A goaf accumulation body void type identification method

Abstract

This application discloses a method for identifying void types in goaf accumulation, belonging to the field of goaf void identification. The method includes: determining the void envelope region corresponding to each void voxel in the spatial distribution data of the goaf accumulation voids; determining the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel based on each void envelope region; and determining the voxel type discrimination label for each void voxel based on the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel. This application effectively captures the local differences in the void structure of complex accumulations by introducing local envelope regions and combining multi-dimensional feature extraction and voxel-level discrimination, improving the refinement of void type identification and avoiding the recognition deficiencies caused by overall scale and coarse-grained indicators.

Description

Technical Field

[0001] This application belongs to the field of goaf void identification, specifically relating to a method, apparatus, equipment, storage medium, and computer program product for identifying the void type of goaf accumulation. Background Technology

[0002] Goaf deposits formed after mining activities typically contain numerous irregular voids, which vary significantly in spatial morphology, connectivity, and local structure. Different types of voids have different impacts on engineering applications such as goaf stability analysis, disaster risk assessment, and seepage simulation. Therefore, it is necessary to identify and differentiate the types of internal voids when performing 3D modeling and structural analysis of the goaf deposits.

[0003] In existing technologies, void type identification is typically based on voxelized 3D data, classifying void regions through methods such as global connected component analysis, local density statistics, or morphological parameter-based methods such as sphericity. These methods generally treat the void region as a whole, classifying types based on statistical characteristics at the overall scale or single morphological indicators, and are suitable for coarse geometric feature analysis of void structures.

[0004] However, when the internal void structure of the accumulation is complex, has significant local variations, or has irregular boundaries, the above process often fails to accurately reflect the structural differences at different locations within the voids. This can lead to insensitivity to local features, unstable type determination, or insufficient detail representation, thus affecting the precision of void type identification and consequently impacting the accuracy of three-dimensional analysis. Summary of the Invention

[0005] This application aims to provide a method, apparatus, equipment, storage medium, and computer program product for identifying void types in goaf accumulation, at least to address the problem of insufficient precision in void type identification.

[0006] In a first aspect, embodiments of this application disclose a method for identifying the void type of goaf accumulation, including: Based on the spatial distribution data of voids in the goaf accumulation, the void envelope region corresponding to each void voxel in the spatial distribution data is determined. Based on each void envelope region, determine the directional feature data, aggregation feature data, and boundary undulation feature data for each void voxel; the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the void boundary voxels in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation; Based on the directional feature data, clustering feature data, and boundary undulation feature data of each void voxel, a voxel type discrimination label is determined for each void voxel.

[0007] Secondly, embodiments of this application also disclose a device for identifying the void type of goaf accumulation, comprising: The neighborhood extraction module is used to determine the void envelope region corresponding to each void voxel in the spatial distribution data based on the spatial distribution data of voids in the goaf accumulation. The feature calculation module is used to determine the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel based on each void envelope region; the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the void boundary voxels in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation; The label determination module is used to determine the voxel type discrimination label of each void voxel based on the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel.

[0008] Thirdly, embodiments of this application also disclose an electronic device, including a processor and a memory, wherein the memory stores a program or instructions that can run on the processor, and the program or instructions, when executed by the processor, implement the steps of the method described in the first aspect.

[0009] Fourthly, embodiments of this application also disclose a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the method described in the first aspect.

[0010] Fifthly, embodiments of this application also disclose a computer program product, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps described in the first aspect.

[0011] In summary, in this embodiment, a corresponding void envelope region is constructed for each void voxel based on the spatial distribution data of voids in the goaf accumulation. This allows the voxel to perform structural analysis within its local spatial neighborhood. With the help of this local range, void identification no longer relies on a single statistical feature at the overall scale, but rather on characterizing the local environment of the voxel itself. This enhances the sensitivity to local structural changes and provides a more targeted spatial basis for subsequent feature extraction. Furthermore, by extracting directional, clustering, and boundary undulation feature data within each void envelope region, the void voxel can obtain fine-grained structural information from multiple dimensions, such as structural orientation consistency, boundary voxel clustering degree, and local boundary undulation. By characterizing and utilizing the local geometric morphology, spatial distribution, and boundary variations reflected by these features, the local differences of complex void structures at different locations are more fully revealed, thereby enhancing the ability to express local structural changes and improving the stability of type identification in areas with complex structural morphology or irregular boundaries. Finally, based on multi-dimensional feature data, voxel-level type identification is performed on each void voxel, enabling the distinction between cracks and pores to be completed at the local scale. This allows the identification process to directly rely on the local structural features of voxels, further identifying structural differences in different local regions within the same connected void volume, reducing instability in type identification caused by the overall scale masking local changes, and improving the ability to express detailed structures. Therefore, the method based on the embodiments of this application, by introducing local envelope regions and combining multi-dimensional feature extraction and voxel-level identification, more effectively captures the local differences in the void structure of complex accumulation bodies, improves the refinement of void type identification, avoids the insufficient identification caused by the overall scale and coarse-grained indicators, and thus improves the accuracy of 3D modeling and structural analysis. Attached Figure Description

[0012] In the attached diagram: Figure 1 This is a flowchart illustrating the steps of a method for identifying the void type of a goaf accumulation body according to an embodiment of this application. Figure 2 This is a flowchart of another method for identifying the void type of goaf accumulation provided in an embodiment of this application; Figure 3 This is a schematic diagram of a voxel extraction process according to an embodiment of this application; Figure 4 This is a program execution logic for determining a weighting coefficient in an embodiment of this application; Figure 5 This is a block diagram of a device for identifying the void type of goaf accumulation provided in an embodiment of this application; Figure 6 This is a block diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, the "and / or" signifies at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0015] like Figure 1 The image shows a method for identifying the void type of a goaf accumulation body provided in an embodiment of this application.

[0016] The method may include the following steps: Step 101: Based on the spatial distribution data of voids in the goaf accumulation, determine the void envelope region corresponding to each void voxel in the spatial distribution data.

[0017] In some embodiments of this application, to enable structural analysis of each void voxel within its local spatial range, it is typically necessary to determine a corresponding void envelope region for the voxel based on the spatial distribution data of voids in the goaf accumulation. By defining a local neighborhood for the voxel in three-dimensional space, it is possible to acquire the set of surrounding void voxels and boundary voxels within a defined range, thereby forming the spatial basis upon which subsequent extraction of directional, clustering, and boundary undulation features depends. The void envelope region represents the set of local void voxels and the set of local boundary voxels contained within a preset scale centered on the voxel. This region reflects the neighborhood structure and geometric environment of the voxel in its local space. With the establishment of this region, voxel-level analysis can be freed from the limitations of the overall scale, allowing local structural differences to be preserved and providing a stable data source for subsequent feature calculations.

[0018] In a specific example, spatially distributed data can be represented in three-dimensional voxel form as a discrete distribution of void voxels and non-void voxels. This allows any void voxel in three-dimensional space to be selected as the analysis object. Around this voxel, a fixed-scale spatial neighborhood can be constructed at its three-dimensional coordinate position, and void voxels and boundary voxels can be extracted from this neighborhood to form the void envelope region corresponding to that voxel. After constructing this region, the spatial distribution information of the voxel within its local neighborhood can be obtained, providing necessary data support for subsequent calculations of directional consistency, boundary voxel aggregation, and local boundary undulations.

[0019] Step 102: Determine the directional feature data, clustering feature data, and boundary undulation feature data of each void voxel based on each void envelope region.

[0020] Among them, the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the boundary voxels of the voids in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation.

[0021] In some embodiments of this application, to enable void voxels to obtain local structural characterization within their corresponding void envelope regions, directional feature data, clustering feature data, and boundary undulation feature data are typically extracted based on the voxel distribution within these regions. As the void envelope region provides a local set of void voxels and a local set of boundary voxels, these features can reflect the geometric morphology and distribution characteristics of voxels in local space from different perspectives. Directional feature data is used to present the directional consistency of the local structure, clustering feature data is used to reflect the concentration trend of boundary voxels around the principal axis, and boundary undulation feature data is used to describe the undulation changes of the local boundary surface. By extracting the above features within the void envelope region, voxels can obtain multi-dimensional structural information at a local scale, providing basic data for subsequent type determination.

[0022] In a specific example, a void voxel can be selected from the spatially distributed data. Based on the void envelope region corresponding to that voxel, its local void voxel set and local boundary voxel set can be obtained. Subsequently, the principal direction vector can be calculated based on the spatial distribution of the local void voxels, thus obtaining directional feature data. Then, the aggregation feature data can be determined based on the distance distribution of the local boundary voxels to the principal axis. Finally, the boundary undulation feature data can be obtained by comparing the local true boundary area with its projected area on the principal plane. The above processing method can also be applied to other void voxels in the spatially distributed data, so that each voxel can obtain three types of feature data reflecting its local directional consistency, the degree of boundary voxel aggregation, and the boundary surface undulation, providing input for subsequent voxel type determination.

[0023] Step 103: Determine the voxel type discrimination label for each voxel based on the directional feature data, clustering feature data, and boundary undulation feature data of each voxel.

[0024] In some embodiments of this application, to enable void voxels to obtain a clear type identification based on their local structural features, directional feature data, clustering feature data, and boundary undulation feature data obtained within the void envelope region are typically used as a comprehensive basis to form a voxel type discrimination label. Since these three types of features respectively reflect local directional consistency, the clustering trend of boundary voxels around the principal axis, and the undulation changes of the local boundary surface, joint analysis of these features can distinguish between void voxels and pore voxels at a local scale. With this discrimination method based on multi-dimensional features, the voxel type label can more closely reflect its local geometry and spatial distribution characteristics, thus providing a structural basis for subsequent spatial distribution label generation.

