An intelligent identification method for potential hazards of open-pit mine slopes based on UAV technology

Abstract

An intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology, which relates to the field of hidden danger identification technology; set measurement points, obtain their vertical point cloud data and inclined point cloud data, construct a point cloud optimization model in combination with measurement parameters, obtain the original point cloud data and measurement parameters of the mine slope, use the point cloud optimization model to obtain optimized point cloud data to construct a three-dimensional model, obtain the real-time image data of the mine slope and generate a real scene model, obtain the slope parameters of the mine slope, obtain the bench height of the mine slope in combination with measurement parameters, construct a slope evaluation model to obtain the slope height of the mine slope, set a height threshold, obtain the super-high area in combination with the slope height, and perform visualization processing on the super-high area; it is beneficial to improve the accuracy of the three-dimensional model and can timely identify and feedback whether the mine slope is super-high.

Description

Technical Field

[0001] The present invention relates to the technical field of hidden danger identification, and specifically to an intelligent identification method for hidden dangers of open-pit mine slopes based on unmanned aerial vehicle technology. Background Art

[0002] Using unmanned aerial vehicles to identify hidden dangers of open-pit mine slopes is an efficient, accurate and low-cost monitoring method. This method combines unmanned aerial vehicle mapping technology, image processing technology and data analysis technology, and has the advantages of strong emergency response ability, high mapping efficiency, high data accuracy, convenient operation, high safety, etc., and plays an important role in mine slope monitoring and disaster prevention;

[0003] In the prior art, when collecting point cloud data of mine slopes, it is often collected obliquely. Therefore, there will be more or less errors in the collected point cloud data, resulting in a lack of accuracy in the constructed three-dimensional model. And in the prior art, whether the mine slope is over-height is often based on simple image recognition, and this method has a large deviation and cannot be accurate to the specific slope position, resulting in inaccurate recognition results. In view of the deficiencies of the prior art, the present invention provides an intelligent identification method for hidden dangers of open-pit mine slopes based on unmanned aerial vehicle technology. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent identification method for hidden dangers of open-pit mine slopes based on unmanned aerial vehicle technology.

[0005] The purpose of the present invention can be achieved through the following technical solutions: An intelligent identification method for hidden dangers of open-pit mine slopes based on unmanned aerial vehicle technology, comprising the following steps:

[0006] Step S1: Set a number of measurement points on the mine slope, use the unmanned aerial vehicle to obtain the vertical point cloud data and inclined point cloud data of each measurement point, obtain the measurement parameters of each inclined point cloud data, and construct a point cloud optimization model in combination with the corresponding vertical point cloud data;

[0007] Step S2: Obtain the original point cloud data of the mine slope and its measurement parameters, use the point cloud optimization model to obtain the optimized point cloud data, and construct a corresponding three-dimensional model;

[0008] Step S3: Obtain the real-time image data of the mine slope, generate a real scene model in combination with the three-dimensional model, obtain the slope parameters of the mine slope in the real scene model, and obtain the bench slope angle and bench height of the mine slope in combination with the measurement parameters;

[0009] Step S4: Construct a slope evaluation model, and use the slope evaluation model to obtain the slope height of the mine slope;

[0010] Step S5: Set a height threshold for the mine slope, obtain the ultra-high region by combining the slope height, and perform visualization processing on the ultra-high region.

[0011] Further, the process of setting several measurement points on the mine slope and using the drone to obtain the vertical point cloud data and inclined point cloud data of each measurement point includes:

[0012] Randomly set several measurement points on a single mine slope and different mine slopes. Taking each measurement point as a reference, obtain the measurement normal of the plane where the measurement point is located on the mine slope, treat the drone as a point for processing, and obtain the measurement distance between the drone and the measurement point;

[0013] Take the angle between the line connecting the drone and the measurement point and the measurement normal as the measurement angle. Take the point cloud data of the measurement point obtained when the drone is on the measurement normal as the vertical point cloud data, and take the point cloud data of the measurement point obtained when the drone moves outside the measurement normal as the inclined point cloud data.

