A mine area geological subsidence risk assessment method and system

Abstract

The embodiment of the present application provides a kind of geological subsidence risk assessment method and system for mining area, belong to numerical simulation risk assessment technical field.The method comprises: obtaining the observation data and structure information of each monitoring point in mining area, and based on the observation data, point-level capsule for expressing the subsidence behavior characteristics of each monitoring point is constructed;Based on the structure information, the structure coupling factor between monitoring points is calculated as the weight adjustment parameter of dynamic routing process in capsule network;The point-level capsule and the structure coupling factor are jointly input into the preset structure coupling capsule network, and according to the weight adjustment parameter, the dynamic routing mechanism is guided to generate structure unit capsule for representing the subsidence response characteristics of different structure units;Based on the structure unit capsule, the subsidence risk level and its spatial distribution of each monitoring point are generated.The present application scheme significantly improves the accuracy and real-time performance of mining area subsidence risk assessment, and provides more reliable decision support.

Description

Technical Field

[0001] This invention relates to the field of numerical simulation risk assessment technology, specifically to a method and system for assessing geological subsidence risk in mining areas. Background Technology

[0002] Mining subsidence is a common and significant geological problem during mining operations. The resulting ground subsidence, cracking, and infrastructure damage often have a major impact on mine safety and operations. Accurate prediction of mining subsidence is essential for effectively managing these risks. Currently, mining subsidence risk assessment methods largely rely on traditional physical models, such as the finite element method (FEA) and geostatistical methods. While these methods can provide some subsidence predictions, they still have significant limitations when dealing with complex mining geological environments.

[0003] The geological complexity of mining areas is one of the major problems that existing methods cannot effectively address. In actual mining areas, the geological environment exhibits significant heterogeneity and complexity, such as fault zones, goafs, and uneven distribution of lithological layers. These factors have a significant impact on the propagation and spread of subsidence. Traditional methods typically assume that the subsidence process is spatially continuous and homogeneous, or rely solely on local physical parameters for modeling. This makes them unable to accurately reflect the subsidence propagation patterns controlled by discontinuous geological structures such as fault zones or goafs. This spatially discontinuous subsidence propagation pattern leads existing subsidence assessment methods to often ignore or underestimate the prediction of high-risk areas and fail to comprehensively identify potential subsidence risks in mining areas.

[0004] The inadequacy of existing assessment methods in handling multi-source data and spatial distribution is also a significant problem. Settlement monitoring data in mining areas typically includes various data types such as settlement rate, displacement changes, groundwater level, lithological classification, and mining disturbance. These data are often spatially unevenly distributed and exhibit complex interrelationships. However, traditional assessment methods mostly only handle single data sources or fail to fully utilize the spatial relationships between data. Therefore, when generating settlement risk prediction results, these methods often ignore the coupling and spatial variability between data within the mining area, resulting in an inaccurate reflection of the spatial distribution and risk hotspots of settlement. More importantly, existing methods struggle to generate spatial heat maps of settlement risk, making it difficult for mine managers to quickly identify high-risk areas during monitoring, assessment, and emergency response deployment.

[0005] Therefore, existing technologies have significant shortcomings in modeling non-uniform settlement propagation under complex geological conditions in mining areas, as well as in multi-source data fusion and spatial distribution representation. This directly affects the accuracy and reliability of settlement risk assessment in mining areas, and consequently impacts the safety management and decision-making efficiency of mining areas. To address these issues, a novel settlement risk assessment method is urgently needed that can overcome the limitations of existing methods, accurately simulate settlement risks in complex mining environments, and provide more effective decision support for mining area management. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for assessing geological subsidence risk in mining areas, so as to at least solve the problems of the inability of existing technologies to accurately handle the non-uniform subsidence propagation caused by complex geological structures in mining areas and the inadequacy of multi-source data fusion and spatial distribution representation.

[0007] To achieve the above objectives, the first aspect of the present invention provides a method for assessing geological subsidence risk in mining areas. The method includes: acquiring observation data and structural information of each monitoring point in the mining area, and constructing point-level capsules based on the observation data to express the subsidence behavior characteristics of each monitoring point; calculating a structural coupling factor between monitoring points based on the structural information, which serves as a weight adjustment parameter for the dynamic routing process in the capsule network; inputting the point-level capsules and the structural coupling factor into a preset structural coupling capsule network, and guiding a dynamic routing mechanism to generate structural unit capsules to characterize the subsidence response characteristics of different structural units according to the weight adjustment parameter; and generating the subsidence risk level and its spatial distribution for each monitoring point based on the structural unit capsules.

[0008] Optionally, the observation data includes any one or more of the following: settlement rate, settlement acceleration, groundwater level disturbance amplitude, lithological coding information, and mining disturbance parameters of the monitoring points; the structural information includes any one or more of the following: spatial distance between monitoring points, fracture structure topology, hydrological channel connectivity, and lithological similarity; the observation data is used to construct the initial state vector of the point-level capsule, and the structural information is used to determine the structural coupling factor to participate in the weight adjustment during the dynamic routing process.