[0025] In a specific example, taking any void voxel in spatially distributed data as an example, the three types of feature data obtained within the void envelope region can be used as input and comprehensively analyzed through preset discrimination rules. When the directional consistency and clustering are strong and the boundary undulation is low, it can be classified as a fissure voxel; if the directional consistency is weak, the clustering is not obvious, and the boundary undulation is high, it can be classified as a void voxel. The same processing method can be applied to all void voxels in spatially distributed data, so that each voxel can obtain a corresponding voxel type discrimination label, which serves as the basic data unit for generating subsequent spatial distribution label data.

[0026] In summary, in this embodiment, a corresponding void envelope region is constructed for each void voxel based on the spatial distribution data of voids in the goaf accumulation. This allows the voxel to perform structural analysis within its local spatial neighborhood. With the help of this local range, void identification no longer relies on a single statistical feature at the overall scale, but rather on characterizing the local environment of the voxel itself. This enhances the sensitivity to local structural changes and provides a more targeted spatial basis for subsequent feature extraction. Furthermore, by extracting directional, clustering, and boundary undulation feature data within each void envelope region, the void voxel can obtain fine-grained structural information from multiple dimensions, such as structural orientation consistency, boundary voxel clustering degree, and local boundary undulation. By characterizing and utilizing the local geometric morphology, spatial distribution, and boundary variations reflected by these features, the local differences of complex void structures at different locations are more fully revealed, thereby enhancing the ability to express local structural changes and improving the stability of type identification in areas with complex structural morphology or irregular boundaries. Finally, based on multi-dimensional feature data, voxel-level type identification is performed on each void voxel, enabling the distinction between cracks and pores to be completed at the local scale. This allows the identification process to directly rely on the local structural features of voxels, further identifying structural differences in different local regions within the same connected void volume, reducing instability in type identification caused by the overall scale masking local changes, and improving the ability to express detailed structures. Therefore, the method based on the embodiments of this application, by introducing local envelope regions and combining multi-dimensional feature extraction and voxel-level identification, more effectively captures the local differences in the void structure of complex accumulation bodies, improves the refinement of void type identification, avoids the insufficient identification caused by the overall scale and coarse-grained indicators, and thus improves the accuracy of 3D modeling and structural analysis.

[0027] Figure 2 This is another method for identifying the void type of goaf accumulation provided in the embodiments of this application.

[0028] The method may include the following steps: Step 201: Determine spatial distribution data based on the sampling data of the accumulation distribution of the goaf accumulation.

[0029] In some embodiments of this application, in order to obtain three-dimensional spatial foundational data that reflects the internal structure of the goaf accumulation, corresponding spatial distribution data is typically determined based on the accumulation distribution sampling data. The accumulation distribution sampling data is used to present the structural state at different locations within the accumulation. By spatially mapping this sampling information, a voxel-based spatial representation can be formed in a three-dimensional coordinate system, enabling subsequent void identification to proceed within a unified spatial framework. This spatial distribution data provides a fundamental spatial basis for subsequently constructing void envelope regions, extracting local structural features, and conducting voxel-level type discrimination, thus providing a clear spatial reference for the void structure analysis process.

[0030] In a specific example, a set of sampling data on the distribution of the goaf can be obtained from within the goaf accumulation. This sampling data may come from voxelized records of the local structure of the accumulation, void phase segmentation results reconstructed by computed tomography (CT), or local state samples extracted from slice sequences. When processing this set of sampling data, its spatial location and corresponding value information can be mapped to a three-dimensional voxel grid, forming a discrete distribution structure composed of voxels in space, thus constituting the corresponding spatial distribution data. The same mapping method can be applied to other forms of sampling data, enabling sampling information from different sources to obtain a consistent voxelized representation in three-dimensional space, providing the necessary data foundation for subsequent construction of void envelope regions and extraction of local structural features.

[0031] Optionally, step 201 includes the following sub-steps: Sub-step 2011: Determine the spatial coordinates of each voxel in the goaf accumulation based on the sampling data of the accumulation distribution of the goaf accumulation.

[0032] In some embodiments of this application, in order to establish a basic coordinate framework in three-dimensional space that reflects the internal structure of the goaf accumulation, the spatial coordinates of each voxel in the accumulation are typically determined based on the accumulation distribution sampling data. The accumulation distribution sampling data is used to present the structural state at different locations within the accumulation. By spatially mapping this sampling information, it can be mapped to a regularized three-dimensional voxel grid, thus clarifying the spatial positional relationship of each voxel. These spatial coordinates provide a unified spatial reference for subsequent identification of void voxels, construction of spatial distribution data, and conduct of local structural analysis.

[0033] In a specific example, a set of sampling data on the distribution of the goaf accumulation can be spatially organized, arranging the sampling points in a three-dimensional coordinate system according to their actual distribution within the accumulation. As the positions of the sampling points in space are mapped one by one, a spatial structural framework composed of voxel units can be formed in a three-dimensional voxel mesh, thus determining the coordinate position of each voxel in the mesh. The same spatial organization method can be applied to other forms of sampling data, ensuring that the structural information obtained by different sampling methods can form a consistent voxelized representation in three-dimensional space, providing the necessary data foundation for subsequent extraction of void voxel position coordinates.

[0034] Sub-step 2012 extracts the void voxel position coordinates of each void voxel in the goaf accumulation from all spatial position coordinates to form spatial distribution data.

[0035] In some embodiments of this application, in order to identify a set of voxels that reflect the internal void structure of the stockpile from the overall spatial framework, the position coordinates of each void voxel in the goaf stockpile are typically extracted from all spatial position coordinates. By filtering the value states of voxels in space, voxels with void attributes can be separated from the overall voxel set, thus clarifying the positional relationships of void voxels in three-dimensional space. Using these void voxel position coordinates, spatial distribution data describing the void distribution within the stockpile can be formed, providing fundamental data for subsequent construction of local spatial extents and void structure analysis.

[0036] In a specific example, after determining the spatial coordinates of the voxels, all voxels in the 3D voxel mesh can be screened one by one. Voxels with void properties are marked as target voxels, and their spatial coordinates are recorded. As the screening process unfolds sequentially throughout the voxel set, a discrete coordinate set composed of void voxels can be formed in 3D space, presenting the void distribution within the packing in a voxelized manner. Based on this discrete coordinate set, corresponding spatial distribution data can be generated, providing the necessary data foundation for subsequent void structure analysis and type identification.

[0037] Optionally, sub-step 2012 includes the following sub-steps: Sub-step 20121 involves performing multiple voxel attribute extraction processes on the spatial location coordinates based on a progressively decreasing scale division threshold, in order to extract multiple sets of voxel-based location coordinates from all spatial location coordinates.

[0038] In some embodiments of this application, in order to identify sets of voxels with void features within an accumulation at different spatial scales, the spatial location coordinates are typically subjected to multiple voxel attribute extraction processes based on progressively decreasing scale division thresholds. By dividing the local spatial relationships of voxels at larger, medium, and smaller scales respectively, the attribute performance of voxels at different scales can be presented, thereby forming multiple sets of voxel extraction location coordinates, providing a multi-scale spatial basis for subsequently constructing a set of void voxel location coordinates.

[0039] For a specific example, please refer to Figure 3 Multi-scale voxel attribute extraction is performed on the spatial location coordinates. First, voxel attribute analysis is performed on the spatial location coordinates at a larger scale. Interactive Threshold Segmentation (ITS) is used to extract fracture regions with significant gray-level differences and strong structural ductility (such as in...). Figure 3(As indicated by the red dashed box in the middle), corresponding to the coordinates of the first set of voxel extraction positions. Subsequently, the same spatial region can be further processed at a medium scale, using Top-hat Morphological Operation (Top-hat) to extract microcrack regions with weak grayscale contrast, unclear boundaries, or fine distribution (such as in...). Figure 3 (As indicated by the yellow dashed box in the image), forming the second set of voxel extraction location coordinates. After further scaling down, the Watershed Algorithm (WA) can be used at even smaller scales to extract discontinuously distributed or excessively small cracks that are close to the background grayscale (such as those in...). Figure 3 (As shown in the green dashed box), this corresponds to the formation of the third set of voxel extraction position coordinates. As the above multi-scale extraction is carried out sequentially, multiple sets of voxel extraction position coordinates can be obtained from all spatial position coordinates, providing multi-scale voxel extraction results for subsequent combination to form a set of gap voxel position coordinates.

[0040] Sub-step 20122 combines the extracted position coordinates of multiple voxels into a set of void voxel position coordinates for spatial position coordinates.

[0041] In some embodiments of this application, in order to obtain a set of voxels that can fully reflect the distribution of voids inside the accumulation, the voxel extraction location coordinates obtained from multiple voxel attribute extractions are usually combined. Since the voxel extraction results at different scales correspond to void regions with different grayscale features and different structural scales, unifying and integrating these voxel extraction location coordinates can form a set of void voxel location coordinates covering void features at multiple scales, providing a complete spatial basis for subsequent determination of voxel void attributes.

[0042] In a specific example, multiple sets of voxel extraction coordinates obtained based on thresholds at different scales can be integrated according to their coordinate relationships in three-dimensional space. For voxel coordinates extracted at larger, medium, and smaller scales respectively, they can be unified into a continuous set of voxel coordinates through methods such as coordinate merging, duplicate coordinate removal, or merging by spatial neighborhood. As multiple sets of voxel extraction coordinates are sequentially added and integrated, a set of void voxel coordinates covering void regions at different scales can be formed, providing a complete coordinate basis for subsequently determining the void properties of each voxel in the accumulation.

[0043] Sub-step 20123: Determine the void properties of each voxel in the goaf accumulation based on the void voxel location coordinate set, and use the spatial Boolean data used to characterize all void properties as spatial distribution data.

