Strip mine slope multi-element engineering geological information identification and fusion modeling method

Abstract

The invention belongs to the field of mining engineering multi-element geological information modeling, and provides a strip mine slope multi-element engineering geological information identification and fusion modeling method, which comprises the following steps: acquiring core images in different areas of a strip mine, and carrying out lithology identification based on the core images to obtain a core lithology data set; acquiring a surface rock stratum image of the strip mine through aerial survey of an unmanned aerial vehicle, and performing lithology identification based on the surface rock stratum image of the strip mine to obtain a surface lithology data set; performing slope water outlet point identification on the surface rock stratum image of the strip mine to obtain a slope water outlet point data set; and carrying out space coordinate quantification on the rock core lithology data set, the surface lithology data set and the slope water outlet point data set, and carrying out fusion modeling to generate a strip mine slope geologic model. According to the method, various engineering geological parameters such as lithology, side slope characteristics, underground water conditions and water outlet point distribution can be effectively integrated, precise quantification of space coordinates of engineering geological characteristics is achieved, and the efficiency and precision of geological modeling are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the field of multi - source geological information modeling in mining engineering, and specifically discloses a method for identifying and fusing multi - source engineering geological information of open - pit mine slopes. Background Technique

[0002] The slope stability in open - pit mining is one of the key issues to ensure mining safety and production efficiency in the mining area. With the continuous increase of mining depth, the geological factors affecting the stability of mine slopes become more complex and variable. Traditional slope stability analysis methods often rely on a single source of geological data, such as borehole data or geological exploration results. However, in practical applications, these data are often incomplete, and there are problems such as uneven spatial distribution or difficult acquisition. Therefore, how to effectively integrate multi - source geological data and establish a more accurate and comprehensive rock mass mechanics model has become an important challenge in current open - pit mine engineering.

[0003] At present, multi - source geological information identification and fusion technology has been applied to some extent in engineering geology and rock mass mechanics research. However, most methods are still based on a single data source and are difficult to handle slope stability problems in complex geological environments. In mine exploitation, it is often necessary to combine various geological parameters such as lithology, slope characteristics, and groundwater, and use different data analysis methods to improve the accuracy and operability of the model. However, existing geological information fusion models generally have problems such as low data - processing efficiency, insufficient fusion accuracy, and ignoring the spatial correlation between data, resulting in poor effects on slope stability analysis and prediction. Therefore, it is very necessary to study and design a new method for identifying and fusing multi - source engineering geological information of open - pit mine slopes, which is based on multi - source geological parameters, and through data fusion technology and modeling algorithms, comprehensively considers the multiple influences of geological parameters to solve the problems existing in current open - pit mine modeling. Summary of the Invention

[0004] In order to solve the problems of low data - processing efficiency, insufficient fusion accuracy, and ignoring the spatial correlation between data in current open - pit mine modeling, the present invention proposes a method for identifying and fusing multi - source engineering geological information of open - pit mine slopes.

[0005] The present invention provides a method for identifying and fusing multi - source engineering geological information of open - pit mine slopes, including the following steps: S1. Collect core images in different areas of the open - pit mine, and perform lithology identification based on the core images to obtain a core lithology data set; S2. Obtain surface rock layer images of the open - pit mine through unmanned aerial vehicle (UAV) aerial survey, and perform lithology identification based on the surface rock layer images of the open - pit mine to obtain a surface lithology data set; S3. Identify the slope water outlet points from the obtained surface rock layer images of the open-pit mine to obtain a slope water outlet point data set; S4. Quantify the spatial coordinates of the core lithology data set obtained in step S1, the surface lithology data set obtained in step S2, and the slope water outlet point data set obtained in step S3, and perform fusion modeling to generate an open-pit mine slope geological model.

