Method for calculating pit bottom water storage of surface mine mining area based on three-dimensional software
By using 3D modeling and intelligent calculation, a high-precision terrain model is generated using DIMINE software and RTK technology. Combined with convolutional neural networks for water level prediction and segmentation, the complexity and inaccuracy of water storage calculation in open-pit mines are solved, achieving efficient and accurate water storage assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-05
- Publication Date
- 2026-03-13
AI Technical Summary
In existing technologies, the calculation of water storage capacity at the bottom of open-pit mines is cumbersome and the results are not accurate enough, which affects safe production and environmental protection.
The 3D modeling software DIMINE, combined with RTK technology and convolutional neural networks, is used to generate a high-precision digital terrain model. The model is then vertically cut using a water level elevation prediction model and a multi-layer graph convolutional network to automatically calculate the water storage capacity at the bottom of the mining area.
It improves the accuracy and efficiency of water storage calculation, reduces manual intervention, enables real-time updates and dynamic monitoring, and provides timely decision support.
Smart Images
Figure CN121659403A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mine assessment technology, specifically to a method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software. Background Technology
[0002] As open-pit mining activities continue, benches descend, leading to a gradual reduction in the working area of the lowest bench. During the flood season, if the drainage system fails to respond effectively and promptly, large amounts of water can easily accumulate at the bottom of the mining area, causing flooding. This flooding problem not only directly threatens the safety of mining operations but can also reduce production efficiency and exacerbate the risk of pollution to the surrounding environment. Currently, many mines still rely primarily on CAD technology for preliminary estimations of water storage at the bottom of the mining area. Introducing advanced technologies such as 3D modeling is of great significance for improving the safety level of open-pit mines, promoting the efficient use of resources, and protecting the ecological environment.
[0003] In existing technologies, open-pit mines require the installation of safety platforms, which can lead to irregular shapes in the mining area. When calculating the water storage capacity, it is necessary to first calculate the water storage capacity of each step at the bottom of the pit and then sum them up. For example, if the height of a mine step is 12 meters and the water storage height is 4 meters, this 4 meters needs to be converted into a horizontal projection based on the slope angle and drawn on a CAD drawing before the calculation can be performed. The steps are complicated and cumbersome, and the calculation results are not accurate enough. Summary of the Invention
[0004] The purpose of this invention is to provide a method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software, so as to solve the technical problems of cumbersome water storage calculation and insufficient accuracy of results in the prior art.
[0005] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution: A method for calculating the water storage capacity at the bottom of an open-pit mine using 3D software includes the following steps: Obtain a CAD current status map of the pit bottom topography of the open-pit mine area, and preprocess the CAD current status map; Using the 3D modeling software DIMINE, a high-precision digital terrain model (DTM) was generated based on the CAD current status map of the pit bottom terrain of the open-pit mine area. Using the water level elevation predicted by the pre-established water level elevation prediction model based on RTK technology as the cutting plane, the digital terrain model (DTM) is vertically cut using the 3D modeling software DIMINE to obtain a cut closed body representing the space at the bottom of the pit below the water level. The volume of the cut closed body is calculated to obtain the water storage capacity at the bottom of the open-pit mine.
[0006] As a preferred embodiment of the present invention, the method for preprocessing the CAD current status drawing includes: Turn off or delete layers that are not related to the terrain at the bottom of the pit, and only keep key elements such as contour lines, elevation points, and boundary lines; Check and verify that the elevation values of the contour lines are continuous and without abrupt changes, and correct any error messages: Ensure that the lines are closed and do not overlap, and avoid broken surfaces or gaps when importing into DIMINE.
[0007] Ensure that the coordinate system of the CAD existing drawing (such as WGS84, UTM or local coordinate system) is consistent with the coordinate system set in the DIMINE software.
[0008] As a preferred embodiment of the present invention, the modeling and generation method of a high-precision digital terrain model (DTM) includes: After importing the existing CAD drawing into DIMINE, use the software's editing functions to cut out the terrain area from the bottom of the pit to a specified height to obtain terrain data.