[0044] In some embodiments of this application, in order to clearly define the porosity properties of each voxel within the stack in three-dimensional space, the properties of all voxels are typically determined based on the set of void voxel position coordinates. By labeling voxels in the void voxel position coordinate set as void voxels and the remaining voxels as non-void voxels, the porosity distribution within the stack can be fully expressed in a voxelized manner. Based on the above property determination results, spatial Boolean data characterizing the porosity properties of all voxels can be constructed, and this spatial Boolean data can be used as spatial distribution data to provide a unified data foundation for subsequent local structural analysis.

[0045] In a specific example, all voxels in the 3D voxel mesh can be traversed, and their spatial positions can be compared with the coordinate set of void voxels. For voxels with matching coordinates, their attributes can be marked as void attributes; for voxels that do not match, their attributes are marked as non-void attributes. As the attributes of all voxels are determined one by one, 3D binary void data composed of 0s and 1s can be formed in 3D space, where 1 represents a void voxel and 0 represents a non-void voxel. This 3D binary void data can serve as spatial distribution data to characterize the void distribution within an accumulation, providing basic data for subsequent void structure analysis and type identification.

[0046] Step 202: Based on the spatial distribution data of voids in the goaf accumulation, determine the void envelope region corresponding to each void voxel in the spatial distribution data.

[0047] The method shown in this step has been explained in step 101 and will not be repeated here.

[0048] Optionally, step 202 includes the following sub-steps: Sub-step 2021 involves extracting multiple voxel coordinates centered on each vacancy voxel to form a voxel matrix corresponding to each vacancy voxel.

[0049] In some embodiments of this application, in order to characterize the voxel structure around the void voxel within a local spatial range, multiple voxel coordinates are typically extracted centered on the spatial coordinates of each void voxel to form a corresponding voxel matrix. By selecting voxel coordinates within the three-dimensional neighborhood of the central voxel according to a preset neighborhood scale, the neighborhood can present a regular cubic array structure in space, thereby providing a stable local spatial basis for subsequently determining the void envelope region.

[0050] In a specific example, the segmented 3D void data can be exported as a slice sequence file. For instance, a tagged image file format (TIFF) can be used, which can relatively completely preserve the grayscale or binary information of the 2D slices, has a clear hierarchical relationship, and is highly compatible with common 3D reconstruction and analysis environments. This allows for layer-by-layer reading and maintains spatial hierarchical consistency during subsequent processing. After obtaining the slice sequence, pixels with a value of 1 in each slice can be traversed one by one, and those that meet the spatial coordinate conditions can be recorded as the spatial coordinates of void voxels. After obtaining the spatial positions of the void voxels, a fixed-size 3D neighborhood can be constructed centered on the coordinates (i, j, k) of each void voxel. For example, three voxel units can be extended in each of the three coordinate axes, centered on i, j, and k, in both positive and negative directions, forming a 7×7×7 cubic neighborhood (i.e., covering the spatial range of i±3, j±3, k±3). The coordinates of all voxels within this neighborhood are then used as the voxel matrix corresponding to that void voxel. As the aforementioned neighborhood construction process unfolds sequentially across all void voxels, a voxel matrix corresponding to each void voxel can be formed, providing the necessary local voxel set for subsequent determination of the void envelope region.

[0051] In fact, the 7×7×7 neighborhood size is the preferred scale selected after comprehensively considering local feature stability, computational accuracy, and computational cost. Compared to a 3×3×3 or 5×5×5 neighborhood, this scale can contain more local void voxels, thus more stably supporting the subsequent calculation of principal directions, principal axis fitting, and boundary undulation parameters; while compared to a 9×9×9 or larger neighborhood, it can avoid excessive averaging of local structures and effectively control computational complexity while maintaining locality. Therefore, in this embodiment, a 7×7×7 neighborhood is preferentially used as the scale for constructing the voxel matrix.

[0052] Sub-step 2022: Determine the spatial region occupied by each voxel matrix as the void envelope region of the corresponding void voxel.

[0053] In some embodiments of this application, after obtaining voxel squares corresponding to each void voxel, the spatial range covered by each voxel square can be defined as the void envelope region of that void voxel. By using the three-dimensional neighborhood of the voxel squares as the local spatial boundary, each void voxel can obtain an independent and fixed local analysis range, thereby providing a consistent spatial basis for subsequent extraction of directional features, aggregation features, and boundary undulation features.

[0054] In a specific example, for a constructed 7×7×7 voxel matrix, the coordinate range corresponding to this matrix in three-dimensional space can be directly used as the void envelope region of the void voxel. For instance, when the spatial coordinates of a void voxel are (i, j, k), its corresponding voxel matrix covers the spatial range of i±3, j±3, k±3. Then, the three-dimensional region formed by all voxels within this range can be regarded as the void envelope region of the void voxel. As the above spatial ranges are determined one by one, each void voxel can obtain its independent local spatial region, providing stable spatial support for subsequent structural feature calculations based on local neighborhoods.

[0055] Step 203: Determine the directional feature data, clustering feature data, and boundary undulation feature data of each void voxel based on each void envelope region.

[0056] Among them, the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the boundary voxels of the voids in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation.

[0057] The method shown in this step has been explained in step 102 and will not be repeated here.

[0058] Optionally, step 203 includes the following sub-steps: Sub-step 2031: Based on the spatial position coordinates of the void voxels contained in each void envelope region, determine the local void voxel coordinate set corresponding to each void envelope region, and based on each local void voxel coordinate set, determine the local boundary voxel coordinate set corresponding to each void envelope region.

[0059] In some embodiments of this application, after obtaining the void envelope regions corresponding to each void voxel, a local void voxel coordinate set corresponding to the void envelope region can be determined based on the spatial position coordinates of the void voxels contained in each void envelope region. By identifying voxels with a value of 1 within the three-dimensional range of the void envelope region, the local void voxels can be explicitly represented in space as a coordinate set. After obtaining the local void voxel coordinate set, a corresponding local boundary voxel coordinate set can be further determined based on this coordinate set, so that each void envelope region has a local void voxel set and a local boundary voxel set for subsequent feature calculations.

[0060] In a specific example, for a given void voxel, its void envelope region can be defined by a 7×7×7 voxel matrix, covering a three-dimensional region of i±3, j±3, and k±3. Within this region, the values ​​of all voxels can be examined one by one, and the coordinates of voxels with a value of 1 can be collected as the local void voxel coordinate set corresponding to that void voxel. Subsequently, the boundary voxels in three-dimensional space can be identified based on this local void voxel coordinate set. For example, voxels adjacent to non-void voxels or located at the outer edge of the local void voxel set can be marked as local boundary voxels, and their spatial coordinates can be collected as the local boundary voxel coordinate set. As the above coordinate sets are determined one by one, each void envelope region can obtain its corresponding local void voxel coordinate set and local boundary voxel coordinate set, providing the necessary local spatial data for subsequent calculations of directional features, clustering features, and boundary undulation features.

[0061] Sub-step 2032: Determine the directional feature data of each void voxel based on the void voxel coordinate set corresponding to each void voxel.

[0062] In some embodiments of this application, after obtaining the local void voxel coordinate sets corresponding to each void voxel, the directional feature data of each void voxel can be determined based on the coordinate sets. By analyzing the distribution of local void voxels in three-dimensional space, the principal direction vector reflecting the extension trend of the local void structure can be obtained, and directional feature data for characterizing directional consistency can be constructed accordingly, so that each void voxel has a numerical description that can reflect the directional characteristics of its local structure.

[0063] In a specific example, for a given void voxel, the distribution of its corresponding local void voxel coordinate set in three-dimensional space can be statistically analyzed. For instance, the principal direction vector can be calculated based on the local void voxel coordinate set to reflect the main extension direction of the local void structure in space. Subsequently, this principal direction vector can be compared with the direction vectors of each voxel in the local void voxel coordinate set, and directional feature data can be constructed based on the consistency of their directions. As the aforementioned directional feature data is determined one by one, each void voxel can obtain directional feature data that reflects the directional characteristics of its local void structure.

[0064] Optionally, sub-step 2032 includes the following sub-steps: Sub-step 20321: Determine the principal direction vector values ​​of the voxels corresponding to each coordinate in the void voxel coordinate set.

[0065] In some embodiments of this application, the principal direction vector values ​​of the voxels corresponding to these coordinates can be determined based on the distribution of each coordinate in the three-dimensional space within the void voxel coordinate set. By analyzing the arrangement trend of void voxels in local space, each voxel can obtain a principal direction vector value reflecting the main extension direction of its local void structure, thereby providing a foundation for subsequent construction of directional feature data.

[0066] In a specific example, the neighborhood distribution of the void voxel coordinate set in three-dimensional space can be analyzed. For instance, the coordinates of each voxel within its local range can be statistically analyzed, and the principal direction vector value of the voxel can be determined based on the spatial extension trend of these coordinate points. For regions with linearly extending local structures, the principal direction vector value can be consistent with the main extension direction of the region; for regions with sheet-like or strip-like local structures, the principal direction vector value can reflect the main arrangement direction of the region. As the above principal direction vector values ​​are determined one by one on all void voxels, a set of principal direction vectors can be formed for subsequent calculation of directional features.

[0067] Sub-step 20322: The normalized mean of the outer product of the central principal direction vector value and each principal direction vector value among all principal direction vector values ​​is determined as the directional feature data.

[0068] In some embodiments of this application, the directional consistency between the principal direction vector value of the central voxel and all principal direction vector values ​​can be determined based on the principal direction vector values ​​of each voxel in the void voxel coordinate set, and directional feature data can be constructed accordingly. The outer product can effectively characterize the consistency between two direction vectors: when the two vectors are aligned or nearly aligned, the magnitude of the outer product approaches zero; when the two vectors deviate significantly, the magnitude of the outer product increases with the deviation angle. Therefore, the outer product can naturally reflect the directional dispersion of the local void structure. To ensure that all principal direction vector values ​​participate in the construction of directional features, all outer product results can be normalized and their mean calculated, so that the directional feature data can comprehensively reflect the overall directional consistency between the central voxel and its neighboring voxels, thereby avoiding the situation where only local extreme directions dominate the evaluation results.