[0006] According to an open-pit mine slope multi-source engineering geological information identification and fusion modeling method of some embodiments of the present application, step S1 includes: S1.1. Collect core photos in different areas of the open-pit mine, perform image processing on the core photos to obtain core images, and record the borehole number, burial depth, and core placement order corresponding to each core image; S1.2. Perform lithology annotation on the processed core images to obtain a core image data set, and divide the core image data set into a training set and a test set for training and testing the Mask R-CNN model; S1.3. Use the training set to pre-train the Mask R-CNN model, and optimize the network parameters of the Mask R-CNN model until the preset target is reached; S1.4. Input the test set into the trained Mask R-CNN model, generate a target mask and output a lithology identification result to obtain the core lithology data set.

[0007] According to an open-pit mine slope multi-source engineering geological information identification and fusion modeling method of some embodiments of the present application, in step S1.2, the core image data set includes core images collected in different areas of the open-pit mine and lithology annotations corresponding to the core images.

[0008] According to an open-pit mine slope multi-source engineering geological information identification and fusion modeling method of some embodiments of the present application, in step S1.3, the pre-training of the Mask R-CNN model using the training set includes: S1.3.1. Set initial parameters and adjust the learning rate; S1.3.2. Input the training set into the Mask R-CNN network for forward propagation to obtain a prediction result; S1.3.3. Compare the prediction result with the data label to obtain a validation loss value; S1.3.4. Perform backpropagation and use the mini-batch gradient descent method to adjust the network parameters; S1.3.5. Loop through S1.3.1 to S1.3.4 in sequence until the loss value reaches the preset target.

[0009] A method for identifying and fusing multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, the step S2 includes: S2.1. Obtain surface photos of the open-pit mine based on UAV oblique photography; S2.2. Make a surface rock layer image of the open-pit mine based on the surface photos of the open-pit mine and perform lithology annotation to obtain a surface rock layer image data set of the open-pit mine; S2.3. Adapt the trained Mask R-CNN model to the surface lithology recognition task through transfer learning, input the surface rock layer image data set of the open-pit mine into the trained Mask R-CNN model, output the surface lithology recognition result, and obtain a surface lithology data set.

[0010] A method for identifying and fusing multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, in the step S2.2, the surface rock layer image data set of the open-pit mine includes surface rock layer images of the open-pit mine, rock types, rock textures, rock structures, and lithology annotations.

[0011] A method for identifying and fusing multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, the step S3 includes: S3.1. Make a water outlet point image based on the surface rock layer image of the open-pit mine and perform water outlet point annotation, convert the slope water outlet point recognition into an object detection problem, obtain a water outlet point image data set, the water outlet point image data set includes water outlet point images under different slope environments and corresponding water outlet interface annotations, and divide the water outlet point image data set into a water outlet point training set and a water outlet point test set; S3.2. Construct a single-stage object detection algorithm, pre-train the single-stage object detection algorithm with the water outlet point training set, and test the trained single-stage object detection algorithm with the water outlet point test set. The single-stage object detection algorithm includes: Backbone backbone network, used to extract multi-scale features from the input water outlet point image; Neck feature fusion module, used to fuse feature information at different levels; Head detection head, used to output the class probability and bounding box position of the water outlet point to obtain a prediction result; Loss Function loss function, used to calculate the difference between the prediction result and the water outlet interface annotation; S3.3. Detect the water outlet points in the water outlet point image data set through the tested single-stage object detection algorithm, divide each surface rock layer image of the open-pit mine into grids, predict multiple bounding boxes for each grid, and output the water outlet point class probability and bounding box offset to obtain a water outlet point detection result; S3.4. Associate the water outlet detection result and the position of the water outlet in the image coordinate system with the geographic coordinate information obtained by the UAV aerial survey to obtain the slope water outlet data set containing spatial position information.