[0009] Optionally, a point-level capsule for expressing the settlement behavior characteristics of each monitoring point is constructed based on the observation data, including: performing uniform scale normalization on the observation data of each monitoring point to obtain corresponding multidimensional features; encoding the normalized multidimensional features into a composite structure containing a state vector and an attitude matrix based on the capsule encoding rule to obtain a point-level capsule for expressing the settlement behavior characteristics of each monitoring point; wherein, the state vector is used to characterize the settlement behavior intensity and change trend of the monitoring point; and the attitude matrix is ​​used to identify the response pattern of settlement behavior in different structural directions.

[0010] Optionally, calculating the structural coupling factor between monitoring points based on the structural information includes: constructing a candidate connection graph between monitoring points based on the structural information, and selecting monitoring point pairs that are directly connected or physically associated on the structural path as a set of point pairs; extracting spatial coupling feature vectors between structural paths for each point pair using graph embedding, and constructing a learnable weighted fusion function based on the spatial coupling feature vectors and historical settlement co-change behavior; adaptively adjusting the fusion coefficients of various structural sub-factors of the weighted fusion function during network training to obtain the structural coupling factor for dynamic routing weight adjustment.

[0011] Optionally, the structurally coupled capsule network includes: a point-level capsule encoding layer at the input layer, a structurally coupled expression layer, and a structural unit response layer; the structurally coupled capsule network transmits information between layers in a non-fixed connection manner through a dynamic routing mechanism; the structurally coupled capsule network introduces a structural coupling factor as a control condition to replace the routing weight calculation based on similarity scores in the capsule network.

[0012] Optionally, the point-level capsules and the structural coupling factor are jointly input into a preset structural coupling capsule network, and a dynamic routing mechanism guided by the weight adjustment parameters is used to generate structural unit capsules to characterize the settlement response features of different structural units. This includes: performing a transformation operation on the point-level capsules to generate a set of prediction vectors for all candidate structural unit capsules; weighting and aggregating the prediction vectors between each point-level capsule and all candidate structural unit capsules according to the weight adjustment matrix constructed by the structural coupling factor to generate a pre-activation input for the structural unit layer; and performing a normalization operation on the pre-activation input based on a nonlinear compression function to obtain the final state vector of the structural unit capsule. The final state vector is composed of multiple sub-feature vectors, which respectively represent the settlement amplitude distribution, settlement change rate trend, and coupling response intensity to disturbances of surrounding structures of the target structural unit.

[0013] Optionally, based on the structural unit capsule, the settlement risk level and its spatial distribution of each monitoring point are generated, including: comparing and mapping the final state vector of the structural unit capsule with a preset risk level classification rule to obtain the risk interval corresponding to each feature value in the structural response region; reflecting the risk interval to each monitoring point according to the distribution position and information contribution ratio of each monitoring point in the corresponding structural unit, and generating a risk heat map in two-dimensional space based on the coordinates of the monitoring points to represent the settlement risk level and its spatial distribution of each monitoring point.

[0014] Optionally, after obtaining the settlement risk level and spatial distribution of each monitoring point, the method further includes: based on the response consistency judgment between the structural unit capsule and its contained monitoring points, identifying monitoring points whose response deviation exceeds a preset threshold and marking them as abnormal response points; identifying the distribution characteristics of each abnormal response point in the structural boundary area to construct a risk drift trend map or generate a list of intervention area suggestions; wherein, the response deviation is determined by the difference between the actual risk level of the monitoring point and the predicted risk value of the structural unit capsule state vector.

[0015] A second aspect of the present invention provides a geological subsidence risk assessment system for mining areas. The system includes: a data acquisition unit for acquiring observation data and structural information of each monitoring point in the mining area, and constructing point-level capsules based on the observation data to express the subsidence behavior characteristics of each monitoring point; a processing unit for calculating structural coupling factors between monitoring points based on the structural information, which serve as weight adjustment parameters for the dynamic routing process in the capsule network; a model building unit for inputting the point-level capsules and the structural coupling factors into a preset structural coupling capsule network, and guiding a dynamic routing mechanism based on the weight adjustment parameters to generate structural unit capsules that characterize the subsidence response characteristics of different structural units; and an evaluation unit for generating the subsidence risk level and its spatial distribution of each monitoring point based on the structural unit capsules.

[0016] On the other hand, the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described method for assessing geological subsidence risk in mining areas.

[0017] Through the above technical solution, this invention constructs a geological subsidence risk assessment method for mining areas based on structurally coupled capsule networks. This method effectively solves the problem of subsidence coupling and propagation caused by complex geological structures in mining areas, which cannot be accurately reflected in existing technologies. By acquiring observation data and structural information of monitoring points in the mining area, and constructing point-level capsules based on the observation data, the subsidence behavior characteristics of each monitoring point can be accurately expressed. Simultaneously, the structural coupling factor between monitoring points is calculated using structural information and used as a weight adjustment parameter in the dynamic routing process, thereby enhancing the model's adaptability to subsidence propagation under non-uniform and complex structures in mining areas. By inputting point-level capsules and structural coupling factors into the structurally coupled capsule network, the dynamic routing mechanism can accurately generate the subsidence response characteristics of different structural units, and then generate the subsidence risk level and its spatial distribution of monitoring points based on the structural unit capsules. This method not only captures the spatial coupling relationship of subsidence but also generates the precise spatial distribution of subsidence risk, significantly improving the accuracy and real-time performance of mining area subsidence risk assessment and providing more reliable decision support.