[0069] In a specific example, the principal direction vector value of the central voxel in the void voxel coordinate set can be cross-producted with the principal direction vector values ​​of each voxel. The results of all cross-products are then normalized, and their mean is calculated. This normalized mean is used as the directional feature data of the central voxel. If the principal direction vector values ​​of neighboring voxels are generally consistent with the direction of the central voxel, the normalized mean is lower; if the direction distribution of neighboring voxels is more dispersed, the normalized mean is correspondingly higher. As the above process is performed on all void voxels one by one, a set of directional feature data for subsequent voxel type determination can be formed.

[0070] Based on the above process of constructing directional feature data, we can first summarize the mathematical expression of directional features so that the relationships between local principal directions can be presented in a unified numerical form. For any void voxel x, its corresponding local void voxel coordinate set can be represented as a local neighborhood. And for each void voxel y, calculate its principal direction vector. Meanwhile, the principal direction vector of the central voxel is denoted as After obtaining these principal direction vectors, it is possible to... and The magnitude of the cross product characterizes the degree of deviation between the two in spatial direction; when the two directions are consistent or nearly consistent, the magnitude of the cross product approaches zero, and as the directional deviation increases, the magnitude of the cross product increases accordingly. Based on this, the local directional deviation of the central voxel x can be expressed as... ,in This represents the number of void voxels within a local neighborhood. Furthermore, it can be applied to all... Calculate the average value to obtain the directional feature data of the central voxel x. : .

[0071] The calculation of the aforementioned directional characteristics depends on the principal direction vectors of the central voxel and its neighboring voxels, which can be obtained by principal component analysis of the local void voxel coordinate set. For any void voxel x, its voxel coordinates can be set as follows: Construct a local neighborhood centered on this voxel. And extract the coordinate set of all void voxels in the neighborhood, denoted as .

[0072] The positions of all void voxels in the local neighborhood can be counted and arranged into a coordinate matrix. Then, based on this, the centroid of the local point set is calculated. And decentralize each point: Finally, all the decentralized points are combined into a matrix. .

[0073] Furthermore, the covariance matrix of the local point set can be calculated based on this. ,in For one Matrix. Eigenvalue decomposition of the covariance matrix: , and according to Sort the eigenvalues, where the largest eigenvalue is... corresponding feature vector This represents the direction of the strongest extension of the local point set. Normalizing this eigenvector yields the principal direction vector of the central voxel x: For other void voxels y in the local neighborhood, the corresponding local void voxel coordinate set can be constructed in the same way, the covariance matrix can be calculated and eigenvalue decomposition can be performed, thereby obtaining the principal direction vector of each voxel. In obtaining and Then, you can substitute it into the formula to calculate. and This enables directional feature data to be generated in a unified manner.

[0074] It should be noted that in the process of constructing directional feature data based on local principal direction vectors, the distribution of eigenvalues ​​of the covariance matrix directly affects the determination of the principal direction vectors, thus affecting the calculation of directional feature data. When performing principal component analysis on the local void voxel coordinate set, if the three eigenvalues ​​of the covariance matrix satisfy... This indicates that the local point set is similarly dispersed in the three directions, and the overall distribution tends to be clumpy or cavity-like. Such structures typically correspond to porous regions, and their principal direction vectors do not have a clear extension trend; therefore, the directional deviation of this region can be directly assigned a value of 0. If two eigenvalues ​​are equal and both greater than the third eigenvalue, i.e. This indicates that the local point set is more dispersed in two directions and less dispersed in the third direction, tending towards a sheet-like or planar distribution. This type of structure usually corresponds to a crack propagation surface, and the directional deviation can be directly assigned a value of 1 to reflect its significant directional extension. If the three eigenvalues ​​show a clear unimodal structure, i.e. This indicates that the local point set mainly extends along a certain direction. Such structures usually correspond to slender cracks or narrow channels. In this case, the principal direction vector can be obtained by continuing to follow the process of covariance matrix and eigenvalue decomposition described above. and And further calculate based on the formula and By classifying the eigenvalue distribution as described above, different types of local structures can be represented in a unified numerical form in the directional feature data, which facilitates automated analysis and discrimination along with other features in subsequent steps.

[0075] Sub-step 2033: Determine the clustering feature data of each void voxel based on the void voxel coordinate set and the local boundary voxel coordinate set corresponding to each void voxel.

[0076] In some embodiments of this application, after obtaining the local void voxel coordinate set and the local boundary voxel coordinate set corresponding to each void voxel, the aggregation characteristic data of each void voxel can be determined based on the two coordinate sets. By analyzing the distribution of local boundary voxels within the local spatial range, a numerical description reflecting the degree of aggregation of local boundary voxels relative to the local void structure can be obtained, so that each void voxel has characteristic data that can characterize its local structural aggregation.

[0077] In a specific example, for a given void voxel, a reference datum for the local void structure can first be determined based on its corresponding local void voxel coordinate set. For example, its geometric center or local principal axis direction can be calculated based on the local void voxel coordinate set. Subsequently, the distance from each boundary voxel to this reference datum can be calculated based on the local boundary voxel coordinate set, and the distance values ​​can be normalized to reflect the degree of aggregation of boundary voxels within the local region. Further, aggregation feature data can be constructed based on the average normalized distance. For example, the difference between the unit value and the normalized average distance can be used as aggregation feature data to characterize the aggregation of local boundary voxels within the local region. As the above aggregation feature data is determined one by one, each void voxel can obtain aggregation feature data that reflects the degree of aggregation of its local void structure.

[0078] Optionally, sub-step 2033 includes the following sub-steps: Sub-step 20331: Determine the principal axis direction vector of the void voxel coordinate set, and determine the principal axis of the void voxel coordinate set based on the principal axis direction vector and the centroid position of the local void voxel coordinate set.

[0079] In some embodiments of this application, the principal axis direction vector of the void voxel coordinate set can be determined based on the overall distribution of the void voxel coordinate set in three-dimensional space. Furthermore, the principal axis of the coordinate set is determined based on the principal axis direction vector and the centroid position of the local void voxel coordinate set. The principal axis direction vector reflects the main extension trend of the coordinate set in space, while the centroid position provides a reference for the overall position of the coordinate set in space. The combination of these two elements forms the principal axis of the coordinate set, which characterizes the overall extension direction of the local void structure, providing a basis for subsequent calculations of the spatial relationship between local boundary voxels and the principal axis.

[0080] In a specific example, the distribution of the void voxel coordinate set in three-dimensional space can be analyzed. Based on the spatial extension trend of each voxel coordinate, the principal axis direction vector of the coordinate set can be determined. The centroid position of the local void voxel coordinate set can be used as the reference point of the principal axis, so that the principal axis direction vector and the centroid position together constitute the principal axis of the void voxel coordinate set. As the above process is performed one by one at each void voxel position, a set of principal axes of the coordinate set can be formed for subsequent off-axis calculation.

[0081] Sub-step 20332: The normalized mean of the distance from each voxel coordinate in the local boundary voxel coordinate set to the principal axis of the coordinate set is determined as the mean of the off-axis degree of the local boundary voxel coordinate set, and the difference between the unit value and the mean of the off-axis degree is determined as the clustered feature data.

[0082] In some embodiments of this application, the mean off-axis degree of the local boundary voxel coordinate set can be determined based on the spatial relationship between the local boundary voxel coordinate set and the principal axis of the coordinate set, and further, clustering feature data can be constructed. The distance of the local boundary voxel to the principal axis can reflect the degree of clustering of the boundary voxels near the principal axis direction: the smaller the distance, the more concentrated the boundary voxels are near the principal axis; the larger the distance, the more obvious the discrete distribution of the boundary voxels in space. In order to make the distance values ​​of different voxel positions comparable, all distances can be normalized to fall into a uniform numerical range, and by calculating the mean of the normalized distance, all boundary voxel coordinates can participate in the evaluation of the degree of clustering of the local structure, thereby avoiding the overall judgment being dominated by only local extreme distances. Finally, the difference between the unit value and the mean off-axis degree can be used as the clustering feature data, so that the clustering feature can reflect the degree of concentration of the boundary voxels in the principal axis direction in a more intuitive way: the smaller the mean off-axis degree, the larger the clustering feature value, indicating that the boundary voxels tend to cluster around the principal axis.

[0083] In a specific example, the coordinates of each voxel in the local boundary voxel coordinate set can be projected onto the principal axis of the coordinate set, and the shortest distance from each voxel coordinate to the principal axis can be calculated. After normalizing all distances, their mean values ​​are calculated, and this mean value is used as the mean off-axis degree of the local boundary voxel coordinate set. The difference between the unit value and this mean off-axis degree is determined as the corresponding clustered feature data. As the above process is performed one by one at all void voxel locations, a set of clustered feature data for subsequent voxel type determination can be formed.

[0084] Based on the above process of constructing clustered feature data, we can first summarize the mathematical expression of clustered features, so that the spatial clustering degree of local boundary voxels relative to the principal axes of the coordinate set can be presented in a unified numerical form. For any void voxel x, its corresponding local boundary voxel coordinate set can be represented as follows: The principal axes of the coordinate set are determined based on the local void voxel coordinate set. For any voxel in the local boundary voxel coordinate set, its shortest distance to the principal axis of the coordinate set can be calculated. And the maximum distance from the local boundary voxel to the principal axis As a normalization scale, the clustering characteristics of the central voxel x can be represented as follows: , in, This represents the number of local boundary voxels. The above mathematical expression transforms the spatial distribution of local boundary voxels around the principal axis into a directly processable numerical form, enabling clustered features to be generated and used in a consistent manner in discrete voxel space.