[0012] A method for identifying and fusing and modeling multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, the step S4 includes: S4.1. Quantify the data in the core lithology data set, find the actual three-dimensional coordinate position according to the borehole number, and quantify the longitude, latitude, elevation, borehole depth, buried depth and corresponding lithology of the core image to obtain a borehole database; S4.2. Quantify the data in the surface lithology data set and the slope water outlet data set. Based on the UAV aerial survey image, extract the target mask through an image recognition algorithm, and combine the UAV attitude angle and the surface elevation point cloud data to quantify the three-dimensional coordinates of the outcropping lithology interface and the three-dimensional coordinates of the slope water outlet to obtain a slope lithology database and a slope water outlet database; S4.3. Integrate the borehole database, the slope lithology database and the slope water outlet database obtained after coordinate quantification to construct a slope layered model database with three-dimensional coordinate information; S4.4. Based on the slope layered model database, generate a lithology layered model through 3Dmine software, and the lithology layered model includes the surface, soil layer, rock layer and coal seam before and after excavation; S4.5. According to the contour coordinates of the lithology boundary in the lithology layered model, add a surface entity in the GIS platform and assign lithology attributes to realize the slope lithology model; S4.6. Based on the slope water outlet coordinates in the slope layered model database, dynamically draw a water outlet range model through WebGIS; S4.7. Integrate the lithology layered model, the slope lithology model and the water point range model into a multi-source information database; S4.8. Dynamically construct and update the open-pit mine slope geological model based on the multi-source information database using Kriging interpolation.

[0013] A method for identifying and fusing and modeling multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, in the step S4.6, the dynamic drawing of the water outlet range model is realized by adding an entity surface through WebGIS of Cesium.

[0014] A method for identifying and fusing and modeling multi-source engineering geological information of an open-pit mine slope according to some embodiments of the present application, in the step S4.8, the dynamic update of the open-pit mine slope geological model is based on the real-time data input of the slope layered model database.

[0015] A method for identifying and fusing multi-source engineering geological information of open-pit mine slopes according to the present invention can effectively integrate various engineering geological parameters such as lithology, slope characteristics, groundwater conditions, and distribution of water outlet points. By constructing a deep learning model, intelligent identification of core lithology, slope rock strata, and slope water outlet points is realized, greatly improving the automation level of geological information acquisition. Innovatively, the UAV aerial survey technology is combined with the image recognition algorithm to accurately quantify the spatial coordinates of engineering geological features. Using the multi-source data fusion method, core data, slope lithology data, and water outlet point data are organically integrated to establish a unified engineering geological information database, overcoming the limitations of single data sources and making the engineering geological model more accurate. Based on geostatistics and geographic information technology, efficient 3D distribution modeling of lithology and characterization technology for the range of water outlet points are developed, significantly improving the efficiency and accuracy of geological modeling. Through the WebGIS platform, 3D visualization display of the engineering geological model is realized, supporting dynamic update and maintenance of the model. This method effectively solves the problems of low efficiency, high manual dependence, and lagging model update of traditional geological exploration methods, provides a complete set of intelligent technology solutions for engineering geological investigation and monitoring of open-pit mine slopes, and also effectively enhances the ability to evaluate the safety of mine exploitation under complex geological conditions, which is of great significance for improving the safety production level of mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic flow chart of a method for intelligent identification and fusion modeling of multi-source engineering geological information of open-pit mine slopes provided in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The following further describes in detail the embodiments of the present invention with reference to the drawings and embodiments. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0018] Embodiment 1. Aiming at the problems existing in current geological modeling, based on the identification of multi-source engineering geological information and considering the extraction of UAV aerial survey image information and multi-source data fusion, a method for intelligent identification and fusion modeling of multi-source engineering geological information of open-pit mine slopes is proposed. This embodiment provides a method for intelligent identification and fusion modeling of multi-source engineering geological information of open-pit mine slopes, as Figure 1 shown, including the following steps: S1. Collect core images in different areas of the open-pit mine, and perform lithology identification based on the core images to obtain a core lithology data set.

[0019] S2. Obtain the surface rock stratum images of the open-pit mine through UAV aerial survey, and perform lithology identification based on the surface rock stratum images of the open-pit mine to obtain a surface lithology data set.

