Intelligent stope structure parameter optimization system and method based on three-dimensional numerical simulation

Abstract

The invention discloses a stope structure parameter intelligent optimization system and method based on three-dimensional numerical simulation, and relates to the technical field of three-dimensional numerical simulation. According to the system, firstly, parameter combinations in a preset range are preliminarily screened and verified by means of a three-dimensional numerical simulation tool through a stope structure parameter combination screening and verification module, and the optimization range is narrowed; inputting the acquired stope structure parameters and mining monitoring data into a constructed response surface model for simulation optimization through a candidate scheme screening and suitability verification module, and verifying the suitability to reduce model fitting error interference; and finally, judging whether the three-dimensional numerical simulation parameters and the constraint conditions are dynamically adjusted and optimized or not through a dynamic adjustment and optimization judgment module according to an adaptability verification result, so that dynamic matching of the three-dimensional numerical simulation parameters and the field geological conditions is realized. And the adaptation precision of the stope structure parameter optimization scheme and the field multi-source dynamic monitoring data is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional numerical simulation technology, and in particular to an intelligent optimization system and method for mining site structure parameters based on three-dimensional numerical simulation. Background Technology

[0002] With the continuous growth in demand for mineral resources and the increasing depletion of easily mined shallow resources, mining is gradually extending to deeper levels. Large metal mines such as the Tonglushan copper-iron mine have entered the mining stage at depths of -365m to -785m or even deeper. Existing technologies construct three-dimensional geomechanical models that integrate geological databases and rock mechanics parameters, and use tools such as FLAC3D to simulate the evolution of stress and displacement fields under different stope spans, segment heights, and filling parameters. Machine learning algorithms (such as genetic algorithms) are then used to perform multi-objective optimization of the simulation results, establishing a response surface model for stope stability and mining efficiency. An intelligent decision-making module matches on-site geological conditions, automatically outputs the optimal parameter scheme, and links with industrial test data to achieve dynamic correction of simulation parameters, ultimately realizing a leap from empirical design to quantitative intelligent optimization of stope structural parameters.

[0003] For example, Chinese invention patent application CN120197360A discloses a method and system for optimizing stope structure parameters in filling mining based on numerical simulation. The method includes: establishing a mine entity model by acquiring the geological conditions of the mine engineering and determining the stope structure parameter scheme; regrouping the merged mine entity model and the stope model; setting initial conditions and mechanical parameters extracted from the geological conditions of the mine engineering to assign parameter values ​​to the merged model; simulating excavation and filling; determining the stability index of the stope under the original rock, void, and filling conditions respectively; and selecting the optimal stope structure parameter scheme.

[0004] For example, Chinese invention patent CN120145699B discloses a method for directional pre-splitting simulation analysis of ultra-thick coal seam fully mechanized caving mining based on three-dimensional modeling. The method includes: conducting geological exploration of the ultra-thick coal seam fully mechanized caving mining area, thereby laying out pre-splitting boreholes and building a three-dimensional model, and determining the detonation sequence based on a predetermined directional pre-splitting path. At the same time, a gradient detonation time difference is designed to form multiple pre-splitting simulation test groups. Then, the three-dimensional model of the ultra-thick coal seam fully mechanized caving mining area is used to carry out detonation under each pre-splitting simulation test group, and the crack propagation path conformity and environmental risk monitoring and assessment are carried out simultaneously during the detonation process.

[0005] The above-mentioned technology has at least the following technical problems: In deep mining scenarios, the environment is characterized by "three highs and one disturbance"—high ground stress, high ground temperature, high karst water pressure (or high seepage pressure), and intense mining disturbances—exacerbating the uncertainty of rock mass deformation and failure. However, existing 3D numerical models largely rely on static geological data and rock mechanics parameters from the initial exploration phase, lacking a dynamic update mechanism. Firstly, the computational power allocation of modeling software (such as FLAC3D and 3DEC) for mesh update algorithms needs to be optimized for deep mining scenarios, causing a lag in the software's response to on-site geological changes and mesh reconstruction. When new geological structures such as concealed faults and fissures are exposed during mining, technicians must manually adjust the mesh topology and boundary conditions, resulting in model adjustments lagging behind on-site changes. Secondly, the assignment of rock mechanics parameters often uses average data from historical rock sample tests as a reference, failing to fully consider the real-time dynamic impact of monitored stress and temperature data on rock mechanics properties, making it difficult to synchronously update the dynamic changes in parameters such as the rock mass's elastic modulus and Poisson's ratio under mining influence.

[0006] This static modeling mode leads to significant deviations between the model and the actual site conditions. As a result, the simulation results of tools such as FLAC3D cannot accurately capture the instantaneous stress concentration and displacement changes in the stope under high ground stress. This, in turn, affects the reliability of the response surface model optimized by the genetic algorithm, resulting in a mismatch between the output parameters such as stope span and filling strength and the actual geological conditions. This not only reduces the timeliness of the intelligent optimization closed loop of the three-dimensional numerical simulation, but also further amplifies the compatibility problem between the intelligent optimization results of the stope structure parameters and the multi-source dynamic monitoring data in the field. There is a problem of low compatibility between the intelligent optimization results of the stope structure parameters and the multi-source dynamic monitoring data in the field. Summary of the Invention

[0007] To address the technical problem of poor compatibility between the intelligent optimization results of stope structure parameters and multi-source dynamic monitoring data in existing technologies, this invention provides an intelligent optimization system and method for stope structure parameters based on three-dimensional numerical simulation. The technical solution is as follows: On the one hand, an intelligent optimization system for stope structure parameters based on three-dimensional numerical simulation is provided. This system includes: a stope structure parameter combination screening and verification module, used to perform preliminary screening and verification of stope structure parameter combinations within a preset stope range during the deep ore body mining stage using three-dimensional numerical simulation tools, in order to narrow down the optimization range of stope structure parameters; a candidate scheme screening and adaptability verification module, used to input the acquired stope structure parameters and mining monitoring data into the constructed response surface model for multi-scenario simulation and target optimization, output candidate parameter optimization schemes, and simultaneously perform adaptability verification to reduce the interference of the fitting error of the response surface model on the optimization of stope parameters; and a dynamic optimization judgment module, used to determine whether to dynamically optimize the three-dimensional numerical simulation parameters and constraints based on the results of the adaptability verification, in order to improve the adaptability between the stope structure parameter scheme and the on-site geological conditions.