[0018] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0020] Figure 1 This is a flowchart of the steps of a method for assessing geological subsidence risk in mining areas provided by one embodiment of the present invention;

[0021] Figure 2 This is a system structure diagram of a geological subsidence risk assessment system for mining areas provided by one embodiment of the present invention. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] Figure 1 This is a flowchart illustrating the steps of a method for assessing geological subsidence risk in mining areas, provided by one embodiment of the present invention. The method includes:

[0024] Step S10: Obtain observation data and structural information of each monitoring point in the mining area, and construct point-level capsules based on the observation data to express the settlement behavior characteristics of each monitoring point.

[0025] Specifically, the observation data includes any one or more of the following: settlement rate, settlement acceleration, groundwater level disturbance amplitude, lithological coding information, and mining disturbance parameters of the monitoring points; the structural information includes any one or more of the following: spatial distance between monitoring points, fracture structure topology, hydrological channel connectivity, and lithological similarity; the observation data is used to construct the initial state vector of the point-level capsule, and the structural information is used to determine the structural coupling factor to participate in the weight adjustment during the dynamic routing process.

[0026] Furthermore, based on the observation data, a point-level capsule is constructed to express the settlement behavior characteristics of each monitoring point, including: performing uniform scale normalization on the observation data of each monitoring point to obtain corresponding multidimensional features; encoding the normalized multidimensional features into a composite structure containing a state vector and an attitude matrix based on the capsule encoding rule to obtain a point-level capsule to express the settlement behavior characteristics of each monitoring point; wherein, the state vector is used to characterize the settlement behavior intensity and change trend of the monitoring point; and the attitude matrix is ​​used to identify the response pattern of settlement behavior in different structural directions.

[0027] In this embodiment of the invention, various types of settlement-related observation data are collected from multi-source monitoring points deployed in the mining area. These observation data include at least: settlement rate (i.e., the rate of change of surface elevation per unit time), settlement acceleration (the rate of change of settlement rate), groundwater level disturbance amplitude (reflecting the degree of influence of local hydrological conditions on stratigraphy), lithological coding information (such as mudstone, sandstone, limestone, etc., and their corresponding numerical representations), and mining disturbance parameters related to the area (e.g., goaf formation time, mining layer depth, mining time series, and its disturbance radius). The above data can be used independently or in combination, depending on the geological characteristics and data availability of the area where the monitoring points are located.

[0028] After obtaining the original observation data, to ensure dimensional consistency and numerical scale uniformity in the subsequent modeling process, all observation data need to undergo uniform scale normalization. Preferably, Z-score normalization or Min-Max normalization methods are used to ensure that each feature dimension falls within a uniform numerical range or a standard normal distribution, thereby avoiding imbalances in network training caused by excessively large data units in one dimension. The normalized observation data forms a multidimensional feature vector, which serves as the basic input for constructing point-level capsules.

[0029] Furthermore, based on the encoding rules of capsule networks, the above-mentioned normalized feature vectors are transformed into capsule units with structural expressive capabilities. Preferably, the point-level capsules are modeled using a composite structure of state vectors and attitude matrices, wherein:

[0030] 1) The state vector (activation vector) is used to represent the intensity and trend of settlement behavior of the monitoring point within the current time window. Its dimension can be selected according to business needs, such as a real number vector with a length of 8 or 16. The vector magnitude can express the degree of settlement, and the direction can guide the dynamic allocation of subsequent routing processes.

[0031] 2) The pose matrix is ​​used to characterize the response pattern of settlement behavior at a monitoring point in different geological structural directions. Preferably, a small-dimensional matrix of 2×2 or 4×4 can be used to encode the spatial projection characteristics and directional sensitivity of settlement features. This matrix reflects the directional trend of settlement changes under controlled geological structures (such as fault dip, bedding direction, hydrological flow direction, etc.) at a certain monitoring point.

[0032] Based on the above structure, the point-level capsule not only has the ability to represent sedimentation intensity, but also has a strong ability to describe spatial response, providing highly expressive atomic units for subsequent dynamic information routing based on structural coupling.

[0033] Furthermore, to improve the consistency and stability of capsule encoding, regularization strategies can be introduced during the construction process. For example, the magnitude of the state vector can be constrained to not exceed a preset threshold, or an L2 norm regularization term can be applied to the pose matrix to avoid numerical explosion or gradient instability. Simultaneously, depending on the actual dimensionality of the observed data, small feedforward networks or shallow convolutional layers can be used to extract and encode features from the observed data to improve the robustness of the capsule representation.

[0034] On the other hand, in addition to observational data, structural information related to the geological relationships between monitoring points also needs to be acquired simultaneously. This structural information preferably includes: the spatial Euclidean distance between any two monitoring points; the topological relationship of the fracture structures between them (e.g., whether they cross faults, fault spacing, fault extension direction, etc.); the connectivity of underground hydrological channels (which can be obtained through numerical simulation or seepage modeling); and lithological similarity indices (obtained by comparing the lithological codes of the rock strata where each monitoring point is located and calculating similarity scores). This structural information can be used to measure the coupling strength between two monitoring points in terms of geological structure, forming the basis for subsequent calculations of structural coupling factors.