[0020] S3. Identify the slope water outlet points from the surface rock stratum images of the open-pit mine obtained in step S2 to obtain a slope water outlet point data set.

[0021] S4. Quantify the spatial coordinates of the core lithology data set obtained in step S1, the surface lithology data set obtained in step S2, and the slope water outlet point data set obtained in step S3, and perform fusion modeling to generate an open-pit mine slope geological model.

[0022] A method for identifying and fusing multi-source engineering geological information of an open-pit mine slope in this embodiment realizes the construction of a model for characterizing the three-dimensional spatial distribution of lithology and the slope water outlet range through the intelligent identification and coordinate quantification of the core lithology of the open-pit mine, the surface lithology of the open-pit mine, and the slope water outlet points of the open-pit mine slope, and combines geostatistical methods and geographic information technology, improving the intelligent level of obtaining engineering geological information while enhancing the model construction efficiency and reliability.

[0023] Embodiment 2. This embodiment provides a method for identifying and fusing multi-source engineering geological information of an open-pit mine slope, including the following steps: S1. Collect core images in different areas of the open-pit mine, and perform lithology identification based on the core images to obtain a core lithology data set.

[0024] As a preference of this embodiment, specifically, step S1 includes: S1.1. Collect core photos in different areas of the open-pit mine, perform image processing on the core photos to obtain core images, and record the borehole number, burial depth, and core placement order corresponding to each core image; S1.2. Perform lithology annotation on the processed core images to obtain a core image data set, and divide the core image data set into a training set and a test set for training and testing the Mask R-CNN model; In step S1.2, the core image data set includes the core images collected in different areas of the open-pit mine and the lithology annotations corresponding to the core images; S1.3. Use the training set to pre-train the Mask R-CNN model, and optimize the network parameters of the Mask R-CNN model until the preset target is reached; In step S1.3, using the training set to pre-train the Mask R-CNN model includes: S1.3.1. Set the initial parameters and adjust the learning rate. More specifically, when adjusting the learning rate, the learning rate can be appropriately lowered; S1.3.2. Input the training set into the Mask R-CNN network for forward propagation to obtain a prediction result. More specifically, after inputting the training set into the Mask R-CNN network, it will go through layer-by-layer calculations for forward propagation to obtain a prediction result; S1.3.3. Compare the prediction result with the data label to obtain the validation loss value; S1.3.4. Perform backpropagation and use mini-batch gradient descent to adjust the network parameters; S1.3.5. Loop through S1.3.1 to S1.3.4 in sequence until the loss value reaches the preset target; In this embodiment, the training process of pre-training the Mask R-CNN model on the training set can be regarded as a process of "iterative fitting". When the validation loss value or prediction accuracy changes little, stop the iteration. The trained Mask R-CNN model can be used to identify single-row cores from core images; S1.4. Input the test set into the trained Mask R-CNN model to generate a target mask and output the lithology identification result, obtaining a core lithology dataset. More specifically, input the test set into the trained Mask R-CNN model, and the trained Mask R-CNN model will fill the cores in the core images in different regions with color blocks to distinguish different lithologies.

[0025] S2. Obtain the surface rock layer images of the open-pit mine through UAV aerial survey, and perform lithology identification based on the surface rock layer images of the open-pit mine to obtain a surface lithology dataset.

[0026] As a preference of this embodiment, specifically, step S2 includes: S2.1. Obtain the surface photos of the open-pit mine based on UAV oblique photography. More specifically, the surface photos of the open-pit mine cover different rock types, textures, and structures in different geological conditions of the open-pit mine; S2.2. Produce the surface rock layer images of the open-pit mine based on the surface photos of the open-pit mine and perform lithology annotation to obtain a surface rock layer image dataset of the open-pit mine; In step S2.2, the surface rock layer image dataset of the open-pit mine includes the surface rock layer images of the open-pit mine, rock types, rock textures, rock structures, and lithology annotations; the surface rock layer image dataset of the open-pit mine can be divided into a surface rock layer image training set and a surface rock layer image test set of the open-pit mine for training and testing the Mask R-CNN model.