[0008] On the other hand, a method for intelligent optimization of stope structure parameters based on three-dimensional numerical simulation is provided. This method includes: Step 1, in the deep ore body mining stage, using three-dimensional numerical simulation tools to preliminarily screen and verify the combination of stope structure parameters within the preset stope range, so as to narrow the optimization range of stope structure parameters; Step 2, inputting the acquired stope structure parameters and mining monitoring data into the constructed response surface model for multi-scenario simulation and target optimization, outputting candidate parameter optimization schemes, and simultaneously performing adaptability verification to reduce the interference of the fitting error of the response surface model on the optimization of stope parameters; Step 3, based on the results of the adaptability verification, determining whether to dynamically adjust the three-dimensional numerical simulation parameters and constraints to improve the adaptability between the stope structure parameter scheme and the on-site geological conditions.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. The stope structure parameter combination screening and verification module uses 3D numerical simulation tools to screen and verify preset parameter combinations during deep mining, narrowing the optimization range. This step can eliminate parameters that are clearly inconsistent with the high-temperature, high-volume, high-altitude, and high-disturbance environment in advance, reducing subsequent computation and alleviating the problem of delayed mesh update response in modeling software. Next, the candidate scheme screening and adaptability verification module inputs stope structure parameters and mining monitoring data into the response surface model for simulation optimization and adaptability verification. This process introduces real-time monitoring data, breaking the limitations of relying on static geological data, reducing the fitting error of the response surface model, and making the basic data for genetic algorithm optimization more closely match the field, reducing the mismatch between parameter schemes and actual geological conditions. Finally, the dynamic optimization judgment module determines whether to optimize simulation parameters and constraints based on the verification results. This dynamically updates rock mechanics parameters, synchronously reflecting the influence of stress and temperature on elastic modulus and Poisson's ratio, solving the defects of static assignment, enabling 3D numerical simulation tools to accurately capture instantaneous stress concentration and displacement mutations, improving the adaptability of optimization results to multi-source dynamic monitoring data in the field, and ensuring the timeliness of the intelligent optimization closed loop of 3D numerical simulation.

[0010] 2. A screening and evaluation index for stope parameters is obtained using a 3D numerical simulation tool. This index is calculated using a stability objective function: taking parameters such as stope span and filling strength as variables, inputting stability data such as rock mass stress and displacement, and superimposing difference integrals and derivative deduction terms, it quantifies the matching degree between the parameter combination and the geological state. This index is compared with a set index; if it meets the standard, it is positively retained; otherwise, it is negatively rejected. This allows for the early screening of unsafe parameter combinations, reducing subsequent invalid calculations and narrowing the scope for optimization. Preliminary verification is performed based on the retained parameter combinations, extracting verification indicators such as rock mass stress increment and cumulative displacement at different time points, and comparing them with dynamic stability thresholds. If all indicators meet the standard, the verification is passed; if a single indicator exceeds the standard, the stope span is reduced proportionally according to the deviation mapping, the response surface model is updated, and the verification is repeated, which can reduce simulation deviations caused by static parameter assignment; if the re-verification still exceeds the standard or all indicators exceed the standard, an early warning is issued; if multiple indicators exceed the standard, they are negatively rejected, which can prevent subsequent simulations from failing to accurately capture instantaneous stress concentration and displacement mutations under high ground stress. The entire process involves quantitative evaluation and dynamic verification, which not only eliminates parameters that do not meet safety requirements in advance and reduces optimization redundancy, but also reduces model deviation caused by static assignment through targeted parameter adjustment and retesting, and can also provide timely warnings of exceeding the standard, thereby improving the safety and adaptability of mining structure parameter optimization.

[0011] 3. By first aligning the initially screened mining structure parameters, mining monitoring data, and geological monitoring data in time and space using a Kalman filter algorithm, constraint vectors are obtained. This ensures data consistency across time and space, avoiding constraint deviations caused by data misalignment. Next, the input dataset is aligned using the constraint vectors as a benchmark. A convolutional neural network is then used to extract target-fusion features and transform them into query vectors. This accurately captures the correlation between parameters and monitoring data, providing a reliable basis for subsequent matching. If the geological complexity exceeds the allowable value, a structural correction coefficient is introduced and coupled with the original constraint vectors to dynamically adjust the constraint vectors. This enhances their adaptability to complex geology and avoids the problem of static constraints being unable to cope with structural changes. Adaptability verification first compares the relative deviation rate between candidate schemes and field parameters, marking schemes exceeding the allowable range that need correction, and quickly locating obviously unsuitable parameters. Then, a historical qualified constraint vector database is called for backtracking verification to accurately find the source of the deviation. Subsequently, a genetic algorithm is used for secondary fine-tuning, and the deviation rate is recalculated. If the target is met, the solution is output, which can further optimize the parameter accuracy; if the target is not met, it is input into the response surface model for simulation verification. If it passes, it is added to the database; if it fails, a failure is indicated. This ensures the reliability of the solution and improves the efficiency of subsequent optimization through data accumulation. The entire process, through dynamic constraint adjustment, precise deviation location, and multiple rounds of optimization verification, effectively reduces static modeling errors, improves the adaptability of parameter solutions to on-site geological conditions, and ensures the accuracy of mining area structure parameter optimization.

[0012] 4. By first obtaining the relative deviation scores of parameters and the matching scores of constraints, and comparing them with the corresponding thresholds, the optimization type is determined: if a single score fails to meet the standard, the corresponding simulation parameters or constraints are optimized; if both fail, simultaneous optimization is performed; otherwise, no optimization is needed. This achieves targeted adjustments and avoids blind optimization. For 3D numerical simulation parameter optimization, for parameters that fail to meet the standards, historical case libraries are used to extract correction patterns. Adjustment coefficients are calculated based on on-site rock mass mechanical response data, and after normalization and summation, the actual adjustment coefficients are obtained and coupled with the original parameters to generate a new combination. If the parameters meet the standards, the new parameters are adopted; otherwise, the dimensions are expanded. This allows for dynamic correction of rock mass parameters, improving the model's adaptability to deep environments and reducing static assignment bias. For constraint optimization, for constraints that fail to meet the standards, historical effective constraint features are extracted, and correction weights are calculated based on on-site structural environment data. After normalization and summation, these weights are coupled with the original constraint vector to generate a new vector. If the constraints meet the standards, the new constraints are adopted; otherwise, the dimensions are expanded. This enhances the coverage of constraints for complex geology and avoids the disconnect between static constraints and the field situation. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A schematic diagram of the intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation provided in an embodiment of the present invention; Figure 2 A flowchart corresponding to the mining site structure parameter combination screening and verification module provided in an embodiment of the present invention; Figure 3 A flowchart corresponding to the candidate scheme screening and adaptability verification module provided in the embodiments of the present invention; Figure 4 This is a flowchart corresponding to the dynamic optimization determination module provided in the embodiments of the present invention; Figure 5 A flowchart of an intelligent optimization method for mining site structure parameters based on three-dimensional numerical simulation provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0016] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0017] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] like Figure 1The diagram shows a structural schematic of an intelligent optimization system for stope structure parameters based on three-dimensional numerical simulation. This system may include: a stope structure parameter combination screening and verification module, used during deep ore body mining to perform preliminary screening and verification of stope structure parameter combinations within a preset stope area using three-dimensional numerical simulation tools, thereby narrowing the optimization range of stope structure parameters; a candidate scheme screening and adaptability verification module, used to input the acquired stope structure parameters and mining monitoring data into a pre-constructed response surface model for multi-scenario simulation and target optimization, outputting candidate parameter optimization schemes and simultaneously performing adaptability verification to reduce the interference of response surface model fitting errors on stope parameter optimization; and a dynamic optimization judgment module, used to determine whether to dynamically optimize the three-dimensional numerical simulation parameters and constraints based on the adaptability verification results, thereby improving the adaptability between the stope structure parameter scheme and the on-site geological conditions.

