculating the coal mining path; the main travel path of the coal mining robot is designed based on the Bellman-Ford algorithm, and the optimal coal mining path is calculated by comprehensively considering the coal seam thickness data matrix, coal seam dip angle distribution matrix and coal quality distribution matrix; a coal mining path navigation map is constructed to form a multi-dimensional weighted graph structure. Develop a unidirectional partitioned data acquisition strategy; The process involves implementing dynamic path planning and coal mining parameter adjustment for coal mining robots. Specifically, the steps of dividing thin coal seams into horizontal layers and establishing a three-dimensional coal mining grid model include: designing working faces along the strike direction; employing a layered mining method; dividing the steeply inclined thin coal seam into multiple horizontal layers according to thickness, with each layer ranging from 0.8 to 1.2 meters in thickness; using multi-feature clustering functions to optimize the coal seam layers; and establishing a three-dimensional coal mining grid model. The step of using a honeycomb hexagonal grid to divide the working face into a basic network for coal mining path calculation involves dividing the working face into multiple layers with side lengths of 0. A 6-meter regular hexagonal unit forms the basic network for coal mining path calculation, improving the accuracy of path planning. The unidirectional zone mining strategy is a coal mining operation method designed for the characteristics of steeply inclined thin coal seams. A complete working face is divided into multiple mining areas of similar area according to the variation characteristics of the coal seam thickness data matrix and the coal seam dip angle distribution matrix. A fixed coal mining direction is set in each mining area. The coal mining robot completes coal mining in a single direction in one mining area before entering the next mining area, avoiding the coal mining robot moving back and forth in the working face and reducing non-productive movement time.

2. The method for unmanned mining of steeply inclined thin coal seams according to claim 1, characterized in that, The honeycomb hexagonal grid is a spatial representation method that divides the coal mining face space into a set of regular hexagonal units with a side length of 0.6 meters. Each hexagonal unit shares a boundary with its six neighboring units. Compared with the traditional rectangular grid, the hexagonal grid has a constant distance between adjacent units in any direction, reducing the cumulative error of directional deviation and improving path smoothness.

3. The method for unmanned mining of steeply inclined thin coal seams according to claim 2, characterized in that, The steps for developing a unidirectional zone mining strategy are as follows: the working face is divided into multiple mining zones, and a unidirectional travel path is set for each mining zone to avoid the coal mining robot repeatedly passing through the same area and improve coal mining efficiency.

4. The method for unmanned mining of steeply inclined thin coal seams according to claim 3, characterized in that, The steps for implementing dynamic path planning for the coal mining robot are as follows: Based on the layer thickness parameter matrix and the set of boundary points for dividing the working face horizontal layers, the coal mining robot collects the gas content distribution matrix according to the real-time coal seam detection data, periodically updates the local area grid weights, recalculates the optimal path, and adapts to changes in the coal seam.

5. The method for unmanned mining of steeply inclined thin coal seams according to claim 4, characterized in that, The steps of the coal mining robot to adjust coal mining parameters to achieve precise cutting and safe and efficient mining are as follows: The coal mining robot adjusts the coal mining parameters and controls the drum height and cutting depth based on the coal seam thickness data matrix, coal seam dip angle distribution matrix, layer thickness parameter matrix and the boundary point set of the working face horizontal layer division, so as to achieve precise cutting and safe and efficient mining for steeply inclined thin coal seams.

6. The method for unmanned mining of steeply inclined thin coal seams according to claim 5, characterized in that, The layered mining method specifically involves dividing a steeply inclined thin coal seam into multiple parallel layers with a thickness of 0.8 to 1.2 meters in the horizontal direction. The coal mining robot completes the mining operation independently in each parallel layer. After completing the mining of one parallel layer, the coal mining robot enters the next parallel layer. This method is suitable for steeply inclined coal seams with an inclination angle greater than 45 degrees. By controlling the range of roof movement through layering, the risk of roof collapse is reduced, and the coal recovery rate is improved.

7. The method for unmanned mining of steeply inclined thin coal seams according to claim 6, characterized in that, The Bellman-Ford algorithm is a graph algorithm for finding the shortest path from a single source. It calculates the shortest path from the starting point to all vertices by performing multiple rounds of relaxation operations on all edges in the graph. In the weighted graph structure designed in coal mining path planning, the edge weights integrate the coal seam thickness coefficient, coal seam dip angle coefficient, and coal quality coefficient. By minimizing the comprehensive cost, it achieves the optimal utilization of resources in the coal mining path.

8. The method for unmanned mining of steeply inclined thin coal seams according to claim 7, characterized in that, The multi-feature clustering function is used to scientifically divide the coal mining face based on multiple physical properties of the coal seam. The input includes a coal seam thickness data matrix, a coal seam dip angle distribution matrix, a coal quality distribution matrix, a fault distribution matrix, and a gas content distribution matrix. The output is a layer thickness parameter matrix and a set of boundary points for dividing the horizontal layers of the working face.

Images