[0014] Further, the process of obtaining the measurement parameters of each item of inclined point cloud data and constructing a point cloud optimization model by combining the corresponding vertical point cloud data includes:

[0015] For a single measurement point, continuously move the drone to obtain multiple items of inclined point cloud data and their corresponding measurement distances and measurement angles, and take the two as measurement parameters;

[0016] Generate a point cloud optimization set according to the inclined point cloud data, measurement parameters and their corresponding vertical point cloud data of different measurement points, and divide it into an optimization training set and an optimization test set;

[0017] Construct a convolutional neural network. Take the different inclined point cloud data, measurement parameters and their corresponding vertical point cloud data in the optimization training set as the input data and output data of the convolutional neural network respectively, and train the convolutional neural network to obtain the first initial convolutional neural network;

[0018] Use the optimization test set to verify the model of the first initial convolutional neural network, and output the first initial convolutional neural network with an error less than or equal to the preset first test error threshold as the point cloud optimization model.

[0019] Further, the process of obtaining the original point cloud data and its measurement parameters of the mine slope, using the point cloud optimization model to obtain the optimized point cloud data, and constructing the corresponding 3D model includes:

[0020] In the actual application scenario, use the drone to collect the original point cloud data of each point on the entire mine slope. Take the corresponding point of the normal where the drone is located when collecting the mine slope on the mine slope as the actual measurement point, and obtain the actual measurement normal;

[0021] Obtain the measurement parameters of each point in the original point cloud data, and input the original point cloud data and its measurement parameters of each point into the point cloud optimization model to obtain the corresponding optimized point cloud data;

[0022] Extract the point cloud features of the mine slope according to the optimized point cloud data of each point, fuse the optimized point cloud data of each item to obtain the optimized point cloud dataset of the mine slope, and use the grid generation algorithm to convert the optimized point cloud dataset into a continuous three-dimensional model.

[0023] Further, the process of obtaining the real-time image data of the mine slope, generating the real scene model in combination with the three-dimensional model, and obtaining the slope parameters of the mine slope in the real scene model includes:

[0024] Use a drone to obtain the real-time image data of the mine slope. The real-time image data includes the pixel values of each point on the mine slope, and map the pixel values of each point to the corresponding positions in the three-dimensional model to generate the real scene model of the mine slope;

[0025] In the real scene model, use the geometric features of the slope points, and based on the division principle of domain similarity, divide the slope points into different regions, use the region growing method to perform hierarchical slope segmentation on the mine slope to obtain different steps, and obtain the step top line and step bottom line of different steps. The slope parameters include the step top line and step bottom line of different steps on the mine slope.

[0026] Further, the process of obtaining the step slope angle and step height of the mine slope in combination with the measurement parameters includes:

[0027] Obtain the measurement parameters of a single point on the step bottom line, use the line corresponding to the measurement distance as the measurement line, use the plane where the point is located on the mine slope as the measurement plane, and use the angle between the measurement line and the measurement plane as the measurement plane angle according to the measurement angle;

[0028] Obtain the horizontal line angle between the measurement line and the horizontal line, obtain the step slope angle of the point according to the measurement plane angle and the horizontal line angle, record the shortest distance from the point to the step top line as the slope distance, obtain the corresponding step height according to the slope distance and the step slope angle, and adopt this method to obtain the step height of each point on the step bottom line and the step slope angles and step heights of different steps on the mine slope.

[0029] Further, the process of constructing a slope evaluation model and obtaining the slope height of the mine slope by using the slope evaluation model includes:

[0030] Generate a slope evaluation set according to the measurement plane angle, horizontal line angle, slope distance and corresponding step height of different points on the step bottom line, and divide it into an evaluation training set and an evaluation test set;

[0031] Construct a convolutional neural network. Take the included angle between the measurement plane and the horizontal line, the included angle between the horizontal line and the slope surface, the slope distance, and their corresponding bench heights at different points in the evaluation training set as the input data and output data of the convolutional neural network respectively, and train the convolutional neural network to obtain a second initial convolutional neural network.

[0032] Use the evaluation test set to verify the model of the second initial convolutional neural network, and output the initial convolutional neural network with an error less than or equal to the preset second test error threshold as the slope evaluation model.

[0033] Input the included angle between the measurement plane and the horizontal line, the included angle between the horizontal line and the slope surface, and the slope distance at each point on the mine slope into the slope evaluation model to obtain the corresponding slope height, where the slope height refers to the minimum height difference between each point on the mine slope and the bench slope top line.