[0085] Furthermore, since the aforementioned section on directional features has already provided a complete mathematical derivation of the process for obtaining the principal direction vector of the local void voxel coordinate set, the same formula will not be repeated here. Only a brief explanation of how to obtain the principal axis direction vector will be given: Specifically, a covariance matrix can be constructed based on the local void voxel coordinate set, and then its eigenvalues ​​can be obtained through eigenvalue decomposition. The eigenvectors and their corresponding eigenvalues ​​are defined, with the eigenvector corresponding to the largest eigenvalue representing the direction of the strongest extension of the local point set. This eigenvector, after normalization, can be used as the principal axis direction vector, and the principal axis of the coordinate set is established with the centroid of the local void voxel coordinate set as the passing point. After obtaining the principal axis, the shortest distance from each boundary voxel to the principal axis can be calculated. Furthermore, based on the aforementioned mathematical expression, clustered feature data are calculated. .

[0086] It should be noted that in the process of constructing the principal axes of the coordinate set based on the local void voxel coordinate set, the distribution of the eigenvalues ​​of the covariance matrix directly affects the determination of the principal axis direction vector, thus affecting the calculation of clustered feature data. When performing principal component analysis on the local void voxel coordinate set, if the three eigenvalues ​​of the covariance matrix satisfy... This indicates that the local structure mainly extends along one direction, while the other two directions are similarly discrete. In this case, the eigenvector corresponding to the largest eigenvalue can still be used as the principal axis direction vector. If two eigenvalues ​​are equal and both are greater than the third eigenvalue, that is... This indicates that the local structure is highly discrete in two directions, but significantly less discrete in the third direction, tending towards a sheet-like or planar distribution. Such structures typically correspond to crack propagation surfaces, and the aggregation characteristic data can be directly assigned a value of 1 to reflect the high concentration of boundary voxels near the principal axes. If the three characteristic values ​​satisfy... This indicates that the local structure exhibits relatively small differences in dispersion across the three directions, tending towards a clustered or cavity-like distribution. Such structures typically correspond to porous regions, and the clustering feature data can be directly assigned a value of 0 without further distance normalization or averaging. Through the above classification of feature value distributions, different types of local structures can be represented in a unified numerical form within the clustering feature data, facilitating automated analysis and discrimination along with other features in subsequent steps.

[0087] Sub-step 2034: Determine the boundary undulation feature data of each void voxel based on the local boundary voxel coordinate set corresponding to each void voxel.

[0088] In some embodiments of this application, after obtaining the local boundary voxel coordinate sets corresponding to each void voxel, the boundary undulation characteristic data of each void voxel can be determined based on the coordinate sets. By analyzing the geometric distribution of the local boundary voxels in three-dimensional space, a numerical description reflecting the degree of undulation of the local boundary surface can be obtained, so that each void voxel has characteristic data that can characterize the smoothness or undulation of its local boundary.

[0089] In a specific example, for a given void voxel, the principal plane of the local boundary voxel in 3D space can be determined based on its corresponding local boundary voxel coordinate set. For instance, a local principal plane reflecting the overall trend of the local boundary can be fitted based on the local boundary voxel coordinate set. Subsequently, the local boundary voxel can be projected onto this principal plane, and the area of ​​the projected region can be calculated. Simultaneously, the residual of the local boundary voxel relative to the principal plane can be calculated based on its offset in the normal direction of the principal plane, and the residual can be normalized to reflect the degree of undulation of the local boundary voxel in the normal direction. Further, boundary undulation feature data can be constructed based on the average value of the normalized residual to characterize the overall undulation level of the local boundary surface. As the above boundary undulation feature data are determined one by one, each void voxel can obtain boundary undulation feature data that reflects the degree of undulation of its local boundary surface.

[0090] Optionally, sub-step 2034 includes the following sub-steps: Sub-step 20341: Determine the principal plane of the coordinate set and the boundary area of ​​the coordinate set based on the local boundary voxel coordinate set.

[0091] In some embodiments of this application, the principal plane of the local boundary voxel coordinate set can be determined based on its distribution in three-dimensional space, and the boundary area of ​​the coordinate set on the principal plane can be further determined. The principal plane reflects the overall morphological trend of the local boundary voxels in space, allowing the main extension direction of the local boundary structure in three-dimensional space to be presented in two-dimensional form; while the boundary area can characterize the overall coverage of the local boundary voxels on the principal plane, providing a basis for subsequent calculation of the projected area of ​​each voxel on the principal plane and further construction of boundary undulation feature data. By jointly determining the principal plane and the boundary area, the overall morphology of the local boundary structure can be quantitatively expressed.

[0092] In a specific example, the distribution of the local boundary voxel coordinate set in three-dimensional space can be analyzed. Based on the spatial extension trend of each voxel coordinate, the principal plane of the coordinate set can be determined, and all boundary voxel coordinates can be projected onto this principal plane to form a projection profile. Then, the boundary area of ​​the coordinate set on the principal plane can be determined according to the outer boundary of the projection profile. As the above process is performed one by one at each void voxel location, a principal plane and its corresponding boundary area set can be formed for subsequent boundary undulation feature calculation.

[0093] Sub-step 20342: Determine the principal plane boundary projection area of ​​each voxel in the local boundary voxel coordinate set based on the principal plane of the coordinate set.

[0094] In some embodiments of this application, the boundary projected area of ​​each voxel in the local boundary voxel coordinate set on the principal plane can be determined based on the principal plane of the coordinate set. The principal plane can unfold the distribution of local boundary voxels in three-dimensional space in two-dimensional form, so that the projected area of ​​the voxels on the principal plane can reflect their spatial occupancy in the local boundary structure. By calculating the projected area of ​​each voxel on the principal plane, the relative contribution of the voxel to the local boundary surface can be obtained, providing the necessary area quantification basis for subsequent construction of boundary undulation feature data.

[0095] In a specific example, the coordinates of each voxel in the local boundary voxel coordinate set can be projected onto the principal plane of the coordinate set, and the boundary projected area of ​​each voxel on the principal plane can be determined based on the projected geometry. As the above process is performed on all boundary voxels one by one, a set of principal plane boundary projected areas can be formed for subsequent boundary undulation feature calculations.

[0096] Sub-step 20343 determines the average normalized residual ratio of the projected area of ​​each principal plane boundary to the boundary area of ​​the coordinate set as the boundary undulation feature data.

[0097] In some embodiments of this application, the boundary undulation characteristic data of the local boundary voxel coordinate set can be determined based on the relationship between the projected area of ​​the local boundary voxel on the principal plane and the boundary area of ​​the coordinate set. The normalized residual ratio between the projected area of ​​the principal plane boundary of each voxel and the boundary area of ​​the coordinate set can reflect the degree of undulation of the voxel in the local boundary structure: when the projected area of ​​the voxel on the principal plane is relatively large, its residual ratio is small, indicating that the voxel is closer to the overall shape of the local boundary; when the projected area is relatively small, its residual ratio is large, indicating that the voxel has more obvious undulation or deviation on the local boundary surface. By averaging the normalized residual ratios of all voxels, the boundary undulation characteristic data can comprehensively reflect the overall undulation level of the local boundary voxels in the principal plane direction, thereby avoiding the evaluation results being dominated only by local extreme voxels, and enabling the boundary undulation characteristics to more stably characterize the overall smoothness or undulation degree of the local boundary structure.

[0098] In a specific example, the ratio of the projected area of ​​the principal plane boundary of each voxel in the local boundary voxel coordinate set to the boundary area of ​​the coordinate set can be calculated. The normalized residual percentage is then obtained by subtracting this ratio from the unit value. The average residual percentage of all voxels is then calculated, and this average is used as the boundary undulation feature data of the local boundary voxel coordinate set. As the above processing is performed sequentially at all void voxel locations, a boundary undulation feature data set for subsequent voxel type determination can be formed.

[0099] Based on the above process of constructing boundary undulation feature data, the degree of boundary undulation of the central voxel x can first be represented in a unified mathematical form. For any void voxel x, its corresponding local boundary voxel coordinate set can be represented as follows: And based on the local true boundary area With the local principal plane projection area Construct the boundary undulation factor. The boundary undulation characteristic data of the central voxel x can be represented as: This expression directly reflects the area difference between the local true boundary and its projection onto the principal plane. The larger the area difference, the more pronounced the boundary undulations. In the above process, to obtain the area difference, it is necessary to calculate the area of ​​the local true boundary and the projected area onto the principal plane separately. The area of ​​the local true boundary can be obtained by counting the number of interfaces between void voxels and solid voxels in the local neighborhood and multiplying them by the area of ​​the corresponding voxel surface; let the dimensions of the voxels in the three directions be... , , Then the local true boundary area can be expressed as ,in , , These represent the number of boundary surfaces parallel to the xy, xz, and yz planes within the local neighborhood, respectively. This area quantization method can directly reflect the true geometric scale of the local boundaries.

[0100] After obtaining the true boundary area, the local boundary voxels can be projected onto the local principal plane to obtain the projected area. The method for obtaining the principal plane normal vector is consistent with the process for obtaining the principal direction vector in the aforementioned directional and clustering features. Both are based on the eigenvalue decomposition of the local point set covariance matrix. Therefore, the formula derivation will not be repeated here, but only briefly explained: Specifically, a covariance matrix can be constructed based on the local boundary voxel coordinate set, and then its eigenvalues ​​can be obtained through eigenvalue decomposition. The eigenvectors corresponding to the smallest eigenvalues ​​can be used as the normal vectors of the local principal plane. After determining the principal plane using this normal vector and the centroids of the local boundary voxel coordinate set, the boundary voxels can be projected onto the principal plane to obtain the projected area. Furthermore, boundary undulation characteristic data are calculated based on the formula. .