[0035] The constructed point-level capsules are used as the underlying representation units and, together with structural information, are input into the routing mechanism of the network. The structural information is used to calculate the structural coupling factor between monitoring points and to weight and adjust the destination paths of point-level capsules during dynamic routing. This approach not only preserves the fine-grained representation of single-point settlement characteristics but also introduces a global coupling mechanism based on geological structural relationships, laying the foundation for multi-point collaborative modeling of settlement risk in mining areas.

[0036] Step S20: Calculate the structural coupling factor between monitoring points based on the structural information, and use it as a weight adjustment parameter for the dynamic routing process in the capsule network.

[0037] Specifically, a candidate connection graph between monitoring points is constructed based on structural information, and monitoring point pairs that are directly connected or physically associated on the structural paths are selected as a set of point pairs. For each point pair, spatial coupling feature vectors between structural paths are extracted through graph embedding, and a learnable weighted fusion function is constructed based on the spatial coupling feature vectors and historical settlement co-change behavior. During network training, the fusion coefficients of various structural sub-factors of the weighted fusion function are adaptively adjusted to obtain structural coupling factors for dynamic routing weight adjustment.

[0038] In this embodiment of the invention, a candidate connection diagram between monitoring points is constructed based on the acquired structural information. This diagram describes the potential coupling relationships between monitoring points in the mining area at the geological structural level. In this diagram, nodes represent monitoring points, and edges represent structural connections. The determination of structural connections is not limited to physical proximity but is based on one or more of the following structural factors: fracture topology (e.g., whether two monitoring points are located within the same fault block or cross faults), hydrological connectivity (e.g., using seepage simulation to determine the hydraulic connectivity between two points), lithological similarity (e.g., using lithologically encoded vector distances to calculate similarity), and consistency of tectonic stress direction. This structural information can be obtained from geological modeling results, seismic interpretation maps, and hydrological analysis models.

[0039] Based on the candidate connectivity graph described above, monitoring point pairs with direct connectivity or physical coupling relationships on the structural path are selected from all possible monitoring point pairs to form a valid point pair set. Judgment criteria include, but are not limited to: fracture spacing less than a threshold, hydrological coupling flux exceeding a background value, and lithological coding similarity higher than a set threshold. This point pair set is used for subsequent coupling factor construction to ensure that the modeling focuses on actual correlations supported by physical mechanisms.

[0040] For each valid pair of points, the spatial coupling features between their structural paths need to be extracted. To achieve this, graph embedding technology is preferred for embedding modeling of the above structural graph, compressing the original graph topology and the structural attributes of each monitoring point into a representation in vector space. During graph embedding, the attribute vectors of nodes include fault zone distribution, goaf boundary, lithology, etc.; adjacency aggregation embedding methods can be used, such as multi-layer graph convolution or node encoding strategies based on random walks, to obtain the structural embedding feature vectors between each pair of monitoring points.

[0041] Furthermore, to incorporate temporal synergy information beyond structural paths, a synergistic change behavior vector can be constructed based on historical settlement observation data between point pairs. Synergistic change behavior describes the synchronicity and correlation characteristics of settlement changes at two monitoring points within a historical time window. For example, it can be used to calculate the Pearson correlation coefficient, maximum dynamic time warping (DTW) similarity, or mutual information of settlement rate changes across multiple time periods. These indicators can capture unstructured but dynamically correlated patterns exhibiting coupled behavior.

[0042] Then, a learnable weighted fusion function is constructed for the aforementioned spatial coupling feature vector and cooperative change behavior vector to integrate the contributions of various structural sub-factors (such as fracture, lithology, and hydrology) and output the final structural coupling factor. This weighted function can be implemented by a shallow neural network, with the input being a concatenation of point-pair structural embedding vectors and cooperative feature vectors, and the output being a normalized structural coupling factor value, typically limited to the range of 0 to 1. To improve the interpretability and stability of training, a regularization term for the fusion coefficient is introduced during network training to control the dominance of different structural factors.

[0043] Throughout the neural network training phase, this fusion function is automatically optimized as the capsule network loss function backpropagates. Its goal is to give highly coupled monitoring point pairs a stronger route assignment weight in the final risk assessment result. For example, if a pair of monitoring points has historically had a high degree of settlement correlation and is located within the same lithological block, its structural coupling factor will be assigned a higher value; conversely, if there are significant structural interruptions, hydrological barriers, or lithological differences between the two points, the coupling factor will be automatically suppressed.

[0044] The resulting structural coupling factor will be introduced during the dynamic routing process to adjust the adaptation scores between point-level capsules and structural unit capsules. That is, the routing probabilities in the original capsule network will be weighted by the structural coupling factor, making it more likely that monitoring point pairs with strong structural coupling will be assigned to the same structural unit during routing, thus achieving physical coupling modeling between non-adjacent points across space.