[0020] Specifically, multiple sets of stope structural parameters and corresponding mining monitoring data (displacement, stress, etc.) are first collected as samples through a combination of field measurements and numerical simulations. Experimental design methods (such as Box-Behnken design) are used as sample data for training. Factor levels are set according to the reasonable range of parameter values ​​to generate experimental combinations covering the key parameter range. Numerical simulations are used to supplement the mining efficiency under each combination to ensure that the data covers the parameter range. With structural parameters as input and mining safety and efficiency indicators as output, quadratic polynomial regression analysis is used to fit the functional relationship for model training. Finally, the accuracy of the model is tested using validation samples, errors are corrected, and the response surface model is finally fitted.

[0021] In this embodiment, the stope structure parameter combination screening and verification module uses three-dimensional numerical simulation to initially screen and verify parameter combinations, significantly narrowing the optimization range, reducing subsequent calculations, and improving optimization efficiency. Furthermore, the preset stope range is based on geological exploration data of deep ore bodies (such as ore body occurrence morphology, burial depth, and lithological distribution) and the overall mining project plan, ensuring targeted screening. The candidate scheme screening and adaptability verification module combines stope structure parameters and mining monitoring data (typically including stope surrounding rock displacement, support structure stress, groundwater level, and ore recovery rate) to perform multi-scenario simulation optimization, simultaneously verifying adaptability and effectively reducing response surface model fitting error interference, ensuring the reliability of candidate schemes. The dynamic optimization judgment module dynamically optimizes simulation parameters and constraints based on the verification results, further improving the adaptability of the scheme to on-site geological conditions, providing strong support for the safe and efficient mining of deep ore bodies.

[0022] like Figure 2The diagram shows a flowchart of the screening and verification module for combination of stope structure parameters provided in this embodiment of the invention. The stope structure parameters are obtained through preliminary screening, and the stability objective function and critical value are set. Finally, a stope parameter screening evaluation index is obtained. It is determined whether the obtained stope parameter screening evaluation index is not less than a preset value. If it is not less than the preset value, the stope parameter is positively retained; if it is less than the preset value, the stope parameter is negatively removed. Preliminary verification is performed using the positively retained parameters, and parameter combination verification is performed based on a set dynamic threshold.

[0023] Further understanding is needed regarding the specific steps of the preliminary screening: First, obtain a screening evaluation index for stope parameters using a 3D numerical simulation tool, and compare it with a pre-set screening evaluation index to determine the stope parameter screening type. This pre-set index is determined by pre-selected personnel based on statistical data from historical safe mining cases and geomechanical theory. The stope parameter screening evaluation index quantifies the matching degree between the stope structure parameter combination and the current geological state. Stope parameter screening types include positive retention and negative rejection. Positive retention indicates that the corresponding stope structure parameter combination meets the safety and stability requirements for deep mining; negative rejection indicates that the corresponding stope structure parameter combination does not meet the safety and stability requirements for deep mining. The comparison with the pre-set screening evaluation index is as follows: if the obtained stope parameter screening evaluation index is greater than or equal to the pre-set index, it is determined as positive retention; otherwise, it is determined as negative rejection.

[0024] Specifically, the method for obtaining the screening and evaluation index of stope parameters is as follows: A stability objective function is invoked, using each stope structural parameter in the stope structural parameter combination as a variable. The stability objective data obtained from the three-dimensional numerical simulation is used as the input to the stability objective function. The stope structural parameters include stope span, filling strength, and support spacing. Stope span refers to the horizontal width of the stope along the strike or dip of the ore body, directly related to the exposed area of ​​the stope and the stress state of the surrounding rock. Filling strength is an indicator of the filling body's ability to resist deformation and failure, affecting the support effect of the stope void. Support spacing is the interval between adjacent support structures, determining the constraint range on the surrounding rock. The stability objective function represents the mapping relationship between the stope structural parameter combination and the stability objective data. The stability objective data includes rock mass stress, displacement, and the range of the plastic zone.

[0025] By differentiating the stability objective function, the stability change rate under each parameter combination is calculated, as well as the adaptability of the stope span, filling strength, and support spacing to the current geological conditions. This avoids the subjective errors of traditional qualitative analysis and allows for a direct comparison of the influence of each parameter. Safety thresholds for stress, displacement, and plastic zone are set. These safety thresholds are based on geomechanical theory, historical safe mining case data, and field rock mass test results to ensure they meet actual mining safety requirements. When the stability target data does not exceed the corresponding safety threshold, and the stability change rate under each parameter combination does not exceed the corresponding safety threshold, the integral of the difference between the stability target data and the safety threshold is output. Otherwise, a deduction term for the derivative of the difference with the safety threshold is output. The integral of the difference and the deduction term are superimposed to obtain the stope parameter screening and evaluation index.