[0027] S2.3. Adapt the trained Mask R-CNN model to the surface lithology identification task through transfer learning, input the surface rock layer image dataset of the open-pit mine into the trained Mask R-CNN model, and output the surface lithology identification result to obtain a surface lithology dataset.

[0028] S3. Identify the slope water outlet points from the surface rock layer images of the open-pit mine obtained in step S2 to obtain a slope water outlet point dataset.

[0029] Preferably, in this embodiment, specifically, step S3 includes: S3.1. Produce a water outlet point image based on the surface rock layer image of the open-pit mine and perform water outlet point annotation, convert the slope water outlet point recognition into an object detection problem, obtain a water outlet point image dataset, the water outlet point image dataset includes water outlet point images and corresponding water outlet interface annotations under different slope environments, and divide the water outlet point image dataset into a water outlet point training set and a water outlet point test set; in this embodiment, the unmanned aerial vehicle (UAV) aerial survey automatically collects high-quality surface rock layer images of the open-pit mine, providing crucial data support for the single-stage object detection algorithm for image recognition. With the collected surface rock layer image data of the open-pit mine, convert the slope water outlet point recognition into an object detection problem, and the intelligent algorithm can automatically identify and classify the objects in the surface rock layer image of the open-pit mine; S3.2. Construct a single-stage object detection algorithm, pre-train the single-stage object detection algorithm using the water outlet point training set, and test the trained single-stage object detection algorithm using the water outlet point test set. The single-stage object detection algorithm includes: Backbone main network, used to extract multi-scale features from the input water outlet point image; Neck feature fusion module, used to fuse feature information at different levels; Head detection head, used to output the class probability and bounding box position of the water outlet point to obtain the prediction result; Loss Function loss function, used to calculate the difference between the prediction result and the water outlet interface annotation; S3.3. Detect the water outlet points in the water outlet point image dataset through the tested single-stage object detection algorithm, divide each surface rock layer image of the open-pit mine into grids, predict multiple bounding boxes for each grid, and output including the water outlet point class probability and bounding box offset to obtain the water outlet point detection result; S3.4. Associate the water outlet point detection result, the position of the water outlet point in the image coordinate system with the geographic coordinate information obtained by the UAV aerial survey to obtain a slope water outlet point dataset containing spatial position information.

[0030] In the single-stage object detection algorithm in this embodiment, directly extract features from the input water outlet point image, and then output the water outlet point class probability and bounding box, skipping the process of generating candidate regions. This single-stage object detection algorithm can perform efficient and accurate object detection on the water outlet point image in real time. The overall architecture of the single-stage object detection algorithm is based on the deep learning convolutional neural network CNN, and its structure is mainly composed of four modules. The four modules of the Backbone main network, Neck feature fusion module, Head detection head, and Loss Function loss function cooperate with each other to jointly achieve the full-process object detection from image input to target output.

[0031] S4. Quantify the spatial coordinates of the core lithology dataset obtained in step S1, the surface lithology dataset obtained in S2, and the slope water outlet point dataset obtained in S3, and perform fusion modeling to generate an open-pit mine slope geological model.