[0045] Through the aforementioned technical approach, this implementation method achieves a deep integration of the complex relationships within the mining area structure with a data-driven routing mechanism. Technically, it enhances the model's ability to identify risks in structurally discontinuous areas such as fault crossings, goaf boundaries, and groundwater disturbances. Furthermore, it makes the modeling of subsidence risk propagation more consistent with the actual geological structure and historical subsidence evolution trends, thus providing an accurate foundation for subsequent refined assessment of subsidence risk levels and identification of regional hotspots.

[0046] Step S30: Input the point-level capsule and the structural coupling factor into a preset structural coupling capsule network, and generate structural unit capsules to characterize the settlement response features of different structural units according to the weight adjustment parameters guided by the dynamic routing mechanism.

[0047] Specifically, the structurally coupled capsule network includes: a point-level capsule encoding layer at the input layer, a structurally coupled expression layer, and a structural unit response layer; the structurally coupled capsule network transmits information between layers in a non-fixed connection manner through a dynamic routing mechanism; the structurally coupled capsule network introduces a structural coupling factor as a control condition to replace the routing weight calculation based on similarity scores in the capsule network.

[0048] In this embodiment of the invention, the input in the point-level capsule coding layer consists of point-level capsules for each monitoring point, constructed through preprocessing. Each point-level capsule comprises a state vector and an attitude matrix, used to characterize the settlement behavior of the current monitoring point. The state vector describes the overall level of settlement intensity or trend, while the attitude matrix reflects the point's response pattern along geological structural directions, such as the distribution differences in settlement response at different dip angles. In the point-level capsule coding layer, each capsule is not directly connected to all capsule units in the next layer; instead, a routing mechanism determines the information transmission path and its degree of attribution.

[0049] In traditional capsule networks, dynamic routing mechanisms typically determine information transmission weights based on the similarity score between the predicted vector and the target capsule activation vector. However, in the scenario of mining area subsidence risk modeling, the physical coupling of structural relationships often plays a more dominant role than pure feature space similarity. Therefore, this implementation introduces a structural coupling factor as a new control variable to replace the traditional similarity score, in order to correct the original coupling weight distribution during the dynamic routing process.

[0050] In the structurally coupled representation layer, activation information from point-level capsules is first transformed through affine transformation to generate corresponding predicted capsule representations. These predicted representations are not directly used to calculate similarity with upper-layer activation vectors; instead, they are combined with the structural coupling factor to participate in the weight update process. Specifically, for each pair of point-level capsules connected to candidate structural unit capsules, the structural coupling factor in the structural graph is used as a weighting correction factor to update the initial adaptation score in dynamic routing. The preferred correction rule is: the original adaptation score is multiplied by the structural coupling factor to obtain a new attribution weight coefficient. In this approach, if a pair of capsules has a high structural coupling strength, its information transmission weight is significantly increased; conversely, its attribution probability is attenuated, limiting the propagation path of information interference.

[0051] In practical implementation, this structural coupling factor can be understood as a structure-aware attention mechanism, enabling the network to prioritize physically meaningful transmission paths among multi-path candidate connections, thereby mitigating the bias problem dominated by pure feature similarity. This mechanism is particularly suitable for subsidence risk assessment modeling in complex geological scenarios in mining areas, such as fault-controlled zones, goaf boundaries, and lithological abrupt change zones.

[0052] The structural unit response layer is responsible for aggregating the activation information of multiple point-level capsules after adjustment by the structural coupling expression layer, generating structural unit capsules that characterize the regional settlement response. Essentially, a structural unit capsule is an aggregated representation of multiple coupled-point capsules. Its output state vector characterizes the settlement risk intensity of the target region, while the attitude matrix expresses the settlement response pattern of the region in various directions. For example, if a structural unit shows a significant settlement trend at all its coupled points in the east-west direction, the corresponding dimension weights of its attitude matrix will be increased, thus giving the risk estimation process directional guidance.

[0053] It should be noted that, during the routing iteration process, to further improve network stability and expressive power, this implementation introduces a regularization constraint term during the update of structural coupling weights to prevent excessive weight concentration or gradient vanishing problems. Furthermore, to adapt to the actual needs of mine settlement assessment, a maximum number of routing iterations can be set and the routing saturation dynamically adjusted to match the information fusion depth required by different structural unit capsules.

[0054] In one possible implementation, taking a typical mining area as an example, 20 monitoring points are set up. For each monitoring point, the settlement rate, settlement acceleration, water level disturbance value, lithology type (coded), and disturbance level corresponding to underground operations are collected. These data constitute the observation data vector for each monitoring point. Simultaneously, fault zone distribution maps, hydrological channel models, and lithology distribution models are extracted from the geological model to construct a structural information matrix between the monitoring points.

[0055] First, a point-level capsule input is constructed based on the observation data of each monitoring point. For each monitoring point i, its observation data vector is... Normalization is performed to obtain the input vector u i Then, through the affine mapping matrix... Generate prediction vectors:

[0056]

[0057] This vector represents the activation prediction of monitoring point i for candidate structural unit capsule j.