[0009] Using DIMINE's modeling capabilities, a high-precision DTM model is generated based on the terrain data, and the accuracy of the DTM model is corrected by adjusting the DTM model parameters.
[0010] As a preferred embodiment of the present invention, the method for determining the water level elevation at the bottom of the pit includes: Obtain information on the location of the bottom of the mining pit, mining duration, and weather conditions; The location of the bottom of the mining pit, the mining duration, and the climate conditions are input into the water level elevation prediction model, which then outputs the water level elevation at the bottom of the pit.
[0011] As a preferred embodiment of the present invention, the method for constructing the water level prediction model includes: RTK technology was used to measure water level elevations at different measurement locations, during mining operations, and under varying climatic conditions. This forms an RTK measurement dataset; Data augmentation is performed on the RTK measurement dataset to obtain the augmented dataset; In the enhanced training set, the convolutional neural network is trained with the measurement location, mining duration, and climate conditions as inputs and the water level elevation as output to obtain the water level elevation prediction model. The water level elevation prediction model is as follows: Where h is the water level elevation output by the water level elevation prediction model, (x,y) is the measurement location, K is the climate condition, T is the mining time, and CNN is the convolutional neural network.
[0012] As a preferred embodiment of the present invention, the loss function of the water level prediction model Where h is the water level elevation output by the water level prediction model. The water level elevation is the corrected RTK measurement, and MSE is the mean square error.
[0013] As a preferred embodiment of the present invention, a method for data augmentation of RTK measurement datasets includes: Each measurement location of the RTK technology is treated as a node, and the water level elevation and measurement location are used as node features of each node. The k-nearest neighbor method is used to obtain the distance between nodes and construct the edges between nodes, resulting in a graph structure composed of nodes and edges. A multi-layer graph convolutional network is constructed based on graph structure. Where A is the adjacency matrix and D is the degree matrix. , These are the node features of the (l+1)th and lth layers, respectively. For the trainable weights of the l-th layer, , I is the identity matrix. Coordinates of the matrix in D The value at that location, for In the matrix coordinates The value at that location, For activation functions; By using a multi-layer graph convolutional network, the feature information of each node is aggregated from its multi-hop neighbor nodes, and the node features are corrected according to spatial consistency. The corrected water level of each node is output by the last convolutional layer. The augmented dataset is obtained by replacing the water level elevation at each measurement location in the RTK measurement dataset with the corrected water level elevation at each node.
[0014] As a preferred embodiment of the present invention, the loss function of the multilayer graph convolutional network... ,in, To preserve fidelity, For smoothness loss, , For hyperparameters, The water level elevation is obtained from the calibrated RTK technology measurement. The water level elevation is obtained by RTK technology. , Let be the water level elevations of the i-th and j-th nodes, respectively, and E be the edge set. It is an L2 norm.
[0015] As a preferred embodiment of the present invention, the method for obtaining the cut closure body includes: The water level elevation predicted by the water level elevation prediction model and corrected by a multi-layer graph convolutional network is used as the cutting benchmark. In DIMINE, the water level elevation corresponds to a horizontal plane. This plane is used as the boundary for cutting, and the mid-section height parameter is adjusted according to the actual terrain features and water storage calculation requirements. In DIMINE, select the DTM model and the water level elevation plane, perform a vertical cut, and automatically generate a closed cut body.
[0016] As a preferred embodiment of the present invention, the method for calculating the water storage capacity at the bottom of the open-pit mine includes: During the cutting process, the DIMINE software calculates and sums the volume of each grid cell in the DTM model below the water level to obtain the water storage volume at the bottom of the pit. ,in It is the area of the i-th grid cell in the DTM model. It is the difference between the water level elevation and the ground elevation at the i-th grid cell.