[0034] Furthermore, set a height threshold for the mine slope. The process of obtaining the over-height area in combination with the slope height and performing visualization processing on the over-height area includes:

[0035] Set corresponding height thresholds for different mine slopes, obtain the actual height of each point on the mine slope. The actual height is equal to the sum of the bench heights of the bench where each point is located and the following benches, minus the slope height of each point, and compare the actual height of each point with the height threshold respectively.

[0036] Take the points with an actual height greater than the height threshold as over-height points, and take the area composed of all over-height points on the mine slope as the over-height area. In the real scene model, highlight the over-height area to achieve visualization processing, generate an over-height signal and feedback it to relevant personnel.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] 1. By obtaining the inclined point cloud data of the measurement points under different measurement parameters and constructing the corresponding point cloud optimization model, the present invention can optimize the original point cloud data collected by the unmanned aerial vehicle, and then construct the corresponding three-dimensional model, which can effectively reduce the influence degree of different measurement parameters on the point cloud data and is beneficial to improving the accuracy of the three-dimensional model.

[0039] 2. By generating a real scene model based on the three-dimensional model, obtaining the slope parameters of the mine slope and the slope height of each point on the mine slope in the real scene model, and then obtaining the actual height of each point, and comparing it with the preset height threshold to obtain the over-height area, the present invention can effectively reduce the deviation in height recognition through image analysis, accurately identify the height to each point on the mine slope, and can timely identify and feedback whether the mine slope is over-height. Description of the Drawings

[0040] Figure 1 is the flow chart of the present invention;

[0041] Figure 2 is the schematic diagram of the cross-section of the mine slope. Specific embodiments

[0042] As Figure 1 shown, an intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology includes the following steps:

[0043] Step S1: Set a number of measurement points on the mine slope, use the UAV to obtain the vertical point cloud data and inclined point cloud data of each measurement point, obtain the measurement parameters of each item of inclined point cloud data, and construct a point cloud optimization model in combination with the corresponding vertical point cloud data;

[0044] Step S2: Obtain the original point cloud data of the mine slope and its measurement parameters, use the point cloud optimization model to obtain the optimized point cloud data, and construct a corresponding three-dimensional model;

[0045] Step S3: Obtain the real-time image data of the mine slope, generate a real scene model in combination with the three-dimensional model, obtain the slope parameters of the mine slope in the real scene model, and obtain the bench slope angle and bench height of the mine slope in combination with the measurement parameters;

[0046] Step S4: Construct a slope evaluation model, and use the slope evaluation model to obtain the slope height of the mine slope;

[0047] Step S5: Set a height threshold for the mine slope, obtain the ultra-high area in combination with the slope height, and perform visualization processing on the ultra-high area.

[0048] It should be further noted that in the specific implementation process, the process of setting a number of measurement points on the mine slope and using the UAV to obtain the vertical point cloud data and inclined point cloud data of each measurement point includes:

[0049] Randomly set a number of measurement points on a single mine slope and different mine slopes. Taking each measurement point as a reference, use the UAV to obtain the vertical point cloud data and inclined point cloud data of each measurement point respectively;

[0050] Taking any measurement point as an example, obtain the normal line of the plane where the measurement point is located on the mine slope, that is, the measurement normal line. Treat the UAV as a point. When the UAV is on the measurement normal line, obtain the straight-line distance between the UAV and the measurement point, that is, the measurement distance;

[0051] Take the angle (acute angle) between the connection line between the UAV and the measurement point and the measurement normal line as the measurement angle. When the UAV is on the measurement normal line, its measurement angle is zero. Record the point cloud data of the measurement point obtained at this time as the vertical point cloud data;

[0052] Move the drone outside the measurement normal. Denote the point cloud data of the measurement points obtained at this time as the inclined point cloud data, and obtain the corresponding measurement distance and measurement angle. Based on the measurement points and the measurement normal, register the vertical point cloud data and the inclined point cloud data to the same coordinate system, and bind each item of point cloud data with its corresponding measurement distance and measurement angle.