[0058] Subsequently, a structural coupling factor matrix was constructed based on fracture topology, hydrological flux, and lithological similarity. Each value represents the monitoring point's... The overall structural coupling strength. The construction method is as follows:

[0059] 1) For monitoring point pairs Calculate whether they are in the same fault block, whether they are hydrologically connected, and whether the lithological coding distance is below the threshold.

[0060] 2) Each feature is converted into a sub-factor in the interval [0,1].

[0061] 3) Set weighting coefficients Define the comprehensive coupling factor:

[0062]

[0063] in, These represent the coupling functions of fracture, hydrology, and lithology, respectively. satisfy .

[0064] In the dynamic routing process, a structural coupling factor is introduced to correct the traditional similarity-driven weight scores. Each pair is initialized. The coupling score is 0, denoted as In each iteration, the route weights are adjusted according to the following update formula:

[0065]

[0066] Where α is the learning step size; It is the current output vector of structural unit capsule j; The inner product of the prediction and the actual output is represented as the fit score.

[0067] The route normalization weights are calculated using the softmax function:

[0068]

[0069] Then, the input aggregation vector of the structural unit capsule j is:

[0070]

[0071] The final output vector is obtained using the squash function:

[0072]

[0073] The `squash` function compresses the output vector length, preserves the direction, and normalizes the magnitude to (0,1), representing the capsule activation intensity. The structural unit capsule output v j The data is further decoded into settlement risk levels for each monitoring point. The decoder uses a fully connected neural network, outputting five-level risk classification labels (extremely low, low, medium, high, and extremely high), and provides the confidence level for each risk category through a softmax function.

[0074] Specifically, the point-level capsules and the structural coupling factor are input into a preset structural coupling capsule network, and a dynamic routing mechanism guided by the weight adjustment parameters is used to generate structural unit capsules that characterize the settlement response features of different structural units. This includes: performing a transformation operation on the point-level capsules to generate a set of prediction vectors for all candidate structural unit capsules; weighting and aggregating the prediction vectors between each point-level capsule and all candidate structural unit capsules according to the weight adjustment matrix constructed by the structural coupling factor to generate a pre-activation input for the structural unit layer; and performing a normalization operation on the pre-activation input based on a nonlinear compression function to obtain the final state vector of the structural unit capsule. The final state vector is composed of multiple sub-feature vectors, which respectively represent the settlement amplitude distribution, settlement change rate trend, and coupling response intensity to disturbances of surrounding structures of the target structural unit.

[0075] In this embodiment of the invention, an affine transformation is performed on the point-level capsules constructed for each monitoring point to obtain multiple sets of prediction vectors in different candidate structural unit directions. These prediction vectors represent the estimates of each monitoring point's response patterns to different structural units. Subsequently, based on a pre-constructed structural coupling factor, a weighted adjustment matrix for controlling routing preferences is generated. This matrix is ​​derived from structural information across multiple dimensions, such as fracture structures, hydrological channels, and lithological similarity. Through weighted fusion of coupling strengths, it reflects the physical linkage strength between different monitoring points. This adjustment matrix does not rely on spatial proximity but rather reflects structural correlation, enabling some non-adjacent monitoring points to establish connections within the network.

[0076] During dynamic routing, the predicted vectors between the point-level capsule and all candidate structural unit capsules are weighted and aggregated according to the aforementioned adjustment matrix to form the pre-activation input of the structural unit layer. This aggregation not only considers the similarity between the predicted content and the current output but also introduces the adjustment effect of the structural coupling factor, so that point pairs with stronger physical coupling occupy a greater weight in information routing. Subsequently, nonlinear compression processing is performed on the aggregated pre-activation input to obtain the final structural unit capsule state vector.

[0077] It is worth noting that the final output structural unit capsule state vector is composed of multiple sub-vectors, each representing the structural unit's response characteristics in different dimensions. Specifically, these include: a sub-vector representing the settlement amplitude to characterize the overall settlement intensity and distribution characteristics of the monitoring point set; a sub-vector reflecting the rate of change of the settlement trend direction and velocity over time; and a sub-vector representing the coupling response intensity to describe the structural unit's sensitivity to external inputs such as mining activities or water level fluctuations. This multi-feature concatenation method endows the structural unit capsule with richer expressive power, facilitating more detailed discrimination and classification in subsequent risk assessments.

[0078] Step S40: Based on the structural unit capsule, generate the settlement risk level and its spatial distribution for each monitoring point.

[0079] Specifically, the final state vector of the structural unit capsule is compared and mapped with the preset risk level classification rules to obtain the risk interval corresponding to each feature value in the structural response region; based on the distribution location and information contribution ratio of each monitoring point in the corresponding structural unit, the risk interval is reflected to each monitoring point, and a risk heat map is generated in two-dimensional space based on the coordinates of the monitoring points to represent the settlement risk level and spatial distribution of each monitoring point.

[0080] In this embodiment of the invention, the output state vector of the structural unit capsule consists of multiple sub-vectors, representing the multidimensional response characteristics of the structural unit in terms of settlement amplitude, rate of change, and structural coupling response. To transform this high-dimensional semantic information into a practical risk level label, a set of risk level classification rules matching mining area engineering experience must first be established. These rules can be formulated based on historical settlement data, existing geological disaster level standards, and regulatory requirements, typically dividing risks into five or three levels, and setting threshold ranges or typical characteristic patterns for each level. By comparing and mapping the various response characteristics output by the structural unit capsule with the above rules, the settlement risk range of the structural unit can be determined.