[0017] Compared with the prior art, the present invention has the following advantages: This invention, through 3D modeling technology, can more accurately reflect the complexity and variability of the terrain at the bottom of the mining area, thereby improving the accuracy of water storage calculation. Furthermore, the automated and intelligent calculation process reduces manual intervention and calculation errors, significantly improving work efficiency. Simultaneously, real-time updates and dynamic monitoring keep the data up-to-date, providing timely support for decision-making. Attached Figure Description
[0018] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.
[0019] Figure 1 A flowchart illustrating the method for calculating the water storage capacity at the bottom of an open-pit mine using three-dimensional software, as provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the terrain data of the pit bottom provided in an embodiment of the present invention; Figure 3 A schematic diagram of the DTM model provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of a cutting closure body provided in an embodiment of the present invention; Figure 5 The diagram shows the calculation results of the water level elevation and water storage capacity at the bottom of the pit provided in the embodiment of the present invention. Detailed Implementation
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0021] like Figure 1 As shown, this invention provides a method for calculating the water storage capacity at the bottom of an open-pit mine using three-dimensional software, comprising the following steps: Obtain a CAD current-state map of the pit bottom topography of the open-pit mine area, and preprocess the CAD current-state map; Using the 3D modeling software DIMINE, a high-precision digital terrain model (DTM) was generated based on the CAD current status map of the pit bottom terrain of the open-pit mine area. Using the water level elevation predicted by the pre-established water level elevation prediction model based on RTK technology as the cutting plane, the digital terrain model (DTM) is vertically cut using the 3D modeling software DIMINE to obtain a cut closed body representing the space at the bottom of the pit below the water level. The volume of the cut closed body is calculated to obtain the water storage capacity at the bottom of the open-pit mine.
[0022] This invention utilizes 3D modeling technology, specifically the powerful 3D modeling and spatial analysis capabilities of DIMINE, to model the bottom of a mine pit and accurately calculate water storage capacity.
[0023] This invention, by integrating a high-precision DTM terrain model with water level elevations enhanced by GNN data and intelligently predicted by CNN, completely changes the traditional simplified model and manual experience estimation method, making the water storage calculation results closer to the actual situation and greatly reducing the error.
[0024] Specifically, this invention obtains a CAD view of the bottom of an open-pit mine. Utilizing the layer management function of the CAD software, all layers unrelated to the terrain (such as annotations and text) are turned off or deleted, retaining only key elements such as contour lines, elevation points, and boundary lines. The contour line elevation values are checked and ensured to be continuous, without logical errors, and with closed, non-overlapping lines. Simultaneously, the coordinate system of the CAD drawing (such as the CGCS2000 national coordinate system) is confirmed to be consistent with the coordinate system set in the DIMINE software, preventing data offset at the source.
[0025] This invention imports preprocessed CAD drawings into DIMINE software. Using the software's "DTM Modeling" function, an initial digital terrain model (DTM) is generated based on contour lines and elevation point data. By adjusting the grid size (e.g., setting it to 0.5m-2m according to terrain complexity) and the interpolation method (e.g., using Kriging interpolation to better reflect spatial correlation), the model is refined, ultimately yielding a high-precision pit bottom DTM model.
[0026] This rigorous preprocessing avoids the problem of "garbage in, garbage out," ensuring the quality of the basic terrain data. This enables the high-precision DTM built upon the terrain data to realistically recreate the complex micro-topography of the pit bottom, laying a solid foundation for subsequent accurate volume calculations and fundamentally surpassing simplified geometric models.
[0027] Methods for preprocessing existing CAD drawings include: Turn off or delete layers that are not related to the topography of the pit bottom (such as labels, text, construction lines, etc.), and only keep key elements such as contour lines, elevation points, and boundary lines; Check and verify that the elevation values of the contour lines are continuous and without abrupt changes, and correct any error messages: Ensure that the lines are closed and do not overlap, and avoid broken surfaces or gaps when importing into DIMINE.