[0032] As an optimization of this embodiment, specifically, step S4 includes: S4.1. Quantify the data in the core lithology dataset into three-dimensional coordinates. According to the borehole number, find the actual three-dimensional coordinate position, and quantify the longitude, latitude, elevation, borehole depth, buried depth, and corresponding lithology of the core image to obtain a borehole database. In this embodiment, in step S1, core images are collected. Each core image contains the corresponding borehole number, buried depth, and core placement order. According to the borehole number in the image, find the actual position of the borehole or the coordinate position recorded in the data, and then the core photos can be quantified in coordinates to obtain a borehole database containing the longitude, latitude, elevation, borehole depth, buried depth, and corresponding lithology of the borehole core image. S4.2. Quantify the data in the surface lithology dataset and the slope water outlet point dataset into three-dimensional coordinates. Based on the UAV aerial survey images, extract the target mask through an image recognition algorithm, and combine the UAV attitude angles and the surface elevation point cloud data to quantify the three-dimensional coordinates of the outcropping lithology interface and the three-dimensional coordinates of the slope water outlet points to obtain a slope lithology database and a slope water outlet point database. In this embodiment, steps S2 and S3 are both based on the recognition of the open-pit mine surface rock layer images by UAV aerial survey. Therefore, the methods and principles for quantifying the coordinates of the outcropping lithology interface and the slope water outlet points are the same. First, extract the mask of the target in the image through an image recognition algorithm. This mask consists of a list of normalized two-dimensional coordinates, indicating the position of the target in the image coordinate system. Calculate according to the shooting direction of the UAV and the actual surface elevation point cloud data. The shooting direction of the UAV is determined by the attitude angles, namely the yaw angle, pitch angle, and roll angle, while the point cloud model of the UAV provides the elevation information of each point on the surface. By combining the geographical coordinates of the UAV shooting, that is, longitude, latitude, and altitude, and the three-dimensional coordinate system of the point cloud model, the position of the target can be accurately calculated in the actual geographical space. S4.3. Integrate the borehole database, slope lithology database, and slope water outlet point database obtained after coordinate quantification to construct a slope layered model database with three-dimensional coordinate information. In this embodiment, specifically, perform coordinate transformation and format conversion on the multi-source engineering geological information of the open-pit mine surface and the open-pit mine interior obtained by different realistic methods, so that they are fused and stored in a unified database under a unified coordinate system. S4.4. Based on the slope layered model database, a lithologic layered model is generated through 3Dmine software. The lithologic layered model includes the surface, soil layer, rock layer, and coal layer before and after excavation; in this embodiment, specifically, using the cross-section of the open-pit mine, the surface topographic map containing the coordinates of the drill hole orifices, and the open-pit mine boundary map, a three-dimensional engineering geological overall model is constructed through 3DMine and Rhino modeling software; S4.5. According to the contour coordinates of the lithologic demarcation line in the lithologic layered model, a surface entity is added in the GIS platform and given a lithologic attribute to realize the slope lithologic model; in this embodiment, specifically, a ground-attached surface entity is added in the GIS platform and given different colors to identify the lithologic attribute, thus realizing the integrated functions of automatic identification of the lithology of the slope rock mass, real-time upload of the contour coordinates, and intelligent visualization of the slope lithology; S4.6. Based on the coordinates of the slope water outlet points in the slope layered model database, a water outlet point range model is dynamically drawn through WebGIS; In this embodiment, specifically, in step S4.6, the dynamic drawing of the water outlet point range model is realized by adding an entity surface through Cesium's WebGIS; more specifically, the modeling of the water point range model is based on the automatic identification of the UAV oblique photography model. After the identification is completed, the coordinates of the water outlet point range will be generated and stored in the database. Then, the visualization of the water outlet point range model can be realized through the front-end code; in Cesium-based WebGIS, a vector surface can be directly drawn in the three-dimensional model by adding an entity surface, and the color, transparency, etc. of the vector surface can be changed through the material attributes; the surface data coordinates can be directly given in the code, but in order to realize the dynamic update of the water outlet point range, the water outlet point coordinates are directly stored in the database, and the database is read through the post request function at the front end to realize the dynamic drawing of the water outlet point range; S4.7. The lithologic layered model, the slope lithologic model, and the water point range model are integrated into the multi-information database; in this embodiment, specifically, the spatio-temporal integration database of the multi-engineering geological information on the surface and inside the open-pit mine is combined with geostatistical inversion to form a dynamic, fast, accurate, and precise modeling method for the engineering geological model. Based on the geological drilling data as the basic data, using the intelligent identification of the lithology of the core image and the intelligent identification of the lithology of the slope surface, a slope layered model database is constructed. After reading the database in 3Dmine software, the lithologic layered model can be directly constructed through commands. The high-precision slope model obtained by the UAV is converted into point cloud data and imported into 3Dmine to construct a DTM surface, which is used to cut the layered model, and finally the constructed lithologic layered model; S4.8. Based on the multi-information database, the geological model of the open-pit mine slope is dynamically constructed and updated using Kriging interpolation; More specifically, the dynamic update of the open-pit slope geological model in step S4.8 is based on the real-time data input of the slope layered model database. In this embodiment, the multi-source engineering geological information databases on the surface and inside the open-pit mine are used as input variables, and the Kriging interpolation is used to quickly construct the engineering geological model, and the engineering geological model is continuously and dynamically modified as the database is updated to maintain its up-to-dateness and accuracy.