[0081] However, since capsule networks employ a many-to-many dynamic routing approach, a single monitoring point often participates in the state construction of multiple structural units. Therefore, it is not feasible to simply assign the risk level of a structural unit directly to a single monitoring point. Instead, it is necessary to further consider the monitoring point's participation in multiple structural units and perform a contribution-based risk mapping. To this end, the routing weight of each monitoring point within the structural units it participates in constructing can be calculated, and this weight can be used as the proportion of its risk information allocation. Ultimately, the settlement risk level of the monitoring point is obtained by weighted fusion of the risk intervals of multiple structural units, ensuring the spatial continuity and physical rationality of the assessment results.

[0082] At the spatial representation level, to visualize risk levels, the risk level of each monitoring point needs to be mapped back to its original coordinates in two-dimensional geographic space, thereby generating a spatial heatmap of subsidence risk. This map uses a color gradient to represent changes in risk intensity, which not only visually presents the concentrated distribution of high-risk areas but also helps identify anomalous risks in boundary transition zones. To further enhance the continuity of spatial distribution, interpolation algorithms, such as the inverse distance weighting method or Gaussian kernel function, can be introduced based on the monitoring point data to perform continuous surface interpolation of risk levels, generating a smoother risk heatmap.

[0083] Preferably, after obtaining the settlement risk level and spatial distribution of each monitoring point, the method further includes: based on the response consistency judgment between the structural unit capsule and its contained monitoring points, identifying monitoring points whose response deviation exceeds a preset threshold and marking them as abnormal response points; identifying the distribution characteristics of each abnormal response point in the structural boundary area to construct a risk drift trend map or generate a list of intervention area suggestions; wherein, the response deviation is determined by the difference between the actual risk level of the monitoring point and the predicted risk value of the structural unit capsule state vector.

[0084] In this embodiment of the invention, each structural unit capsule has integrated its associated point-level capsule information through the aforementioned dynamic routing process, forming a high-dimensional state vector representing the composite response mode of its overall settlement behavior. Simultaneously, each monitoring point also possesses its independently generated settlement risk level value, derived from its own observation data and its feature contribution within the capsule network. To determine whether there is an abnormal deviation between the two, the state vector output by the structural unit capsule needs to be mapped to the risk level space to obtain the predicted risk value of the structural unit in the direction of the corresponding monitoring point. The difference between this predicted value and the actual risk level of the monitoring point is calculated to obtain a numerical deviation index.

[0085] If the response deviation of a monitoring point exceeds a preset threshold range (which can be set based on historical data distribution or experience), then the point is determined to have behavioral anomalies within the current structural unit. Such points are marked as "abnormal response points," indicating that their settlement trend has failed to maintain consistency with the overall pattern of the structural unit to which they belong, and may be additionally affected by local lithological changes, sudden water level disturbances, or human mining activities.

[0086] Furthermore, to reveal the spatial distribution characteristics of anomalies, this method performs spatial clustering and boundary region analysis on all anomaly response points, focusing on identifying whether they are clustered along structural unit edges or across structural interfaces. If most anomalies are found to be located at structural edges, a "risk drift" phenomenon may exist, meaning that the original risk center area has spatially shifted. Based on this phenomenon, a risk drift trend map can be generated to express the potential migration paths of risk hotspots. Simultaneously, based on the density and spatial connectivity of anomalies, high-priority intervention areas can be automatically selected, and a recommendation list can be output as a basis for decision-making regarding engineering monitoring intensification, early warning level upgrades, or construction avoidance.

[0087] Figure 2 This is a system structure diagram of a geological subsidence risk assessment system for mining areas provided by one embodiment of the present invention. Figure 2As shown, this invention provides a geological subsidence risk assessment system for mining areas. The system includes: a data acquisition unit for acquiring observation data and structural information of each monitoring point in the mining area, and constructing point-level capsules based on the observation data to express the subsidence behavior characteristics of each monitoring point; a processing unit for calculating the structural coupling factor between monitoring points based on the structural information, which serves as a weight adjustment parameter for the dynamic routing process in the capsule network; a model building unit for inputting the point-level capsules and the structural coupling factor into a preset structural coupling capsule network, and guiding a dynamic routing mechanism based on the weight adjustment parameter to generate structural unit capsules that characterize the subsidence response characteristics of different structural units; and an evaluation unit for generating the subsidence risk level and its spatial distribution of each monitoring point based on the structural unit capsules.

[0088] The present invention also provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described method for assessing geological subsidence risk in mining areas.