[0026] The stability objective function can be expressed as a multi-parameter coupled mapping relationship, for example: f(L,S,D)=k1×L+(-k2)×S+(-k3)×D+C. Where L is the stope span, S is the filling strength, and D is the support spacing; k1, k2, and k3 are the parameter influence coefficients of the stope span, filling strength, and support spacing, respectively. k1, k2, and k3 are all greater than 0. k1 is presented with a positive sign in the stability objective function, but k2 and k3 are negative, reflecting the direction of the corresponding parameter's influence on stability. That is, when the stope span L increases, the stability objective function f increases (corresponding to a decrease in stability, such as deterioration of stress, displacement, and other indicators); when the filling strength S increases and the support spacing D decreases (intensified support), f decreases (corresponding to an increase in stability). If a certain mine has k1=0.5, k2=0.3, k3=0.2, and C=2, when L=8m, S=10MPa, and D=2m, f=0.5×8+(-0.3)×10+(-0.2)×2+2==2.6, this value corresponds to a stress of 25MPa, a displacement of 3mm, and a plastic zone of 0.8m³ in the three-dimensional simulation, thus realizing the quantitative correlation between parameter combination and stability data.

[0027] The specific steps of the preliminary verification are as follows: Based on the combination of stope structure parameters retained after preliminary screening, the rock mass stress increment, displacement accumulation, and plastic zone expansion rate at different time points during the dynamic response of the stope under the current geological conditions are obtained as verification indicators; dynamic stability thresholds for the verification indicators are set, and the simulated extracted indicator values ​​are compared with the dynamic stability thresholds one by one. The dynamic stability thresholds include: the set rock mass stress increment (5%), the set displacement accumulation (2mm / 24h), and the set plastic zone expansion rate (0.1m³ / h); if the verification indicators at all time points do not exceed the dynamic stability thresholds, the parameter combination is determined to have passed the preliminary verification; if a single verification indicator exceeds the corresponding dynamic stability threshold, the deviation of the corresponding single verification indicator is used to further verify the dynamic stability of the stope. The reduction ratio of the stope span is mapped from the field database. After updating the stope span of the constructed response surface model, it is re-verified to reduce the numerical simulation deviation of the response surface model caused by the static assignment of the corresponding stope structural parameters. If a single verification index still exceeds the corresponding dynamic stability threshold after re-verification, a dynamic threshold exceedance warning is issued; otherwise, the parameter combination is deemed to have passed the preliminary verification. If all verification indices exceed the corresponding dynamic stability threshold, a dynamic threshold exceedance warning is issued. If there are multiple verification indices that exceed the corresponding dynamic stability threshold but not all of them, the parameter combination is converted to negative parameter elimination to avoid the subsequent numerical simulation results failing to accurately capture instantaneous stress concentration and displacement mutation under high ground stress due to the synergistic exceedance of multiple indices.

[0028] In this embodiment, the stability objective function is represented as follows: After the effectiveness of the stope structure parameters is verified, a suitable fitting method is selected to construct the function based on the data characteristics. A quadratic polynomial fitting method is chosen to establish an explicit function with stope span, filling strength, and support spacing as independent variables, and stress, displacement, and plastic zone range as dependent variables. The value corresponding to the set dynamic stability threshold is not stable and can be modified according to the application purpose of the current scenario in practical applications. Based on the deviation of the corresponding single verification index, the reduction ratio of the stope span is mapped in the stope database. This is achieved through statistical analysis of historical data with the same geological conditions and index type. The deviation value within each group is used as a variable. A mapping relationship between index deviation input and span reduction ratio output is established through nonlinear fitting, and these correlation models are bound to the corresponding geological condition labels.

[0029] Preliminary verification was conducted based on initial screening. Dynamic monitoring ensured the stability of the parameter combinations, and the adaptability of the combinations was examined over time. Changes in geological conditions during mining were considered to ensure that the verified parameter combinations remained stable not only in the initial state but also maintained safe performance under dynamic changes. Differential processing for different exceedance scenarios avoided misjudging superior parameter combinations due to slight exceedances of a single indicator, and also prevented the risk of multiple indicators exceeding limits simultaneously. This provided a more accurate and reliable parameter set for the generation of subsequent candidate solutions, enhancing the overall optimization system's adaptability to complex geological conditions.

[0030] like Figure 3 The diagram shows the flowchart corresponding to the candidate scheme screening and adaptability verification module provided in this embodiment of the invention. Through multi-scenario simulation and target optimization, spatiotemporal alignment is performed, and constraint vector constraints are applied based on the spatiotemporal alignment. By obtaining the target-fusion features and query vector, it is determined whether the current geological complexity is greater than the set allowable operational complexity. If it is greater, dynamic optimization is performed; if it is not greater, adaptability verification is performed. Dynamic optimization is coupled through a correction factor before adaptability verification is performed. The relative deviation rate is used for determination. Whether the obtained deviation rate is greater than the preset deviation rate is determined. If it is greater, simulation verification is performed; otherwise, the final parameter optimization scheme is determined.

[0031] Furthermore, multi-scenario simulation and target optimization are conducted, specifically including: using the initially screened stope structure parameters and current mining monitoring data as the input dataset, supplemented by geological monitoring data from deep mining environments, and performing spatiotemporal alignment using the Kalman filter algorithm to simultaneously obtain constraint vectors for the current geological conditions. These constraint vectors represent constraint indicators resulting from the spatiotemporal alignment of the current geological condition vector and the mining target vector. Based on the spatiotemporal alignment corresponding to the constraint vectors, the input dataset undergoes spatiotemporal alignment processing. Simultaneously, a convolutional neural network is used for multi-dimensional feature extraction to obtain target-fusion features containing the monitoring trends corresponding to the current stope structure parameters. Specifically, the spatiotemporally aligned stope structure parameters and geological monitoring data are combined to present a time-space-feature network structure. Local correlation features are extracted through multi-layer convolutional operations, followed by pooling and deep convolution to achieve feature aggregation. The pooling layer aggregates local features, while the deep convolutional layer, based on the local features extracted from the shallow layer, further captures global correlations across time and space. Finally, the connection layer integrates these features to obtain multi-dimensional features. The dimensional features are transformed into query vectors to measure the matching degree with the constraints of each manufacturer. The target-fusion feature represents the synergy between the current mining site structure parameters and mining monitoring data in the spatiotemporal dimension. If the geological structure complexity of the mining site area corresponding to the current mining stage is found to be greater than the set allowable complexity, a structural correction coefficient is introduced. The constraint vector is dynamically optimized in combination with the query vector to enhance the adaptability of the constraint vector to the current geological conditions. The structural correction coefficient and the query vector are coupled to generate a directional correction weight matrix. For the specific value of the constraint matching degree in the query vector, the weight of the correction factor in the corresponding dimension of the constraint vector is increased to strengthen the correction. The coupling operation uses the basic constraint value × (1 + structural correction coefficient × query vector). Otherwise, the obtained constraint vector is directly used for adaptability verification. The structural correction coefficient represents the result of mapping the difference between the geological structure complexity and the set allowable complexity in the mining site database. The set allowable complexity is pre-set based on the geological structure complexity of the currently obtained mining site area on the basis of 3D modeling.