[0033] The embodiments of the present invention are given for purposes of illustration and description, and are not exhaustive or limit the invention to the disclosed form. Many modifications and variations are obvious to those of ordinary skill in the art. The embodiments are chosen and described in order to best explain the principles of the invention and its practical application, and to enable those of ordinary skill in the art to understand the invention and design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope, characterized in that, It includes the following steps: S1. Collect core images in different areas of the open-pit mine, perform lithology identification based on the core images, and obtain a core lithology dataset; S2. Obtain the surface rock layer images of the open-pit mine through UAV aerial survey, perform lithology identification based on the surface rock layer images of the open-pit mine, and obtain a surface lithology dataset; S3. Identify the slope water outlet points from the surface rock layer images of the open-pit mine obtained in step S2, and obtain a slope water outlet point dataset; S4. Quantify the spatial coordinates of the core lithology dataset obtained in step S1, the surface lithology dataset obtained in S2, and the slope water outlet point dataset obtained in S3, and perform fusion modeling to generate a geological model of the open-pit mine slope.

2. The method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 1, wherein, The step S1 includes: S1.

1. Collect core photos in different areas of the open-pit mine, perform image processing on the core photos to obtain core images, and record the borehole number, burial depth, and core placement order corresponding to each core image; S1.

2. Perform lithology annotation on the processed core images to obtain a core image dataset, and divide the core image dataset into a training set and a test set; S1.

3. Use the training set to pre-train the Mask R-CNN model, and optimize the network parameters of the Mask R-CNN model until the preset target is reached; S1.

4. Input the test set into the trained Mask R-CNN model, generate a target mask and output the lithology identification result, and obtain the core lithology dataset.

3. A method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 2, characterized in that, In the step S1.2, the core image dataset includes core images collected in different areas of the open-pit mine and lithology annotations corresponding to the core images.

4. A method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 2, characterized in that, In the step S1.3, the pre-training of the Mask R-CNN model using the training set includes: S1.3.

1. Set initial parameters and adjust the learning rate; S1.3.

2. Input the training set into the Mask R-CNN network for forward propagation to obtain a prediction result; S1.3.

3. Compare the prediction result with the data label to obtain a validation loss value; S1.3.

4. Perform backpropagation and use the mini-batch gradient descent method to adjust the network parameters; S1.3.

5. Loop through S1.3.1 to S1.3.4 in sequence until the loss value reaches the preset target.

5. The method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 2, wherein, The step S2 includes: S2.

1. Obtain surface photos of the open-pit mine based on UAV oblique photography; S2.

2. Make surface rock layer images of the open-pit mine based on the surface photos of the open-pit mine and perform lithology annotation to obtain a surface rock layer image dataset of the open-pit mine; S2.

3. Adapt the trained Mask R-CNN model to the surface lithology identification task through transfer learning, input the surface rock layer image dataset of the open-pit mine into the trained Mask R-CNN model, and output the surface lithology identification result to obtain a surface lithology dataset.