[0101] It should be noted that in the process of constructing the principal plane based on the local boundary voxel coordinate set and calculating the boundary undulation feature data, the distribution of the eigenvalues ​​of the covariance matrix directly affects the determination of the principal plane normal vector, thus affecting the calculation of the boundary undulation feature data. When performing principal component analysis on the local boundary voxel coordinate set, if the three eigenvalues ​​of the covariance matrix satisfy... This indicates that the local boundary point set is highly discrete in two directions, but less discrete in the third direction, tending towards a sheet-like or planar distribution. This type of structure typically corresponds to a crack propagation surface, and the boundary undulation feature data can be directly assigned a value of 0 to reflect its relatively smooth boundary surface. If the feature value satisfies... This indicates that the local boundary point set mainly extends along one direction, while the dispersion in the other two directions is smaller and similar, and the overall distribution tends to be linear or elongated channel-like. This type of structure usually corresponds to elongated cracks, and the boundary undulation feature data can also be assigned a value of 0. If the three feature values ​​satisfy... This indicates that the local boundary point set exhibits relatively small differences in dispersion across the three directions, tending towards a clumping, cavity-like, or spherical distribution. Such structures typically correspond to porous regions, and the boundary undulation feature data can be directly assigned a value of 1 without further area difference calculation. Through the above classification processing of feature value distribution, different types of local structures can be represented in a unified numerical form in the boundary undulation feature data, facilitating automated analysis and discrimination along with other features in subsequent steps.

[0102] Step 204: Determine the voxel type discrimination label for each voxel based on the directional feature data, clustering feature data, and boundary undulation feature data of each voxel.

[0103] The method shown in this step has been explained in step 103 and will not be repeated here.

[0104] Optionally, step 204 includes the following sub-steps: Sub-step 2041 uses the weighted average of directional feature data, clustered feature data, and boundary undulation feature data as the voxel type discrimination index value for void voxels.

[0105] In some embodiments of this application, after obtaining directional feature data, aggregation feature data, and boundary undulation feature data corresponding to each void voxel, a voxel type discrimination index value for distinguishing different void structure types can be constructed based on the above three types of feature data. By weighted combination of the three types of feature data, each type of feature can play a corresponding role in the index construction process according to its importance to the local structure discrimination, thereby forming a discrimination index that can comprehensively reflect the local structural attributes of void voxels.

[0106] In a specific example, for a given void voxel, its directional feature data, clustering feature data, and boundary undulation feature data can be acquired separately, and then weighted according to preset weights. For instance, the directional feature data, clustering feature data, and boundary undulation feature data can be multiplied by their respective weight coefficients, and the weighted results can be summed to obtain the voxel type discrimination index value for that void voxel. As these index values ​​are determined one by one, each void voxel can obtain a voxel type discrimination index value that comprehensively reflects its local structural directional features, clustering features, and boundary undulation features, providing a foundation for subsequent determination of voxel type discrimination labels.

[0107] Sub-step 2042: If the voxel type discrimination index value is greater than the preset crack discrimination threshold, the voxel type discrimination label of the void voxel is determined as the crack label.

[0108] In some embodiments of this application, after obtaining the voxel type discrimination index value for each void voxel, the index value can be judged based on a preset crack determination threshold. When the voxel type discrimination index value of a void voxel is greater than the crack determination threshold, the voxel type discrimination label of the void voxel can be determined as a crack label. In this way, void voxels with significant local structural features, strong directionality and aggregation, or high degree of boundary undulation can be classified as crack voxels during the discrimination process, thereby achieving effective identification of crack structures in three-dimensional space.

[0109] In a specific example, for a given void voxel, a voxel type discrimination index value can be calculated based on its directional feature data, clustering feature data, and boundary undulation feature data according to preset weights. For instance, when the voxel type discrimination index value of the void voxel is 0.78, and the preset fracture determination threshold is 0.65, since the index value is greater than the threshold, the voxel type discrimination label of the void voxel can be determined as a fracture label. As the above discrimination process is carried out sequentially on all void voxels, void voxels with higher index values ​​can be identified as fracture voxels in three-dimensional space, providing a foundation for subsequent fracture structure extraction and analysis.

[0110] Sub-step 2043: If the voxel type discrimination index value is less than or equal to the crack discrimination threshold, the voxel type discrimination label of the void voxel is determined as the pore label.

[0111] In some embodiments of this application, after obtaining the voxel type discrimination index value for each void voxel, the index value can be judged based on a preset crack determination threshold. When the voxel type discrimination index value of a void voxel is less than or equal to the crack determination threshold, the voxel type discrimination label of the void voxel can be determined as a pore label. In this way, void voxels with weak local structural directionality, insignificant aggregation, or low boundary undulation can be classified as pore voxels during the discrimination process, thereby achieving effective differentiation of non-crack void structures in three-dimensional space.

[0112] In a specific example, for a given void voxel, a voxel type discrimination index value can be calculated based on its directional feature data, clustering feature data, and boundary undulation feature data according to preset weights. For instance, when the voxel type discrimination index value of the void voxel is 0.42, and the preset fracture determination threshold is 0.65, since the index value is less than or equal to the threshold, the voxel type discrimination label of the void voxel can be determined as a pore label. As the above discrimination process is carried out sequentially on all void voxels, void voxels with lower index values ​​can be identified as pore voxels in three-dimensional space, providing a foundation for the subsequent expression and analysis of pore structures.

[0113] Based on the above quantification logic, the three types of features are further organized into a unified fracture tendency function. For any void voxel, the value of its directional characteristic data is The value of clustered feature data The value of the boundary undulation feature data is In this case, corresponding weighting coefficients will be designed. , , To adjust the relative contributions of the three types of features in the overall discrimination: ,in This expression unifies the three types of features to the same scale, enabling them to participate in structure type determination in a consistent manner: The stronger the directionality, the higher the aggregation, and the smoother the boundaries, the more the structure tends to become fractured. , and The direction of change (i.e., monotonicity) is consistent; in addition, the sum of the three terms is divided by 3 to ensure that the tendency index always falls within the [0,1] interval, which facilitates subsequent threshold judgment. Through this weighted combination method, the three types of features can play a role in the overall index according to their importance, so that the index value can comprehensively reflect the directional features, clustering features and boundary features of the local structure.

[0114] Optionally, in some embodiments of this application, the weight coefficients of the weighted mean obtained from the directional feature data, clustered feature data, and boundary undulation feature data can be adaptively inferred by selecting some voxels with clearly defined directional feature data, clustered feature data, and boundary undulation feature data. In this case, this application further includes the following additional steps: Step 2001: Determine the locations of multiple target voxels from the voids in the goaf accumulation; each target voxel location has corresponding target directional feature data, target aggregation feature data, and target boundary undulation feature data.

[0115] In some embodiments of this application, to enable directional feature data, clustering feature data, and boundary undulation feature data to participate in weighted combination in an adaptive manner during the construction of voxel type discrimination index, several representative target voxel locations can first be selected from the voids of the goaf accumulation. Each target voxel location has corresponding target directional feature data, target clustering feature data, and target boundary undulation feature data, which can serve as a reference sample for subsequent estimation of weight coefficients. By selecting these target voxel locations, the solution process for the weight coefficients can reflect the actual feature distribution of the void structure in the goaf accumulation.

[0116] In a specific example, based on the spatial distribution of voids in the goaf accumulation, several voxel locations with significant differences in local structure can be selected as target voxel locations. For example, voxel locations with high directional feature data, moderate aggregation feature data, and low boundary undulation feature data can be selected, or voxel locations with weak directional feature data, high aggregation feature data, or significant boundary undulation feature data can be selected. For each selected target voxel location, its target directional feature data, target aggregation feature data, and target boundary undulation feature data can be recorded separately. As the above target voxel locations are determined one by one, multiple sets of one-to-one corresponding target feature data can be formed, providing a basic sample for subsequent estimation of the weights of directional feature data, aggregation feature data, and boundary undulation feature data.

[0117] Step 2002: Based on the crack determination threshold and multiple sets of one-to-one target directional feature data, target clustering feature data, and target boundary undulation feature data, determine the first weight for the directional feature data, the second weight for the clustering feature data, and the third weight for the boundary undulation feature data, so that under the first weight, the second weight, and the third weight, all the one-to-one target directional feature data, target clustering feature data, and target boundary undulation feature data satisfy the normalization convex condition.

[0118] In some embodiments of this application, the first weight of the directional feature data, the second weight of the clustered feature data, and the third weight of the boundary undulation feature data can be determined based on the crack determination threshold and multiple sets of one-to-one corresponding target directional feature data, target clustered feature data, and target boundary undulation feature data. By generating candidate weights, normalizing the candidate weights, calculating the weighted mean of the target feature data, and verifying the weighted mean group by group, the three types of weights determined in the end can satisfy the normalization convex condition on all target feature data, thereby ensuring that the three types of feature data can participate in the weighted summation in a stable and reasonable manner when constructing the voxel type discrimination index value in the subsequent process.

[0119] In a specific example, it can be followed Figure 4 The program execution logic shown determines three types of weights. Specifically, a set of candidate weights can be generated in step S1. And normalize it to meet the requirements. Subsequently, in step S2, a set of target directional feature data that has not yet been verified is selected from multiple target voxel locations. Target clustering characteristic data and target boundary undulation feature data In step S3, the target feature data and candidate weights are weighted and combined to calculate the weighted mean. In step R1, it is determined whether the weighted mean satisfies the following conditions. If the weighted mean satisfies the above interval conditions, proceed to step S4, where the same set of candidate weights is used to calculate the weighted mean for each of the previously verified target feature data, and in step R2, it is determined whether all verified weighted means satisfy the above conditions. If all the above conditions are met, then in step R3, it is determined whether the verification of all target voxels has been completed; if all target voxels have been verified, then the current candidate weights can be used to determine the next step. Determine the final weights , and ; If there are still unverified target voxels, return to S2 and continue execution (step R3); if the weighted mean does not fall within the 0 to 1 interval during step R1 or step R2, return to step S1 to regenerate new candidate weights and repeat the above verification process. Through the above iterative execution logic, a set of weight combinations that satisfies the normalization convexity condition for all target directional feature data, target clustering feature data, and target boundary undulation feature data can be finally obtained, thereby determining the final values ​​of each weight.