[0053] It should be further noted that in the specific implementation process, the process of obtaining the measurement parameters of each item of inclined point cloud data and constructing a point cloud optimization model in combination with the corresponding vertical point cloud data includes:

[0054] For a single measurement point, continuously move the drone to obtain multiple items of inclined point cloud data and their corresponding measurement distances and measurement angles. Take the two as measurement parameters, and use the same method to separately obtain the vertical point cloud data and multiple items of inclined point cloud data and their corresponding measurement parameters of each measurement point;

[0055] Generate a point cloud optimization set according to the inclined point cloud data, measurement parameters and corresponding vertical point cloud data of different measurement points, and divide the obtained point cloud optimization set into an optimization training set and an optimization test set;

[0056] Construct a convolutional neural network. Use the different inclined point cloud data and measurement parameters in the optimization training set as the input data of the convolutional neural network, and use the corresponding vertical point cloud data in the optimization training set as the output data of the convolutional neural network. Train the convolutional neural network to obtain the first initial convolutional neural network;

[0057] Use the optimization test set to verify the model of the first initial convolutional neural network, and output the first initial convolutional neural network with an error less than or equal to the preset first test error threshold as the corresponding point cloud optimization model.

[0058] It should be further noted that in the specific implementation process, the process of obtaining the original point cloud data and its measurement parameters of the mine slope, using the point cloud optimization model to obtain the optimized point cloud data, and constructing the corresponding three-dimensional model includes:

[0059] In the actual application scenario, use the drone to collect the point cloud data of the entire mine slope. Denote the collected point cloud data as the original point cloud data of the mine slope. Take the corresponding point of the normal line where the drone is located when collecting the mine slope as the actual measurement point, and obtain the corresponding actual measurement normal line;

[0060] The original point cloud data includes the point cloud data of the actual measurement points and the point cloud data of other points except the actual measurement points, and obtain the measurement parameters corresponding to each point, including the measurement distance and the measurement angle. Input the original point cloud data and its measurement parameters of each point into the point cloud optimization model, and output the corresponding optimized point cloud data through the point cloud optimization model;

[0061] The number of actual measurement points is variable and determined by the size of the mine slope. The optimized point cloud data under different actual measurement points can be integrated into a unified coordinate system and registered using a registration algorithm. Taking the optimized point cloud data under a single actual measurement point as an example;

[0062] Extract the point cloud features of the mine slope based on the optimized point cloud data of each point. The point cloud features include corner points, edges, and surface normals. Fuse the optimized point cloud data after the above processing to obtain an optimized point cloud dataset of the mine slope, and use a mesh generation algorithm to convert the optimized point cloud dataset into a continuous mesh model, that is, a three-dimensional model.

[0063] It should be further noted that in the specific implementation process, the process of obtaining the real-time image data of the mine slope and generating a real scene model in combination with the three-dimensional model, and obtaining the slope parameters of the mine slope in the real scene model includes:

[0064] Taking any mine slope as an example, use a drone to obtain the real-time image data of the mine slope. The real-time image data contains the pixel values of each point on the mine slope. Generate a real scene model of the mine slope by mapping the pixel values of each point to the corresponding positions in the three-dimensional model;

[0065] In the real scene model, use the geometric features of the slope points (such as slope, aspect, elevation, etc.) for division. The slope points refer to specific positions on the terrain surface, and their geometric features are used to describe the terrain undulation and morphological changes. Based on the division principle of domain similarity, divide the slope points with similar geometric features into the same category and area;

[0066] Use the region growing method to perform hierarchical slope segmentation on the mine slope to obtain different steps. The region growing method is an image segmentation technology that aggregates pixel points according to the similar properties of pixels within the same region. Use the concave hull algorithm to extract the slope contour, and fit the step line based on the Bezier curve to obtain the corresponding step top line and step bottom line. The slope parameters include the step top line and step bottom line of different steps on the mine slope.

[0067] It should be further noted that in the specific implementation process, the process of obtaining the bench slope angle and bench height of the mine slope in combination with the measurement parameters includes:

[0068] Taking any point on the bench bottom line as an example, obtain the measurement parameters of this point. Take the line corresponding to the measured distance as the measurement line, and take the plane where this point is located on the mine slope as the measurement plane. According to the measurement angle, the included angle between the measurement line and the measurement plane can be obtained, denoted as the measurement plane included angle;

[0069] Obtain the angle between the measurement line and the horizontal line, denoted as the horizontal line angle. Obtain the bench slope angle of this point based on the measurement plane angle and the horizontal line angle. The bench slope angle, the measurement plane angle, and the horizontal line angle are in the same plane, and their sum is 180 degrees;

[0070] Obtain the shortest distance from this point to the bench crest line, denoted as the slope distance. Obtain the corresponding bench height based on the slope distance and the bench slope angle. Use the same method to obtain the bench heights of the points on the bench bottom line of this bench and the bench slope angles and bench heights of different benches on the mine slope.