6. The method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 5, characterized in that, In the step S2.2, the surface rock layer image dataset of the open-pit mine includes surface rock layer images of the open-pit mine, rock types, rock textures, rock structures, and lithology annotations.

7. A method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 5, characterized in that, The step S3 includes: S3.

1. Generate a water outlet point image based on the surface rock layer image of the open-pit mine and perform water outlet point annotation, convert the identification of slope water outlet points into an object detection problem, obtain a water outlet point image dataset, where the water outlet point image dataset includes water outlet point images under different slope environments and corresponding water outlet interface annotations, and divide the water outlet point image dataset into a water outlet point training set and a water outlet point test set; S3.

2. Construct a single-stage object detection algorithm, pre-train the single-stage object detection algorithm using the water outlet point training set, and test the trained single-stage object detection algorithm using the water outlet point test set. The single-stage object detection algorithm includes: A Backbone main network for extracting multi-scale features from the input water outlet point image; A Neck feature fusion module for fusing feature information at different levels; A Head detection head for outputting the class probability and bounding box position of the water outlet point to obtain a prediction result; A Loss Function for calculating the difference between the prediction result and the water outlet interface annotation; S3.

3. Detect the water outlet points in the water outlet point image dataset using the tested single-stage object detection algorithm, divide each surface rock layer image of the open-pit mine into grids, predict multiple bounding boxes for each grid, and output the class probability of the water outlet point and the bounding box offset to obtain the water outlet point detection result; S3.

4. Associate the water outlet point detection result, the position of the water outlet point in the image coordinate system with the geographic coordinate information obtained by UAV aerial survey to obtain the slope water outlet point dataset containing spatial position information.

8. A method for identifying and fusing and modeling multi-source engineering geological information of an open-pit mine slope according to claim 7, characterized in that, The steps S4 include: S4.

1. Quantify the three-dimensional coordinates of the data in the core lithology dataset, find the actual three-dimensional coordinate position according to the borehole number, and quantify the longitude, latitude, elevation, borehole depth, buried depth and corresponding lithology of the core image to obtain a borehole database; S4.

2. Quantify the three-dimensional coordinates of the data in the surface lithology dataset and the slope water outlet point dataset. Based on the UAV aerial survey image, extract the target mask through an image recognition algorithm, and combine the UAV attitude angle and the surface elevation point cloud data to quantify the three-dimensional coordinates of the outcropping lithology interface and the three-dimensional coordinates of the slope water outlet point to obtain a slope lithology database and a slope water outlet point database; S4.

3. Integrate the borehole database, slope lithology database and slope water outlet point database obtained after coordinate quantification to construct a slope layered model database with three-dimensional coordinate information; S4.

4. Based on the slope layered model database, generate a lithology layered model through 3Dmine software, and the lithology layered model includes the surface, soil layer, rock layer and coal seam before and after excavation; S4.

5. According to the contour coordinates of the lithology dividing line in the lithology layered model, add a surface entity and assign a lithology attribute in the GIS platform to implement the slope lithology model; S4.

6. Based on the slope water outlet point coordinates in the slope layered model database, dynamically draw a water outlet point range model through WebGIS; S4.

7. Integrate the lithologic layered model, the slope lithology model, and the water point range model into the multi-information database; S4.

8. Dynamically construct and update the open-pit slope geological model based on the multi-information database using Kriging interpolation.

9. A method for identifying and fusing and modeling multi-source engineering geological information of an open-pit mine slope according to claim 8, characterized in that, In step S4.6, the dynamic drawing of the water point range model is realized by adding an entity surface through Cesium's WebGIS.

10. A method for identifying and fusing and modeling multi - element engineering geological information of an open - pit mine slope according to claim 8, characterized in that, In step S4.8, the dynamic update of the open-pit slope geological model is based on the real-time data input of the slope layered model database.

Images