[0032] The structural correction coefficient is standardized and transformed into a correction factor that can be directly applied to the constraint vector to quantify the correction magnitude of the constraint vector when the current geological structural complexity exceeds the allowable range. The correction factor is coupled with the original constraint vector to obtain the dynamically adjusted constraint vector. The result of constraining the mining structure parameters based on the dynamically adjusted constraint vector is used as a candidate parameter optimization scheme, so that the corrected constraint vector can more accurately reflect the constraint requirements under complex geological conditions and improve its adaptability to the current geological state.

[0033] The specific process of adaptability verification is as follows: The stope structure parameters in the candidate parameter optimization scheme are compared with the actual stope structure parameters monitored in real time using 3D laser scanning and visual monitoring technologies to obtain the relative deviation rate of each stope structure parameter. When the deviation corresponding to the relative deviation rate exceeds the preset allowable deviation range, it is marked as a parameter scheme to be corrected; otherwise, the adaptability verification is deemed successful. Based on the parameter scheme to be corrected, a historical database of qualified constraint vectors is called to compare the difference dimensions between the current constraint vector and the historical optimal constraint vector, and a backtracking verification of the corresponding constraint vector is performed to locate the source of deviation, such as insufficient correction of geological structures or monitoring data fusion errors. After the constraint vector backtracking verification is completed, the candidate parameter optimization scheme is fine-tuned a second time based on a genetic algorithm. The stope structure parameters in the candidate scheme are used as optimization variables, and a small number of random perturbations are generated. The parameters are combined in a variable manner to form a combined scheme including the original scheme and the nearest neighbor variants. The fitness function takes the deviation of each parameter as the core input, selects the one with the highest fitness, and iteratively recalculates its fitness and deviation to obtain the deviation corresponding to the relative deviation rate of each mining area structural parameter. If the re-obtained deviation does not exceed the preset allowable deviation, the fitness verification is deemed successful, and the final parameter optimization scheme is output. Otherwise, the candidate parameter optimization schemes that exceed the preset allowable deviation are input into the constructed response surface model for simulation verification, the simulation verification results are output, and the preset personnel are prompted to re-check the correction logic of the constraint vectors and perform secondary positioning of the deviation source. If the simulation verification result is qualified, the constraint vector database is added as a new historical qualified sample, and the corresponding final parameter optimization scheme is output. Otherwise, a parameter optimization scheme adaptation failure prompt is sent.

[0034] In this embodiment, spatiotemporal alignment is achieved by defining the temporal attributes of each data point and using the time cycle of the mining cycle as the baseline time sequence to perform time-series calibration on the dynamic monitoring data. Ultimately, all data are synchronized to the same time interval (such as each mining cycle), ensuring that the data correspond one-to-one in the time dimension. Spatial alignment is based on the three-dimensional model, where spatial data from different sources are mapped to the corresponding grid cells of the three-dimensional coordinate system through coordinate mapping transformation. The acquisition of constraint vectors is based on geological condition vectors and mining target vectors. Using spatiotemporal alignment as a unified standard, the spatiotemporally aligned geological condition vectors and mining target vectors are integrated. For example, the magnitude of ground stress in the geological condition vector and the safety threshold of the mining span in the mining target vector are integrated into the maximum allowable span of the mining area based on the current ground stress.

[0035] The structural correction coefficient is a key piece of information contained in the historical combination (including geological background information, quantified values ​​of geological structural complexity, quantified values ​​of allowable complexity, corresponding complexity differences, the structural correction coefficient used at that time, and the effect of constraint vector adjustment). It searches for historical data from the same period within the combination that are similar to the current complexity difference. If no completely identical combination exists, the structural correction coefficient is calculated based on the summation and average. The preset allowable deviation range is pre-set according to the current mining structure.

[0036] This example enhances the adaptability of the stope structure parameter optimization scheme to complex mining environments through refined data processing and dynamic constraint adjustment. It performs spatiotemporal alignment of the initially screened stope structure parameters, mining monitoring data, and supplementary geological monitoring data, eliminating misalignments in time series and spatial distribution between different data sources. This establishes a unified and reliable spatiotemporal benchmark for subsequent analysis. Furthermore, multi-dimensional feature extraction accurately captures the synergistic relationship between stope structure parameters and mining monitoring data in the spatiotemporal dimension, providing feature support for the construction of constraint vectors that aligns with actual engineering scenarios. The structural correction coefficients obtained through stope database mapping dynamically adjust key indicators in the constraint vectors, ensuring that constraints respond promptly to changes in geological structures and making the generated candidate parameter optimization schemes more consistent with the actual geological conditions of the current mining area.

[0037] Through multi-level verification and iterative optimization, the engineering reliability and practicality of the mining structure parameter optimization scheme are enhanced. The schemes that deviate from the allowable range and need to be corrected are identified, thus avoiding mining safety risks that may be caused by parameter deviations from the source. The schemes to be corrected are verified by calling a historical constraint vector database with qualified adaptability. By comparing the difference dimensions between the current constraint vector and the historical optimal constraint vector, the source of the deviation can be accurately located, providing a clear direction for subsequent adjustments and providing more reference for optimization under similar geological conditions.

[0038] like Figure 4The flowchart shown is for the dynamic optimization judgment module provided in this embodiment of the invention. The relative deviation score of the parameters and the matching score of the constraints in the obtained adaptability verification quantization parameters are compared with their corresponding set thresholds. If only the relative deviation score of the parameters is not up to standard, three-dimensional numerical simulation parameter optimization is performed. If only the matching score of the constraints is not up to standard, constraint optimization is performed. If neither is up to standard, simultaneous optimization of the two parameters is performed. The constraint optimization couples the actual corrected weights with the constraints and re-determines whether the obtained constraint matching score is up to standard. If it is up to standard, a new constraint is applied with the current constraint vector. Otherwise, a constraint vector dimension expansion prompt is given. If the corresponding relative deviation score of the parameters is not up to standard, the actual adjusted parameters are obtained and coupled, the relative deviation score of the parameters is obtained again and re-determined. If the obtained value still exceeds the threshold, a simulation parameter dimension expansion prompt is given. Otherwise, the current parameters are used as a new simulation parameter combination.