[0028] Ensure that the coordinate system of the CAD existing drawing (such as WGS84, UTM or local coordinate system) is consistent with the coordinate system set in the DIMINE software.
[0029] High-precision digital terrain models (DTMs) can be generated through the following methods: After importing the existing CAD drawing into DIMINE, use the software's editing functions to crop the terrain area from the bottom of the pit to a specified height to obtain terrain data, such as... Figure 2 As shown; Using DIMINE's modeling capabilities, high-precision DTM models can be generated based on terrain data, such as... Figure 3 As shown, the accuracy of the DTM model is corrected by adjusting the DTM model parameters (such as grid size, interpolation method, etc.).
[0030] Furthermore, in order to improve the automation / intelligence level of DIMINE software modeling, this invention establishes a neural network model for automatically predicting key water level elevation parameters, and at the same time establishes a multi-layer graph convolutional network to correct the water level elevation parameters, thereby ensuring the accuracy of water level elevation parameter prediction and realizing high-precision and highly intelligent acquisition of water level elevation parameters.
[0031] Methods for determining the water level elevation at the bottom of the pit include: Obtain information on the location of the bottom of the mining pit, mining duration, and weather conditions; The location of the bottom of the mining pit, the mining duration, and the climate conditions are input into the water level elevation prediction model, which then outputs the water level elevation at the bottom of the pit.
[0032] Methods for constructing water level elevation prediction models include: RTK technology was used to measure water level elevations at different measurement locations, during mining operations, and under varying climatic conditions. This forms the RTK measurement dataset, also known as the raw dataset. RTK measures the elevation of the water surface, and its basic calculation formula is: Elevation = Geodetic Height - Geostationary Anomaly ζ, where: Earth height (h): The height relative to the WGS84 ellipsoid, directly measured by the RTK receiver.
[0033] Height anomaly (ζ): The difference between geodetic height (based on an ellipsoid) and normal height (based on a quasi-geoid, i.e., what we usually call elevation). This value can be obtained through a regional quasi-geoid refinement model.
[0034] The specific steps and calculation process of this invention are as follows: Setting up a base station: Erect an RTK base station at a control point with known coordinates.
[0035] Enter the precise planar coordinates (X, Y) and normal altitude (i.e., elevation) of the control point.
[0036] The base station begins receiving satellite signals and sends differential correction information to the rover via a data link (such as a radio or network).
[0037] Rover station measurement and real-time calculation: The surveyor places the centering rod of the RTK rover station on the water surface at the bottom of the pit (ensuring the water surface is calm and the rod tip is in contact with the water surface).
[0038] The rover station performs two tasks simultaneously: receiving satellite signals and receiving differential signals from the base station.
[0039] Real-time calculation of geodetic height: The mobile station's internal processor uses real-time differential processing to calculate the precise planar coordinates (X, Y) and geodetic height of the current pole tip position. The accuracy can reach the centimeter level.
[0040] Elevation transformation (key step): The RTK handheld device (controller) has a pre-set elevation anomaly model (or geoid model). This model may be a localized mesh model or a parametric model.
[0041] Once the rover station measures the (X, Y, geodetic height) of a point, the handheld device software will automatically interpolate the corresponding elevation anomaly (ζ) based on the plane coordinates of that point.
[0042] Calculating water level elevation: The handheld device software automatically calculates the normal elevation (i.e., altitude) of the point using the formula: Water level elevation = Geodetic height - ζ. Result: The surveyor directly reads the converted, usable water level elevation value on the handheld device screen, while simultaneously recording the measurement time.
[0043] Multiple measuring points are set up at different locations in the water at the bottom of the pit (such as near the shore or in the central area) to obtain the average elevation of the water surface, because the water surface may not be absolutely level.