[0120] Step 205: Generate spatial distribution label data for the goaf accumulation based on the voxel type discrimination labels of all void voxels.

[0121] In some embodiments of this application, to ensure a unified type representation of the void structure in the goaf accumulation in three-dimensional space, spatial distribution label data is typically generated based on the voxel type discrimination labels of all void voxels. By mapping the position of each void voxel in space to its type label, a spatial annotation structure composed of multiple types of voxels can be formed in the three-dimensional voxel mesh, allowing for the spatial differentiation between fracture-type voxels and pore-type voxels within the accumulation. This spatial distribution label data provides a unified spatial basis for subsequent structural analysis, visualization, or further parameter extraction, enabling a clear and intuitive presentation of the type distribution relationship of the void structure within the accumulation.

[0122] In a specific example, all void voxels in the spatially distributed data can be sequentially written with their corresponding voxel type labels according to their three-dimensional coordinates, so that the original voxel values ​​and type labels form a joint representation in the same spatial location. Once all voxels have been labeled, an extended tensor composed of the original spatially distributed data and type labels can be obtained in the three-dimensional voxel mesh, allowing the crack-like and pore-like voxels within the packing body to be presented in space as discretized labels. Based on this extended tensor, complete spatially distributed label data can be generated, providing a unified data foundation for subsequent structural analysis and applications.

[0123] In summary, in this embodiment, a corresponding void envelope region is constructed for each void voxel based on the spatial distribution data of voids in the goaf accumulation. This allows the voxel to perform structural analysis within its local spatial neighborhood. With the help of this local range, void identification no longer relies on a single statistical feature at the overall scale, but rather on characterizing the local environment of the voxel itself. This enhances the sensitivity to local structural changes and provides a more targeted spatial basis for subsequent feature extraction. Furthermore, by extracting directional, clustering, and boundary undulation feature data within each void envelope region, the void voxel can obtain fine-grained structural information from multiple dimensions, such as structural orientation consistency, boundary voxel clustering degree, and local boundary undulation. By characterizing and utilizing the local geometric morphology, spatial distribution, and boundary variations reflected by these features, the local differences of complex void structures at different locations are more fully revealed, thereby enhancing the ability to express local structural changes and improving the stability of type identification in areas with complex structural morphology or irregular boundaries. Finally, based on multi-dimensional feature data, voxel-level type identification is performed on each void voxel, enabling the distinction between cracks and pores to be completed at the local scale. This allows the identification process to directly rely on the local structural features of voxels, further identifying structural differences in different local regions within the same connected void volume, reducing instability in type identification caused by the overall scale masking local changes, and improving the ability to express detailed structures. Therefore, the method based on the embodiments of this application, by introducing local envelope regions and combining multi-dimensional feature extraction and voxel-level identification, more effectively captures the local differences in the void structure of complex accumulation bodies, improves the refinement of void type identification, avoids the insufficient identification caused by the overall scale and coarse-grained indicators, and thus improves the accuracy of 3D modeling and structural analysis.

[0124] refer to Figure 5 This application illustrates a device 30 for identifying the void type of a goaf accumulation body, comprising: The neighborhood extraction module 301 is used to determine the void envelope region corresponding to each void voxel in the spatial distribution data based on the spatial distribution data of voids in the goaf accumulation. The feature calculation module 302 is used to determine the directional feature data, clustering feature data, and boundary undulation feature data of each void voxel based on each void envelope region. The directional feature data is positively correlated with the structural orientation consistency of the voids in the goaf accumulation, the clustering feature data is positively correlated with the degree of clustering of the void boundary voxels in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation. The label determination module 303 is used to determine the voxel type discrimination label of each voxel based on the directional feature data, clustering feature data and boundary undulation feature data of each voxel.

[0125] Optionally, the neighborhood extraction module 301 includes: The neighborhood partitioning submodule is used to extract multiple voxel coordinates centered on each void voxel to form a voxel matrix corresponding to each void voxel. The neighborhood extraction submodule is used to determine the spatial region occupied by each voxel 3D matrix as the void envelope region of the corresponding void voxel.

[0126] Optionally, the feature calculation module 302 includes: The coordinate set submodule is used to determine the local void voxel coordinate set corresponding to each void envelope region based on the spatial position coordinates of the void voxels contained in each void envelope region, and to determine the local boundary voxel coordinate set corresponding to each void envelope region based on each local void voxel coordinate set. The directional feature submodule is used to determine the directional feature data of each void voxel based on the void voxel coordinate set corresponding to each void voxel. The clustering feature submodule is used to determine the clustering feature data of each void voxel based on the void voxel coordinate set and the local boundary voxel coordinate set corresponding to each void voxel, respectively. The boundary undulation feature submodule is used to determine the boundary undulation feature data of each void voxel based on the local boundary voxel coordinate set corresponding to each void voxel.

[0127] Optional, the directional feature submodule includes: The principal direction calculation unit is used to determine the principal direction vector value of each voxel corresponding to each coordinate in the void voxel coordinate set; The directional feature unit is used to determine the normalized mean of the cross product of the central main direction vector value and each of the other main direction vector values ​​as directional feature data.

[0128] Optional, the clustering feature submodule includes: The principal axis determination unit is used to determine the principal axis direction vector of the void voxel coordinate set, and to determine the principal axis of the void voxel coordinate set based on the principal axis direction vector and the centroid position of the local void voxel coordinate set. Clustered feature units are used to determine the normalized mean of the distance from each voxel coordinate in the local boundary voxel coordinate set to the principal axis of the coordinate set as the mean off-axis degree of the local boundary voxel coordinate set, and the difference between the unit value and the mean off-axis degree is determined as the clustered feature data.

[0129] Optional, the boundary undulation feature submodule includes: Boundary area unit, used to determine the principal plane of the coordinate set and the boundary area of ​​the coordinate set based on the local boundary voxel coordinate set; Boundary projected area unit, used to determine the principal plane boundary projected area of ​​each voxel in the local boundary voxel coordinate set based on the principal plane of the coordinate set; The boundary undulation feature unit is used to determine the average normalized residual ratio of the projected area of ​​each principal plane boundary to the boundary area of ​​the coordinate set as the boundary undulation feature data.

[0130] Optionally, the label determination module 303 includes: The index calculation submodule is used to use the weighted average of directional feature data, clustered feature data, and boundary undulation feature data as the voxel type discrimination index value of the void voxels. The crack label submodule is used to determine the voxel type identification label of the void voxel as a crack label when the voxel type identification index value is greater than the preset crack identification threshold. The pore tagging submodule is used to identify the voxel type identification tag of a void voxel as a pore tag when the voxel type identification index value is less than or equal to the crack identification threshold.

[0131] Optionally, the identification device 30 for identifying the void type of the goaf accumulation also includes: The reference selection module is used to determine the locations of multiple target voxels from the voids in the goaf accumulation; each target voxel location has corresponding target directional feature data, target aggregation feature data, and target boundary undulation feature data; The parameter back-calculation module is used to determine the first weight for the directional feature data, the second weight for the clustered feature data, and the third weight for the boundary undulation feature data based on the crack determination threshold and multiple sets of one-to-one corresponding target directional feature data, target clustered feature data, and target boundary undulation feature data, so that under the first weight, the second weight, and the third weight, all the one-to-one corresponding target directional feature data, target clustered feature data, and target boundary undulation feature data satisfy the normalization convex condition.

[0132] Optionally, the identification device 30 for identifying the void type of the goaf accumulation also includes: The distribution generation module is used to determine spatial distribution data based on the sampled data of the accumulation distribution of the goaf accumulation.

[0133] Optionally, the distributed generation module includes: The coordinate determination submodule is used to determine the spatial coordinates of each voxel in the goaf accumulation based on the sampling data of the accumulation distribution of the goaf accumulation. The distribution generation submodule is used to extract the void voxel position coordinates of each void voxel in the goaf accumulation from all spatial position coordinates to form spatial distribution data.

[0134] Optionally, the distributed generation submodule includes: The extraction unit is used to perform multiple voxel attribute extraction processes on spatial location coordinates based on a progressively decreasing scale division threshold, so as to extract multiple sets of voxel extraction location coordinates from all spatial location coordinates. The combination unit is used to combine the extracted position coordinates of multiple voxels into a set of void voxel position coordinates for spatial position coordinates; The generation unit is used to determine the void properties of each voxel in the goaf accumulation based on the void voxel position coordinate set, and uses the spatial Boolean data used to characterize all void properties as spatial distribution data.

[0135] Optionally, the identification device 30 for identifying the void type of the goaf accumulation also includes: The tagging module is used to generate spatial distribution tag data for goaf accumulations based on the voxel type of all void voxels.

[0136] In summary, in this embodiment, a corresponding void envelope region is constructed for each void voxel based on the spatial distribution data of voids in the goaf accumulation. This allows the voxel to perform structural analysis within its local spatial neighborhood. With the help of this local range, void identification no longer relies on a single statistical feature at the overall scale, but rather on characterizing the local environment of the voxel itself. This enhances the sensitivity to local structural changes and provides a more targeted spatial basis for subsequent feature extraction. Furthermore, by extracting directional, clustering, and boundary undulation feature data within each void envelope region, the void voxel can obtain fine-grained structural information from multiple dimensions, such as structural orientation consistency, boundary voxel clustering degree, and local boundary undulation. By characterizing and utilizing the local geometric morphology, spatial distribution, and boundary variations reflected by these features, the local differences of complex void structures at different locations are more fully revealed, thereby enhancing the ability to express local structural changes and improving the stability of type identification in areas with complex structural morphology or irregular boundaries. Finally, based on multi-dimensional feature data, voxel-level type identification is performed on each void voxel, enabling the distinction between cracks and pores to be completed at the local scale. This allows the identification process to directly rely on the local structural features of voxels, further identifying structural differences in different local regions within the same connected void volume, reducing instability in type identification caused by the overall scale masking local changes, and improving the ability to express detailed structures. Therefore, the method based on the embodiments of this application, by introducing local envelope regions and combining multi-dimensional feature extraction and voxel-level identification, more effectively captures the local differences in the void structure of complex accumulation bodies, improves the refinement of void type identification, avoids the insufficient identification caused by the overall scale and coarse-grained indicators, and thus improves the accuracy of 3D modeling and structural analysis.