[0039] Furthermore, determining whether to dynamically optimize the 3D numerical simulation parameters and constraints involves the following steps: obtaining adaptability verification quantification parameters, including relative deviation scores for evaluating the degree of deviation between candidate schemes and field data, and constraint matching scores for judging the effectiveness of constraint vectors; the relative deviation score represents the ratio of the actual deviation value of each mining area structural parameter in the final optimized parameter scheme to the corresponding allowable deviation value, such as the ratio of the actual deviation of the mining area span of 2m to the allowable deviation of 3m; the constraint matching score represents the ratio of the number of dimensions corresponding to the constraint vectors in the final optimized parameter scheme to the total number of dimensions corresponding to the constraint vectors; comparing the relative deviation score and the constraint matching score with the corresponding thresholds, recording the quantification parameter items that do not meet the standards, to determine the dynamic optimization type, which includes 3D numerical simulation parameter optimization, constraint optimization, and simultaneous optimization of 3D numerical simulation parameters and constraints.

[0040] Specifically, determining the dynamic optimization type is as follows: if only the relative deviation score of the parameters fails to meet the standard, it is determined to be parameter optimization for three-dimensional numerical simulation; if only the constraint matching score fails to meet the standard, it is determined to be constraint optimization; if both the relative deviation score of the parameters and the constraint matching score fail to meet the standard, it is determined to be synchronous optimization of three-dimensional numerical simulation parameters and constraints; otherwise, it is determined that dynamic optimization of three-dimensional numerical simulation parameters and constraints is not required.

[0041] Specifically, the parameter optimization process for three-dimensional numerical simulation is as follows: For simulation parameters whose relative deviation scores do not meet the standards (such as rock elastic modulus, Poisson's ratio, and internal friction angle), the historical parameter optimization case library is consulted to extract parameter correction rules under similar geological conditions; based on the current field monitoring of rock mass mechanical response data and parameter correction rules, the adjustment coefficients of each simulation parameter are calculated, normalized, and then summed to obtain the actual adjustment coefficients. The rock mass mechanical response data includes real-time stress increment, displacement rate, and plastic zone distribution area. This parameter correction rule is the direction and proportion of adjustment required when the mechanical response is abnormal due to deviation under specific geological scenarios. For example, the parameter correction rule for Poisson's ratio, taking weak rock geology as an example, often revolves around the displacement rate anomaly. In general, when the displacement rate monitored on-site (such as the roof subsidence rate and the lateral displacement rate of the sidewall) exceeds the simulated predicted value (abnormal mechanical response), and the relative deviation score of the Poisson's ratio parameter does not meet the standard, the simulated value of Poisson's ratio is usually too low. Specifically, the correction is to increase the simulated value of Poisson's ratio by a corresponding proportion in the same direction for every certain proportion that the displacement rate exceeds the simulated predicted value. The obtained actual adjustment coefficient is coupled with the original three-dimensional numerical simulation parameters to generate a new combination of simulation parameters. The relative deviation score of the parameters is re-obtained. If the re-obtained relative deviation score of the parameters is not less than the corresponding threshold, the new combination of simulation parameters is used as the optimized three-dimensional numerical simulation parameters. Otherwise, a prompt for expanding the simulation parameter dimension is sent to enhance the adaptability of the response surface model to the deep mining environment.

[0042] The constraint optimization process is as follows: For constraint items whose constraint matching scores do not meet the standard (such as geological structure constraints and mining target constraints), the historical qualified constraint vector database is called to extract effective constraint features under similar geological conditions; based on the current field monitoring of the stope structure environment data, the correction weights of each dimension of the constraint vector under the effective constraint features are calculated, normalized, and then summed to obtain the actual correction weights. The stope structure environment data includes fault activity frequency, geostress value, and high geothermal distribution gradient. Effective constraint features can be geological structure stability constraints and mining target adaptability constraints, etc.; the obtained actual correction weights are coupled with the original constraint vector to generate a new constraint vector. The constraint matching score is re-obtained. If the re-obtained constraint matching score is not less than the corresponding threshold, the new constraint vector is used as the optimized constraint condition; otherwise, a constraint vector dimension expansion prompt is sent to improve the coverage of the constraint vector for the complex environment of deep mining.

[0043] In this embodiment, the thresholds corresponding to the relative deviation score of parameters and the matching score of constraints are preset based on the final parameter optimization scheme and the purpose under the current three-dimensional numerical simulation scenario. The actual adjustment coefficients are calculated based on the optimization records related to the same geological conditions in the case library, combined with the monitored data, to obtain the dynamic adjustment coefficients of each simulation parameter. The method for obtaining the correction weights is similar.

[0044] By combining historical experience with actual field data, the simulation parameters' fit to the geological environment of deep mining was improved, reducing the deviation between numerical simulation results and engineering reality. During the optimization process, a historical parameter optimization case library was consulted to extract parameter correction patterns under similar geological conditions, providing reliable empirical basis for adjustments and avoiding deviations in adjustment direction due to a lack of reference. After coupling the adjustment coefficients with the original simulation parameters to generate new parameter combinations, the adaptability of the response surface model to the deep mining environment was further enhanced. This ensured that the final optimized simulation parameters not only matched historical successful experience but also fit the actual rock mass state of the current mining area, providing a high-quality simulation foundation for the accurate optimization of subsequent mining area structure parameters.

[0045] The combination of constraint condition optimization and overall dynamic optimization judgment process improves the coverage and adaptability of constraint vectors to the complex environment of deep mining, while ensuring the accuracy of optimization direction and avoiding ineffective adjustments. By calling the historical qualified constraint vector database to extract effective constraint features, it can quickly locate the constraint direction that meets the engineering requirements under similar geological conditions. Combined with the current monitored stope structure environment data, the correction weights can accurately reflect the importance of different constraint items in the current deep mining environment. The newly generated constraint vectors are more in line with actual needs, which not only improves optimization efficiency, but also ensures that the final output stope structure parameter optimization scheme meets the requirements, and reduces the mining risks caused by unsuitable constraints or simulation parameter deviations.

[0046] like Figure 5 The flowchart shown is for an intelligent optimization method for stope structure parameters based on three-dimensional numerical simulation. The processing flow of this method may include the following steps: Step 1, in the deep ore body mining stage, preliminary screening and verification of stope structure parameter combinations within the preset stope range are performed using three-dimensional numerical simulation tools; Step 2, the acquired stope structure parameters and mining monitoring data are input into the constructed response surface model for multi-scenario simulation and target optimization, outputting candidate parameter optimization schemes and simultaneously performing adaptability verification; Step 3, based on the results of the adaptability verification, it is determined whether to dynamically adjust the three-dimensional numerical simulation parameters and constraints to improve the adaptability between the stope structure parameter scheme and the on-site geological conditions.