[0044] Data augmentation is performed on the RTK measurement dataset to obtain the augmented dataset; The method for data augmentation of RTK measurement datasets according to the present invention includes: Each measurement location of the RTK technology is treated as a node, and the water level elevation and measurement location are used as node features of each node. The k-nearest neighbor method is used to obtain the distance between nodes and construct the edges between nodes, resulting in a graph structure composed of nodes and edges. A multi-layer graph convolutional network is constructed based on graph structure. Where A is the adjacency matrix and D is the degree matrix. , These are the node features of the (l+1)th and lth layers, respectively. For the trainable weights of the l-th layer, , I is the identity matrix. Coordinates of the matrix in D The value at that location, for In the matrix coordinates The value at that location, For activation functions; By using a multi-layer graph convolutional network, the feature information of each node is aggregated from its multi-hop neighbor nodes, and the node features are corrected according to spatial consistency. The corrected water level of each node is output by the last convolutional layer. The augmented dataset is obtained by replacing the water level elevation at each measurement location in the RTK measurement dataset with the corrected water level elevation at each node.
[0045] Loss function of multi-layer graph convolutional networks ,in, To preserve fidelity, For smoothness loss, , These are hyperparameters and can be set as needed. The water level elevation is obtained from the calibrated RTK technology measurement. The water level elevation is obtained by RTK technology. , Let be the water level elevations of the i-th and j-th nodes, respectively, and E be the edge set. It is an L2 norm.
[0046] This invention corrects and enhances the measurement results obtained by RTK technology, thereby effectively identifying and correcting RTK measurement anomalies caused by instantaneous fluctuations (such as wind and equipment errors), generating a more reasonable and smoother water level field in space, providing high-quality data labels for subsequent water level elevation prediction model training, thus ensuring the prediction performance of the water level elevation prediction model at the training data level and avoiding its degradation due to poor data quality.
[0047] This invention treats each measurement point as a node in a graph structure, with node features including water level elevation and coordinates. The k-nearest neighbor algorithm is used to construct edges between nodes, forming the graph structure. This graph is then input into a multi-layer graph convolutional network (GCN). GCN uses a message-passing mechanism to allow each node to aggregate information from its neighbors. Its loss function... This includes fidelity loss (ensuring the corrected value doesn't deviate too far from the original measurement) and smoothness loss (forcing spatially adjacent node water level values to be close). Through training, the GNN outputs the water level elevation after spatial consistency correction. This creates an enhanced dataset.
[0048] Among these factors, fidelity loss aims to ensure that the corrected water level elevation does not differ too much from the original measurement, especially for high-precision measurement points. Smoothness loss aims to ensure that the water level elevations of adjacent nodes are similar (because the water surface is smooth).
[0049] In the enhanced training set, the convolutional neural network is trained with the measurement location, mining duration, and climate conditions as inputs and the water level elevation as output to obtain the water level elevation prediction model. The water level elevation prediction model is Where h is the water level elevation output by the water level elevation prediction model, (x,y) is the measurement location, K is the climate condition, T is the mining time, and CNN is the convolutional neural network.
[0050] Loss function of water level prediction model Where h is the water level elevation output by the water level prediction model. The water level elevation is the corrected RTK measurement, and MSE is the mean square error.
[0051] This invention takes the measurement location (x, y), mining duration (T), and climate conditions (K) from the augmented dataset as input, and the corrected water level elevation as input. To output the target, a convolutional neural network (CNN) is trained to establish a prediction model h = CNN(K, T, (x, y)). Thus, in practical applications, only the current location of the pit bottom, the planned mining duration, and weather forecast information need to be input; the model can then predict the future pit bottom water level, achieving a leap from static measurement to dynamic prediction.
[0052] Moreover, this invention utilizes a CNN model to capture the complex nonlinear relationship between multiple factors such as location, time, and climate and water level, achieving a prediction accuracy far exceeding that of simple empirical formulas based on a single factor.
[0053] This invention establishes a water level elevation prediction model that can directly predict water level elevation based on measurement location, mining duration, and climate conditions. By training with an enhanced dataset, the prediction accuracy can be effectively improved, thereby achieving the goal of improving efficiency and accuracy in the 3D modeling process.