[0137] Reference Figure 6 The electronic device 500 may include one or more of the following components: processing component 502, memory 504, power supply component 506, multimedia component 508, audio component 510, input / output (I / O) interface 512, sensor component 514, and communication component 516.

[0138] Processing component 502 typically controls the overall operation of electronic device 500, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 502 may include one or more processors 520 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 502 may include one or more modules to facilitate interaction between processing component 502 and other components. For example, processing component 502 may include a multimedia module to facilitate interaction between multimedia component 508 and processing component 502.

[0139] Memory 504 is used to store various types of data to support the operation of electronic device 500. Examples of this data include instructions for any application or method operating on electronic device 500, contact data, phonebook data, messages, pictures, multimedia, etc. Memory 504 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0140] Power supply component 506 provides power to various components of electronic device 500. Power supply component 506 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 500.

[0141] Multimedia component 508 includes an interface that provides an output interface between electronic device 500 and user. In some embodiments, the interface may include a liquid crystal display (LCD) and a touch panel (TP). If the interface includes a touch panel, the interface may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may not only sense the boundaries of touch or swipe actions but also detect the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 508 includes a front-facing camera and / or a rear-facing camera. When electronic device 500 is in an operating mode, such as shooting mode or multimedia mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0142] Audio component 510 is used to output and / or input audio signals. For example, audio component 510 includes a microphone (MIC) used to receive external audio signals when electronic device 500 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 504 or transmitted via communication component 516. In some embodiments, audio component 510 also includes a speaker for outputting audio signals.

[0143] Input / output (I / O) interface 512 provides an interface between processing component 502 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0144] Sensor assembly 514 includes one or more sensors for providing state assessments of various aspects of electronic device 500. For example, sensor assembly 514 may detect the on / off state of electronic device 500, the relative positioning of components such as the display and keypad of electronic device 500, changes in position of electronic device 500 or a component of electronic device 500, the presence or absence of user contact with electronic device 500, orientation or acceleration / deceleration of electronic device 500, and temperature changes of electronic device 500. Sensor assembly 514 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 514 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 514 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0145] Communication component 516 facilitates wired or wireless communication between electronic device 500 and other devices. Electronic device 500 can access wireless networks based on communication standards, such as WiFi, carrier networks (such as 2G, 3G, 4G, or 5G), or combinations thereof. In one exemplary embodiment, communication component 516 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 516 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0146] In an exemplary embodiment, the electronic device 500 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to implement the methods provided in the embodiments of this application.

[0147] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 504 including instructions, which can be executed by a processor 520 of an electronic device 500 to perform the above-described method. For example, the non-transitory storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0148] In an exemplary embodiment, the electronic device 500 may also be provided as a server, including a processing component 502, which further includes one or more processors, and memory resources represented by memory 504 for storing instructions, such as applications, that can be executed by the processing component 502. The applications stored in memory 504 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 502 is configured to execute instructions to perform the methods provided in the embodiments of this application.

[0149] Electronic device 500 may also include a power supply component 506 configured to perform power management of electronic device 500, a wired or wireless communication component 516 configured to connect electronic device 500 to a network, and an input / output (I / O) interface 512. Electronic device 500 may operate on an operating system stored in memory 504, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0150] This application also provides a computer program product, including a computer program, which, when executed by a processor, implements the method of this application embodiment.

[0151] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims below.

[0152] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the claimed technical solutions.

[0153] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0154] It will be readily apparent to those skilled in the art that any combination of the above embodiments is feasible. Therefore, any combination of the above embodiments is an implementation scheme of this application. However, due to space limitations, this specification will not describe them in detail here.

[0155] It should be noted that, unless otherwise expressly stated, the methods provided in the embodiments of this application are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. Based on the above description, the required structure for constructing a system having the solutions of this application is obvious. Furthermore, this application is not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0156] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0157] Similarly, it should be understood that, for the purpose of simplification and aiding understanding of one or more aspects of the application, various features of the application are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the application. However, this disclosure should not be construed as reflecting an intention that the claimed application requires more features than are expressly recited in each of the claimed technical solutions. Rather, as reflected in the claimed technical solutions, the application aspects comprise fewer features than all those in the single embodiment disclosed above. Therefore, the claimed technical solutions following the specific implementation are thus expressly incorporated into that specific implementation, wherein each claimed technical solution is itself a separate embodiment of the application.

[0158] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination of all features disclosed in this application and all processes or units of any method or device so disclosed can be employed. Unless expressly stated otherwise, each feature disclosed in this application may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0159] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. For example, in the claimed technical solutions, any one of the claimed embodiments can be used in any combination.

[0160] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the methods according to the embodiments of this application. This application can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such an implementation of this application can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0161] In another embodiment of the present invention, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute the method of the embodiments of the present application.

[0162] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0163] It should be noted that, for the sake of simplicity, the method embodiments of this application are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the embodiments of this application.

[0164] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims below.

[0165] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the claimed technical solutions.

[0166] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A method for identifying the type of voids in a gob pile, characterized by, include: Based on the spatial distribution data of voids in the goaf accumulation, the void envelope region corresponding to each void voxel in the spatial distribution data is determined. Based on each void envelope region, determine the directional feature data, aggregation feature data, and boundary undulation feature data for each void voxel; the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the void boundary voxels in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation; Based on the directional feature data, clustering feature data, and boundary undulation feature data of each void voxel, a voxel type discrimination label is determined for each void voxel.

2. The method of identifying the type of voids in a gob pile according to claim 1, wherein, The determination of directional feature data, clustering feature data, and boundary undulation feature data for each void voxel based on each void envelope region includes: Based on the spatial position coordinates of the void voxels contained in each void envelope region, a local void voxel coordinate set corresponding to each void envelope region is determined, and a local boundary voxel coordinate set corresponding to each void envelope region is determined based on the local void voxel coordinate set. Based on the set of void voxel coordinates corresponding to each void voxel, the directional feature data of each void voxel are determined. Based on the void voxel coordinate set and the local boundary voxel coordinate set corresponding to each void voxel, the aggregation characteristic data of each void voxel are determined respectively. The boundary undulation feature data of each void voxel are determined based on the local boundary voxel coordinate set corresponding to each void voxel.

3. The method of identifying the type of voids in a gob pile according to claim 2, wherein, The step of determining the directional feature data of each void voxel based on the void voxel coordinate set corresponding to each void voxel includes: Determine the principal direction vector value of each voxel corresponding to each coordinate in the void voxel coordinate set; The normalized mean of the outer product of the central main direction vector value and each of the main direction vector values ​​is determined as the directional feature data.

4. The method for identifying the void type of goaf accumulation as described in claim 2, characterized in that, The determination of the clustering feature data of each void voxel based on the void voxel coordinate set and the local boundary voxel coordinate set corresponding to each void voxel includes: Determine the principal axis direction vector of the void voxel coordinate set, and determine the principal axis of the void voxel coordinate set based on the principal axis direction vector and the centroid position of the local void voxel coordinate set; The normalized mean of the distances from each voxel coordinate in the local boundary voxel coordinate set to the principal axis of the coordinate set is determined as the mean off-axis degree of the local boundary voxel coordinate set, and the difference between the unit value and the mean off-axis degree is determined as the clustered feature data.

5. The method for identifying the void type of goaf accumulation as described in claim 2, characterized in that, The step of determining the boundary undulation feature data of each void voxel based on the local boundary voxel coordinate set corresponding to each void voxel includes: The principal plane of the coordinate set and the boundary area of ​​the coordinate set are determined based on the local boundary voxel coordinate set. Based on the principal plane of the coordinate set, determine the principal plane boundary projection area of ​​each voxel in the local boundary voxel coordinate set; The average normalized residual ratio of the projected area of ​​each principal plane boundary to the boundary area of ​​the coordinate set is determined as the boundary undulation feature data.

6. The method for identifying the void type of goaf accumulation as described in claim 1, characterized in that, The method for identifying the void type of goaf accumulation also includes: The spatial distribution data is determined based on the sampling data of the accumulation distribution of the goaf accumulation.

7. A device for identifying the void type of a goaf accumulation, characterized in that, include: The neighborhood extraction module is used to determine the void envelope region corresponding to each void voxel in the spatial distribution data based on the spatial distribution data of voids in the goaf accumulation. The feature calculation module is used to determine the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel based on each void envelope region; the directional feature data is positively correlated with the structural orientation of the voids in the goaf accumulation, the aggregation feature data is positively correlated with the degree of aggregation of the void boundary voxels in the goaf accumulation relative to the principal axis, and the boundary undulation feature data is negatively correlated with the degree of undulation of the local boundary surface of the goaf accumulation; The label determination module is used to determine the voxel type discrimination label of each void voxel based on the directional feature data, aggregation feature data, and boundary undulation feature data of each void voxel.

8. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method for identifying the void type of goaf accumulation as described in any one of claims 1 to 6.

9. An electronic device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method for identifying the void type of goaf accumulation as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product stores a computer program, which, when executed by a processor, implements the steps of the method for identifying the void type of goaf accumulation as described in any one of claims 1 to 6.

Images