[0071] It should be further noted that in the specific implementation process, the process of constructing a slope evaluation model and using the slope evaluation model to obtain the slope height of the mine slope includes:

[0072] Generate a slope evaluation set based on the measurement plane angles, horizontal line angles, slope distances, and their corresponding bench heights of different points on the bench bottom line. Divide the obtained slope evaluation set into an evaluation training set and an evaluation test set;

[0073] Construct a convolutional neural network. Use the measurement plane angles, horizontal line angles, and slope distances of different points in the evaluation training set as the input data of the convolutional neural network, and use the corresponding bench heights in the evaluation training set as the output data of the convolutional neural network. Train the convolutional neural network to obtain a second initial convolutional neural network;

[0074] Use the evaluation test set to verify the model of the second initial convolutional neural network, and output the initial convolutional neural network with an error less than or equal to the preset second test error threshold as the corresponding slope evaluation model;

[0075] Input the measurement plane angle, horizontal line angle, and slope distance of any point on the mine slope into the slope evaluation model to generate the corresponding slope height. The slope height refers to the minimum height difference between any point on the mine slope and the bench crest line;

[0076] The difference from the bench height is that the bench height is for the points on the bench bottom line and represents the minimum height difference between the corresponding points on the bench bottom line and the bench crest line. The slope height is for any point on the mine slope and represents the minimum height difference between the corresponding point and the bench crest line.

[0077] It should be further noted that in the specific implementation process, set a height threshold for the mine slope. The process of combining the slope height to obtain the ultra-high area and performing visualization processing on the ultra-high area includes:

[0078] Set corresponding height thresholds for different mine slopes, obtain the actual height of each point on the mine slope. The actual height is equal to the sum of the bench heights of the bench where each point is located and the benches below, minus the slope height of each point. Compare the actual height of each point with the height threshold respectively. If the actual height is greater than the height threshold, mark this point as a high point. If the actual height is less than or equal to the height threshold, do not perform any other operations on it;

[0079] Mark the area composed of all high points on the mine slope as a high area, highlight the high area in the real - scene model for visualization processing, and generate a corresponding high signal, and feedback the high signal to relevant personnel.

[0080] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology, characterized in that, It includes the following steps: Step S1: Set a number of measurement points on the mine slope, use a drone to obtain the vertical point cloud data and inclined point cloud data of each measurement point, obtain the measurement parameters of each item of inclined point cloud data, and construct a point cloud optimization model in combination with the corresponding vertical point cloud data; Step S2: Obtain the original point cloud data and its measurement parameters of the mine slope, use the point cloud optimization model to obtain the optimized point cloud data, and construct a corresponding 3D model; Step S3: Obtain the real-time image data of the mine slope, generate a real scene model in combination with the 3D model, obtain the slope parameters of the mine slope in the real scene model, and obtain the bench slope angle and bench height of the mine slope in combination with the measurement parameters; Step S4: Construct a slope evaluation model, and use the slope evaluation model to obtain the slope height of the mine slope; Step S5: Set a height threshold for the mine slope, obtain the ultra-high area in combination with the slope height, and perform visualization processing on the ultra-high area; Among them, the process of obtaining the vertical point cloud data and the inclined point cloud data includes: Randomly set a number of measurement points on a single mine slope and different mine slopes. Taking each measurement point as a reference, obtain the measurement normal of the plane where the measurement point is located on the mine slope. Treat the drone as a point for processing, and obtain the measurement distance between the drone and the measurement point; Take the angle between the connection line of the drone and the measurement point and the measurement normal as the measurement angle. Take the point cloud data of the measurement point obtained when the drone is on the measurement normal as the vertical point cloud data, and take the point cloud data of the measurement point obtained when the drone moves outside the measurement normal as the inclined point cloud data; The process of obtaining the measurement parameters and constructing the point cloud optimization model includes: For a single measurement point, continuously move the drone to obtain multiple items of inclined point cloud data and their corresponding measurement distances and measurement angles, and use the two as measurement parameters; Generate a point cloud optimization set according to the inclined point cloud data, measurement parameters and their corresponding vertical point cloud data of different measurement points, and divide it into an optimization training set and an optimization test set; Construct a convolutional neural network. Take the different inclined point cloud data, measurement parameters and their corresponding vertical point cloud data in the optimization training set as the input data and output data of the convolutional neural network respectively, and train the convolutional neural network to obtain a first initial convolutional neural network; Use the optimization test set to verify the model of the first initial convolutional neural network, and output the first initial convolutional neural network whose error is less than or equal to the preset first test error threshold as the point cloud optimization model.