[0089] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0090] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0091] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A method for assessing geological subsidence risk in mining areas, characterized in that, The method includes: Observational data and structural information from each monitoring point in the mining area are acquired, and point-level capsules are constructed based on the observational data to express the settlement behavior characteristics of each monitoring point; wherein, Based on the observation data, a point-level capsule is constructed to express the settlement behavior characteristics of each monitoring point. This includes: performing uniform scale normalization on the observation data of each monitoring point to obtain corresponding multidimensional features; encoding the normalized multidimensional features into a composite structure containing a state vector and an attitude matrix based on capsule encoding rules to obtain a point-level capsule expressing the settlement behavior characteristics of each monitoring point; wherein the state vector is used to characterize the settlement behavior intensity and trend of the monitoring point; and the attitude matrix is ​​used to identify the response patterns of settlement behavior in different structural directions. Based on the structural information, the structural coupling factor between monitoring points is calculated and used as a weight adjustment parameter for the dynamic routing process in the capsule network; wherein, The calculation of structural coupling factors between monitoring points based on the structural information includes: constructing a candidate connection graph between monitoring points based on the structural information, and selecting monitoring point pairs that are directly connected or physically related on the structural paths as a set of point pairs; extracting spatial coupling feature vectors between structural paths for each point pair using graph embedding, and constructing a learnable weighted fusion function based on the spatial coupling feature vectors and historical settlement co-change behavior; adaptively adjusting the fusion coefficients of various structural sub-factors of the weighted fusion function during network training to obtain structural coupling factors for dynamic routing weight adjustment. The point-level capsules and the structural coupling factor are input into a preset structural coupling capsule network, and a dynamic routing mechanism is guided by the weight adjustment parameters to generate structural unit capsules that characterize the settlement response features of different structural units; wherein, The structurally coupled capsule network includes: a point-level capsule encoding layer at the input layer, a structurally coupled expression layer, and a structural unit response layer; the structurally coupled capsule network transmits information between layers in a non-fixed connection manner through a dynamic routing mechanism; the structurally coupled capsule network introduces a structural coupling factor as a control condition to replace the routing weight calculation based on similarity scores in the capsule network; The point-level capsules and the structural coupling factor are input into a preset structural coupling capsule network, and a dynamic routing mechanism guided by the weight adjustment parameters is used to generate structural unit capsules to characterize the settlement response features of different structural units. This includes: performing a transformation operation on the point-level capsules to generate a set of prediction vectors for all candidate structural unit capsules; weighting and aggregating the prediction vectors between each point-level capsule and all candidate structural unit capsules according to the weight adjustment matrix constructed by the structural coupling factor to generate a pre-activation input for the structural unit layer; and performing a normalization operation on the pre-activation input based on a nonlinear compression function to obtain the final state vector of the structural unit capsule. The final state vector is composed of multiple sub-feature vectors, which respectively represent the settlement amplitude distribution, settlement change rate trend, and coupling response intensity to disturbances of surrounding structures of the target structural unit. Based on the aforementioned structural unit capsule, the settlement risk level and its spatial distribution at each monitoring point are generated, including: The final state vector of the structural unit capsule is compared and mapped with the preset risk level classification rules to obtain the risk interval corresponding to each feature value in the structural response region. Based on the distribution location and information contribution ratio of each monitoring point in its corresponding structural unit, the risk range is reflected to each monitoring point, and a risk heat map is generated in two-dimensional space based on the coordinates of the monitoring points to represent the settlement risk level and spatial distribution of each monitoring point.

2. The method for assessing geological subsidence risk in mining areas according to claim 1, characterized in that, The observation data includes: Any one or more of the following: settlement rate, settlement acceleration, groundwater level disturbance amplitude, lithological coding information, and mining disturbance parameters at the monitoring point; The structural information includes any one or more of the following: spatial distance between monitoring points, topological relationship of fracture structures, connectivity of hydrological channels, and lithological similarity. The observation data is used to construct the initial state vector of the point-level capsule, and the structural information is used to determine the structural coupling factor to participate in the weight adjustment during the dynamic routing process.

3. The method for assessing geological subsidence risk in mining areas according to claim 1, characterized in that, After obtaining the settlement risk level and its spatial distribution at each monitoring point, the method further includes: Based on the consistency of response between the structural unit capsule and the monitoring points it contains, monitoring points whose response deviation exceeds a preset threshold are identified and marked as abnormal response points; Identify the distribution characteristics of each anomalous response point in the structural boundary region to construct a risk drift trend map or generate a list of recommended intervention areas; among which, The response deviation is determined by the difference between the actual risk level of the monitoring point and the predicted risk value of the structural unit capsule state vector.

4. A geological subsidence risk assessment system for mining areas, characterized in that, The system is used to execute the geological subsidence risk assessment method for mining areas as described in any one of claims 1-3, and the system includes: The acquisition unit is used to acquire observation data and structural information of each monitoring point in the mining area, and to construct a point-level capsule based on the observation data to express the settlement behavior characteristics of each monitoring point; The processing unit is used to calculate the structural coupling factor between monitoring points based on the structural information, which serves as the weight adjustment parameter for the dynamic routing process in the capsule network. The model building unit is used to input the point-level capsules and the structural coupling factor into a preset structural coupling capsule network, and to guide the dynamic routing mechanism to generate structural unit capsules that characterize the settlement response features of different structural units according to the weight adjustment parameters. An evaluation unit is used to generate the settlement risk level and its spatial distribution for each monitoring point based on the structural unit capsule.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the method for assessing geological subsidence risk in mining areas as described in any one of claims 1-3.

Images