[0054] Methods for obtaining cut closed bodies include: The water level elevation predicted by the water level elevation prediction model and corrected by a multi-layer graph convolutional network is used as the cutting benchmark. In DIMINE, the water level elevation corresponds to a horizontal plane. This plane is used as the boundary for cutting, and the mid-section height parameter is adjusted according to the actual terrain features and water storage calculation requirements. In DIMINE, select the DTM model and the water level elevation plane, perform a vertical cut, and automatically generate a closed cut volume, such as... Figure 4 As shown. (That is, the space at the bottom of the pit below the water level) like Figure 5 As shown, the calculation method for the water storage capacity at the bottom of an open-pit mine includes: During the cutting process, the DIMINE software calculates and sums the volume of each grid cell in the DTM model below the water level to obtain the water storage volume at the bottom of the pit. ,in It is the area of the i-th grid cell in the DTM model. It is the difference between the water level elevation and the ground elevation at the i-th grid cell.
[0055] This invention automates the entire process, from water level prediction to volume calculation, within a digital platform, significantly improving work efficiency and freeing up manpower. Furthermore, the 3D visualization model allows technicians to visually inspect the cutting results, and the volume calculation results are based on real terrain, ensuring scientific reliability.
[0056] This invention utilizes graph neural networks (GNNs) to enhance the spatial consistency of raw RTK measurement data, effectively filtering out noise and outliers, thus solving the problem of unstable data quality in traditional measurement methods. The entire process integrates modern surveying, 3D modeling, and artificial intelligence technologies, achieving a high level of intelligence.
[0057] This invention realizes the full-process digitalization and automation of data processing, model building, water level prediction and volume calculation, avoiding cumbersome manual operation and intervention, and shortening the calculation cycle from several days to several hours or even less, providing timely and efficient data support for mine production decisions.
[0058] This invention is not only applicable to open-pit mines, but its technical framework (3D modeling + AI prediction + automated calculation) can also be widely applied to water storage calculation and water level monitoring in reservoirs, tailings ponds and other fields after adaptive adjustments, and has broad market application prospects.
[0059] This invention, through 3D modeling technology, can more accurately reflect the complexity and variability of the terrain at the bottom of the mining area, thereby improving the accuracy of water storage calculation. Furthermore, the automated and intelligent calculation process reduces manual intervention and calculation errors, significantly improving work efficiency. Simultaneously, real-time updates and dynamic monitoring keep the data up-to-date, providing timely support for decision-making.
[0060] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.
Claims
1. A method for calculating the water storage capacity at the bottom of an open-pit mine using three-dimensional software, characterized in that, Includes the following steps: Obtain a CAD current status map of the pit bottom topography of the open-pit mine area, and preprocess the CAD current status map; Using the 3D modeling software DIMINE, a high-precision digital terrain model (DTM) was generated based on the CAD current status map of the pit bottom terrain of the open-pit mine area. Using the water level elevation predicted by the pre-established water level elevation prediction model based on RTK technology as the cutting plane, the digital terrain model (DTM) is vertically cut using the 3D modeling software DIMINE to obtain a cut closed body representing the space at the bottom of the pit below the water level. The volume of the cut closed body is calculated to obtain the water storage capacity at the bottom of the open-pit mine.
2. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 1, characterized in that: The method for preprocessing the CAD current status drawing includes: Turn off or delete layers that are not related to the topography of the pit bottom, and only keep the key elements such as contour lines, elevation points, and boundary lines; Check and verify that the elevation values of the contour lines are continuous and without abrupt changes, and correct any error messages: Ensure that the lines are closed and do not overlap, and avoid broken surfaces or gaps when importing DIMINE; Ensure that the coordinate system of the existing CAD drawing is consistent with the coordinate system set in the DIMINE software.
3. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 2, characterized in that: High-precision digital terrain models (DTMs) can be generated through the following methods: After importing the existing CAD map into DIMINE, use the software's editing function to cut out the terrain area from the bottom of the pit to a specified height to obtain terrain data; Using DIMINE's modeling capabilities, a high-precision DTM model is generated based on the terrain data, and the model accuracy of the DTM is corrected by adjusting the model parameters of the DTM.
4. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 3, characterized in that: Methods for determining the water level elevation at the bottom of the pit include: Obtain information on the location of the bottom of the mining pit, mining duration, and weather conditions; The location of the bottom of the mining pit, the mining duration, and the climate conditions are input into the water level elevation prediction model, which then outputs the water level elevation at the bottom of the pit.
5. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 4, characterized in that: The method for constructing the water level prediction model includes: RTK technology was used to measure water level elevations at different measurement locations, during mining operations, and under varying climatic conditions. This forms an RTK measurement dataset; Data augmentation is performed on the RTK measurement dataset to obtain the augmented dataset; In the enhanced training set, the convolutional neural network is trained with the measurement location, mining duration, and climate conditions as inputs and the water level elevation as output to obtain the water level elevation prediction model. The water level elevation prediction model is as follows: Where h is the water level elevation output by the water level elevation prediction model, (x,y) is the measurement location, K is the climate condition, T is the mining time, and CNN is the convolutional neural network.
6. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software, as described in claim 5, is characterized in that: The loss function of the water level prediction model Where h is the water level elevation output by the water level prediction model. The water level elevation is the corrected RTK measurement, and MSE is the mean square error.
7. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 6, characterized in that: Methods for data augmentation of RTK measurement datasets include: Each measurement location of the RTK technology is treated as a node, and the water level elevation and measurement location are used as node features of each node. The k-nearest neighbor method is used to obtain the distance between nodes and construct the edges between nodes, resulting in a graph structure composed of nodes and edges. A multi-layer graph convolutional network is constructed based on graph structure. Where A is the adjacency matrix and D is the degree matrix. , These are the node features of the (l+1)th and lth layers, respectively. For the trainable weights of the l-th layer, , I is the identity matrix. Coordinates of the matrix in D The value at that location, for In the matrix coordinates The value at that location, For activation functions; By using a multi-layer graph convolutional network, the feature information of each node is aggregated from its multi-hop neighbor nodes, and the node features are corrected according to spatial consistency. The corrected water level of each node is output by the last convolutional layer. The augmented dataset is obtained by replacing the water level elevation at each measurement location in the RTK measurement dataset with the corrected water level elevation at each node.
8. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 7, characterized in that: The loss function of the multi-layer graph convolutional network ,in, To preserve fidelity, For smoothness loss, , For hyperparameters, The water level elevation is obtained from the calibrated RTK technology measurement. The water level elevation is obtained by RTK technology. , Let be the water level elevations of the i-th and j-th nodes, respectively, and E be the edge set. It is an L2 norm.
9. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software, as described in claim 8, is characterized in that: Methods for obtaining cut closed bodies include: The water level elevation predicted by the water level elevation prediction model and corrected by a multi-layer graph convolutional network is used as the cutting benchmark. In DIMINE, the water level elevation corresponds to a horizontal plane. This plane is used as the boundary for cutting, and the mid-section height parameter is adjusted according to the actual terrain features and water storage calculation requirements. In DIMINE, select the DTM model and the water level elevation plane, perform a vertical cut, and automatically generate a closed cut body.
10. The method for calculating the water storage capacity at the bottom of an open-pit mine based on three-dimensional software according to claim 9, characterized in that: The method for calculating the water storage capacity at the bottom of the open-pit mine includes: During the cutting process, the DIMINE software calculates and sums the volume of each grid cell in the DTM model below the water level to obtain the water storage volume at the bottom of the pit. ,in It is the area of the i-th grid cell in the DTM model. It is the difference between the water level elevation and the ground elevation at the i-th grid cell.