2. The intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology according to claim 1, characterized in that, The process of obtaining the optimized point cloud data and constructing the 3D model includes: In the actual application scenario, use a drone to collect the original point cloud data of each point on the entire mine slope. Take the corresponding point of the normal where the drone is located on the mine slope during the collection of the mine slope as the actual measurement point, and obtain the actual measurement normal; Obtain the measurement parameters of each point in the original point cloud data, and input the original point cloud data and its measurement parameters of each point into the point cloud optimization model to obtain the corresponding optimized point cloud data; Extract the point cloud features of the mine slope based on the optimized point cloud data of each point, fuse the optimized point cloud data to obtain the optimized point cloud dataset of the mine slope, and use the grid generation algorithm to convert the optimized point cloud dataset into a continuous three-dimensional model.

3. The intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology according to claim 2, wherein, The process of generating a real scene model and obtaining slope parameters includes: Use a drone to obtain the real-time image data of the mine slope. The real-time image data contains the pixel values of each point on the mine slope. Map the pixel values of each point to the corresponding positions in the three-dimensional model to generate the real scene model of the mine slope; In the real scene model, use the geometric features of the slope points and based on the division principle of domain similarity, divide the slope points into different regions, use the region growing method to perform hierarchical slope segmentation on the mine slope to obtain different steps, and obtain the step top line and step bottom line of different steps. The slope parameters include the step top line and step bottom line of different steps on the mine slope.

4. An intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology according to claim 3, characterized in that, The process of obtaining the step slope angle and step height includes: Obtain the measurement parameters of a single point on the step bottom line. Take the line corresponding to the measured distance as the measurement line, and take the plane where the point is located on the mine slope as the measurement plane. Take the angle between the measurement line and the measurement plane as the measurement plane angle according to the measurement angle; Obtain the horizontal line angle between the measurement line and the horizontal line. Obtain the step slope angle of the point according to the measurement plane angle and the horizontal line angle. Denote the shortest distance between the point and the step top line as the slope distance. Obtain the corresponding step height according to the slope distance and the step slope angle. Adopt this method to obtain the step height of each point on the step bottom line and the step slope angles and step heights of different steps on the mine slope.

5. The intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology according to claim 4, characterized in that, The process of constructing a slope evaluation model and obtaining the slope height includes: Generate a slope evaluation set according to the measurement plane angle, horizontal line angle, slope distance and the corresponding step height of different points on the step bottom line, and divide it into an evaluation training set and an evaluation test set; Construct a convolutional neural network. Take the measurement plane angle, horizontal line angle, slope distance and the corresponding step height of different points in the evaluation training set as the input data and output data of the convolutional neural network respectively, and train the convolutional neural network to obtain the second initial convolutional neural network; Use the evaluation test set to verify the model of the second initial convolutional neural network, and output the initial convolutional neural network with an error less than or equal to the preset second test error threshold as the slope evaluation model; Input the measurement plane angle, horizontal line angle, and slope distance of each point on the mine slope into the slope evaluation model to obtain the corresponding slope height. The slope height refers to the minimum height difference between each point on the mine slope and the step top line.

6. The intelligent identification method for hidden dangers of open-pit mine slopes based on UAV technology according to claim 5, characterized in that, The process of obtaining the ultra-high area and performing visualization processing on it includes: Set corresponding height thresholds for different mine slopes, obtain the actual height of each point on the mine slope. The actual height is equal to the sum of the step heights of the steps where the point is located and the steps below, minus the slope height of the point. Compare the actual height of each point with the height threshold respectively; Points with an actual height greater than the height threshold are regarded as ultra-high points, and the area formed by all ultra-high points on the mine slope is regarded as the ultra-high area. In the real-scene model, the ultra-high area is highlighted for visualization processing, an ultra-high signal is generated and fed back to relevant personnel.

Images