[0047] In this embodiment, a progressive optimization process for stope structure parameters is established, which improves the efficiency of parameter optimization and the adaptability of the solution. The first step is to conduct preliminary screening and verification through three-dimensional numerical simulation, eliminating unsuitable items from the preset parameter combinations, narrowing the optimization range, and reducing redundant calculations in subsequent steps. The second step combines mining monitoring data and response surface models to conduct multi-scenario simulation optimization, and simultaneously performs adaptability verification to reduce the interference of model fitting errors on the optimization results and produce more feasible candidate solutions. The third step dynamically adjusts the simulation parameters and constraints based on the verification results, further ensuring that the final solution conforms to the on-site geological conditions, and comprehensively ensuring the safety and applicability of the parameter solution in deep ore body mining.

[0048] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, 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., infrared, wireless, microwave, etc.) means. A 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 includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0049] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0050] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0051] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0052] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0053] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation, characterized in that, include: The stope structure parameter combination screening and verification module is used to conduct preliminary screening and verification of stope structure parameter combinations within a preset stope range through three-dimensional numerical simulation tools during the deep ore body mining stage, so as to narrow down the optimization range of stope structure parameters. The candidate scheme screening and adaptability verification module is used to input the acquired stope structure parameters and mining monitoring data into the constructed response surface model for multi-scenario simulation and target optimization, output candidate parameter optimization schemes, and simultaneously perform adaptability verification to reduce the interference of the fitting error of the response surface model on the optimization of stope parameters. The dynamic optimization judgment module is used to determine whether to dynamically optimize the three-dimensional numerical simulation parameters and constraints based on the results of the adaptability verification, so as to improve the adaptability between the mining area structure parameter scheme and the on-site geological conditions.

2. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 1, characterized in that, The specific steps of the preliminary screening are as follows: The selection and evaluation index of the stope parameters is obtained by using a three-dimensional numerical simulation tool and compared with the set parameter selection and evaluation index to determine the type of stope parameter selection. The mining parameter screening and evaluation index is used to quantify the degree of matching between the combination of mining structure parameters and the current geological conditions. The filtering types for the stope parameters include positive retention of stope parameters and negative removal of stope parameters; The positive retention of the mining parameters indicates that the corresponding combination of mining structure parameters meets the safety and stability requirements of deep mining. The negative elimination of the stope parameters indicates that the corresponding combination of stope structure parameters does not meet the safety and stability requirements of deep mining. The comparison with the set parameter screening and evaluation index is specifically as follows: If the obtained screening and evaluation index of the mining parameters is greater than or equal to the set screening and evaluation index, the mining parameters are determined to be positively retained; otherwise, the mining parameters are determined to be negatively removed.

3. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 2, characterized in that, The specific method for obtaining the screening and evaluation index of the mining parameters is as follows: The stability objective function is invoked with each stope structure parameter in the stope structure parameter combination as a variable, and the stability objective data obtained from the three-dimensional numerical simulation is used as the input of the stability objective function. The stope structure parameters include stope span, filling strength and support spacing. The stability objective function represents the mapping relationship between the combination of stope structure parameters and the stability objective data, which includes rock mass stress, displacement, and plastic zone range. By differentiating the stability objective function, the rate of change of stability under each parameter combination is calculated, which characterizes the degree of adaptation of each mining area structural parameter to the current geological condition. Set the safety threshold values ​​for stress, displacement, and plastic zone. When the stability target data does not exceed the corresponding safety threshold value and the stability change rate under each parameter combination does not exceed the corresponding safety threshold value, output the integral of the difference between the stability target data and the safety threshold value; otherwise, output the derivative deduction term of the difference between the stability target data and the safety threshold value. The difference integral and the difference derivative deduction term are superimposed to obtain the mining parameter screening and evaluation index.

4. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 3, characterized in that, The specific steps of the preliminary verification are as follows: Based on the combination of mining structure parameters retained after preliminary screening, the rock mass stress increment, displacement accumulation and plastic zone expansion rate at different time points during the dynamic response of the mining under the current geological conditions are obtained as verification indicators. Set dynamic stability thresholds for verification indicators, and compare the simulated extracted indicator values ​​with the dynamic stability thresholds one by one. The dynamic stability thresholds include: set rock mass stress increment, set displacement accumulation, and set plastic zone expansion rate. If the verification metrics at all time points do not exceed the dynamic stability threshold, the parameter combination is deemed to have passed the preliminary verification. If a single verification indicator exceeds the corresponding dynamic stability threshold, the reduction ratio of the stope span is mapped in the stope database based on the deviation of the corresponding single verification indicator. The stope span of the constructed response surface model is then updated and re-verified to reduce the numerical simulation deviation of the response surface model caused by the static assignment of the corresponding stope structure parameters. If, after re-verification, a single verification indicator still exceeds the corresponding dynamic stability threshold, a dynamic threshold exceeding warning will be issued; otherwise, the parameter combination will be deemed to have passed the initial verification. If all verification indicators exceed the corresponding dynamic stability threshold, a dynamic threshold exceedance warning will be issued. If a single verification index exceeds the corresponding dynamic stability threshold but not all of them do, then the combination of parameters will be converted into a negative parameter elimination to avoid the subsequent numerical simulation results failing to accurately capture instantaneous stress concentration and displacement mutation under high ground stress due to multiple indexes exceeding the limit in tandem.

5. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 1, characterized in that, The multi-scenario simulation and target optimization specifically include: The preliminary screened mining structure parameters and the mining monitoring data of the current mining stage are used as the input dataset, supplemented by geological monitoring data in the deep mining environment. The Kalman filter algorithm is used for spatiotemporal alignment to synchronously obtain the constraint vector of the current geological conditions. The constraint vector represents the constraint index after spatiotemporal alignment of the current geological condition vector and the mining target vector. Based on the spatiotemporal alignment of the constraint vector, the input dataset is spatiotemporally aligned. At the same time, a convolutional neural network is used to extract multi-dimensional features to obtain target-fusion features that contain the monitoring trend corresponding to the current mining structure parameters. These features are then transformed into query vectors to measure the matching degree between the input dataset and the constraints of each manufacturer. The target-fusion features represent the synergy between the current mining structure parameters and the mining monitoring data in the spatiotemporal dimension. If the geological structure complexity of the mining area corresponding to the current mining stage is detected to be greater than the set allowable complexity, a structural correction coefficient is introduced and the constraint vector is dynamically optimized in combination with the query vector to enhance the adaptability of the constraint vector to the current geological conditions. Otherwise, the obtained constraint vector is directly used for adaptability verification. The structural correction coefficient represents the result of mapping the difference between the geological structure complexity and the set allowable complexity in the mining area database.

6. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 5, characterized in that, The introduction of constructing correction coefficients for dynamic optimization of the constraint vector is specifically as follows: The structural correction coefficients are standardized and then converted into correction factors that act on the constraint vectors, so as to quantify the magnitude of the correction to the constraint vectors when the current geological structural complexity exceeds the allowable range. The correction factor is coupled with the original constraint vector to obtain the dynamically adjusted constraint vector. The result of constraining the stope structure parameters based on the dynamically adjusted constraint vector is used as a candidate parameter optimization scheme, so that the corrected constraint vector accurately reflects the constraint requirements under complex geological conditions.

7. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 6, characterized in that, The specific process for performing the compatibility check is as follows: The mining structure parameters in the candidate parameter optimization scheme are compared with the actual mining structure parameters monitored in real time on site to obtain the relative deviation rate of each mining structure parameter. When the deviation of the relative deviation rate exceeds the preset allowable deviation, it is marked as a parameter scheme to be corrected; otherwise, the adaptability verification is deemed to have passed. Based on the parameter scheme to be corrected, call the historical constraint vector database with qualified adaptability, compare the difference dimensions between the current constraint vector and the historical best constraint vector, and perform backtracking verification of the corresponding constraint vector to locate the source of deviation. After the constraint vector backtracking verification is completed, the candidate parameter optimization scheme is fine-tuned a second time based on the genetic algorithm to re-obtain the deviation range corresponding to the relative deviation rate of each mining area structural parameter. If the re-obtained deviation range does not exceed the preset allowable deviation range, the adaptability verification is determined to be passed and the final parameter optimization scheme is output. Conversely, if the deviation exceeds the preset allowable deviation range, the candidate parameter optimization scheme will be input into the constructed response surface model for simulation verification, the simulation verification results will be output, and the preset personnel will be prompted to re-check the constraint vector correction logic and perform secondary location of the deviation source. If the simulation verification result is qualified, the constraint vector database is added as a new historical qualified sample, and the corresponding final parameter optimization scheme is output; otherwise, a parameter optimization scheme adaptation failure prompt is sent.

8. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 1, characterized in that, The process of determining whether to dynamically optimize the parameters and constraints of the three-dimensional numerical simulation includes: Obtain the adaptability verification quantification parameters, which include the relative deviation score of parameters used to evaluate the degree of deviation between the candidate solution and the field data, and the constraint matching score used to judge the effectiveness of the constraint vector; The relative deviation score of the parameters represents the ratio of the actual deviation value of each mining area structure parameter to the allowable deviation value of the corresponding parameter in the final parameter optimization scheme; The constraint matching score represents the ratio of the number of dimensions corresponding to the constraint vectors in the final parameter optimization scheme to the total number of dimensions corresponding to the constraint vectors. The relative deviation score of parameters and the matching score of constraints are compared with the corresponding thresholds, and the quantitative parameter items that fail to meet the standards are recorded to determine the dynamic optimization type. The dynamic optimization type includes three-dimensional numerical simulation parameter optimization, constraint optimization, and synchronous optimization of three-dimensional numerical simulation parameters and constraints. The determination of the dynamic optimization type specifically includes: If only the relative deviation score of the parameters fails to meet the standard, it is determined to be parameter tuning for three-dimensional numerical simulation. If only the constraint matching score fails to meet the standard, it is determined to be constraint tuning. If both the relative deviation score of the parameters and the constraint matching score fail to meet the standard, it is determined to be synchronous tuning of the three-dimensional numerical simulation parameters and constraints. Otherwise, it is determined that no dynamic optimization of the three-dimensional numerical simulation parameters and constraints is required.

9. The intelligent optimization system for mining site structure parameters based on three-dimensional numerical simulation as described in claim 8, characterized in that, The specific process for optimizing the parameters in the three-dimensional numerical simulation is as follows: For simulated parameter items whose relative deviation scores do not meet the standards, the historical parameter optimization case library is called to extract the parameter correction rules under similar geological conditions; Based on the current field monitoring data of rock mass mechanical response and parameter correction rules, the adjustment coefficients of each simulated parameter are calculated, and after normalization, they are summed to obtain the actual adjustment coefficients. The rock mass mechanical response data includes real-time stress increment, displacement rate and plastic zone distribution area. The actual adjustment coefficients are coupled with the original three-dimensional numerical simulation parameters to generate a new combination of simulation parameters. The relative deviation score of the parameters is re-acquired. If the re-acquired relative deviation score of the parameters is not less than the corresponding threshold, the new combination of simulation parameters is used as the optimized three-dimensional numerical simulation parameters. Otherwise, a prompt for expanding the dimension of simulation parameters is sent to enhance the adaptability of the response surface model to the deep mining environment. The specific process for optimizing the constraints is as follows: For constraint terms whose constraint matching scores do not meet the standards, the historical qualified constraint vector database is called to extract effective constraint features under similar geological conditions; Based on the current on-site monitoring data of the mining area structure environment, the correction weights of each dimension of the constraint vector under the effective constraint characteristics are calculated, and after normalization, they are summed to obtain the actual correction weights. The mining area structure environment data includes fault activity frequency, ground stress value and high ground temperature distribution gradient. The actual corrected weights are coupled with the original constraint vector to generate a new constraint vector. The constraint matching score is then re-acquired. If the re-acquired constraint matching score is not less than the corresponding threshold, the new constraint vector is used as the optimized constraint. Otherwise, a constraint vector dimension expansion prompt is sent to improve the coverage of the constraint vector for the complex environment of deep mining.

10. A method for intelligent optimization of stope structure parameters based on three-dimensional numerical simulation, applied to the intelligent optimization system for stope structure parameters based on three-dimensional numerical simulation as described in any one of claims 1-9, comprising the following steps: Step 1: In the deep ore body mining stage, a three-dimensional numerical simulation tool is used to conduct preliminary screening and verification of the combination of mining structure parameters within the preset mining area, so as to narrow down the optimization range of mining structure parameters. Step 2: Input the acquired stope structure parameters and mining monitoring data into the constructed response surface model for multi-scenario simulation and target optimization, output candidate parameter optimization schemes, and simultaneously perform adaptability verification to reduce the interference of the fitting error of the response surface model on the optimization of stope parameters. Step 3: Based on the results of the adaptability verification, determine whether to dynamically optimize the three-dimensional numerical simulation parameters and constraints to improve the adaptability between the mining area structure parameter scheme and the on-site geological conditions.

Images