Intelligent early warning method for slope landslide

Abstract

The application discloses an intelligent early warning method for slope landslide, and comprises the following steps: a. obtaining high-definition open slope orthographic images and terrain raster data; b. installing a slope radar on the opposite side of the open slope; c. installing a first intelligent cruise camera; d. increasing the number of cameras, and synthesizing panoramic photos of the early warning area by using photos taken by i cameras; e. increasing the number of cameras according to the monitoring period of the slope radar; f. determining the three-dimensional coordinate range and the central three-dimensional coordinate of the covered area of each photo; g. screening historical and real-time photos corresponding to the perspective of the early warning area, screening interference factors in the area based on an image artificial intelligence algorithm, and realizing intelligent early warning. The application establishes a mapping relationship between the camera perspective and the three-dimensional coordinates of the mine slope based on computer vision, verifies the early warning information quickly by using an artificial intelligence image processing algorithm, greatly improves the work efficiency, and reduces the labor intensity of personnel.

Description

TECHNICAL FIELD

[0001] The present application relates to a kind of intelligent early warning method for open-pit mine slope landslide using slope radar and camera, belong to mining technical field. BACKGROUND

[0002] With the gradual improvement of the situation of safety production in China, people's understanding and expectation of safety are also higher and higher, especially with the sustained and rapid development of mining industry in recent years, the number of mine employees is also considerable, and the state attaches great importance to mine safety production, and open-pit mine slope landslide monitoring and early warning technology has developed rapidly. There are signs before slope landslide, including slope cracking, rock rolling, slope extrusion deformation and other characteristics. At present, slope radar is generally used in domestic major mines to monitor these characteristics, and slope radar can monitor the slope monitoring area in real time, with full-time, all-weather, long-distance, large-scale, continuous spatial coverage, non-contact, surface and sub-millimeter level. The basic principle of slope radar is based on ground-based synthetic aperture radar differential interferometric measurement technology (DInSAR). Through ground-based synthetic aperture radar technology, high resolution is realized in distance direction by pulse compression, and high resolution is realized in azimuth direction by beam sharpening, so as to obtain two-dimensional high-resolution image of observation area; through differential interferometric measurement technology, the sequence of two-dimensional high-resolution images obtained at different times in the same target area is combined, and the high-precision deformation information of the measured area is obtained by using the phase difference of each pixel point in the image. Automatic monitoring is realized by using network remote control system, and when the slope deformation, deformation rate and deformation acceleration reach the set early warning threshold level, disaster warning will be given. However, due to the existence of many interference factors in the current open pit, such as mine production equipment and personnel, ground stress release and rock relaxation caused by rock mass excavation, instantaneous loosening and deformation of rock mass caused by blasting impact and vibration, rock mass change caused by drilling, production and transportation operation in mine pit, and radar system error, these interference factors cause high false alarm rate when radar early warning is carried out according to the monitoring and early warning threshold, and the radar itself does not have the function of checking whether the early warning information is accurate or not. Usually, technical personnel need to patrol the radar early warning area one by one to verify the early warning situation, so there are problems of low work efficiency and large verification workload, and a large number of early warning false information brings heavy burden to production units and technical personnel. SUMMARY

[0003] The purpose of the present application is to overcome the drawbacks of the prior art and provide an intelligent early warning method for slope landslide to quickly verify the early warning information, improve work efficiency and reduce the labor intensity of verification personnel.

[0004] To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0005] An intelligent early warning method for slope landslide, the method comprises the following steps:

[0006] a. Obtain high-definition open slope orthographic images and terrain raster data through unmanned aerial oblique photogrammetry, and one-to-one correspondence between any pixel of the slope orthographic image and the terrain raster data;

[0007] b. Install a slope radar on the opposite side of the open slope, so that the monitoring area of the slope radar corresponds to the shooting range of the obtained high-definition open slope orthographic image;

[0008] c. Install the first intelligent cruise camera:

[0009] Install the first intelligent cruise camera on the opposite side of the open slope, obtain 360° view intelligent cruise shooting photos of the open slope, use n to represent the number of photos, each photo corresponds to a view angle, manually select the view angles covering the radar monitoring area from n view angles, assume m1, and sequentially number the m1 view angles;

[0010] d. When the first camera cannot cover the early warning area, increase the number of cameras, and use the photos taken by i cameras to synthesize the panoramic photo of the early warning area:

[0011] Use the photos of m1 view angles to synthesize the panoramic photo of the early warning area, check whether there is an uncovered area, if there is an uncovered area, install the second intelligent cruise camera on the opposite side of the uncovered area, and repeat step c for the camera to obtain photos of m2 view angles; repeat the above process until the ith camera is installed, and a complete slope panoramic photo is synthesized by using photos of m1+m2+…+m i view angles, then step d stops;

[0012] e. Increase the number of cameras according to the slope radar monitoring period:

[0013] Use j to represent the slope radar monitoring period, record the time used by the first to ith camera to cruise all view angles, assume that the time used by the kth camera is T minutes, k=1,2, …, i, when T>j, supplement additional cameras at the kth camera, the additional camera and the kth camera each cruise half of the m k photos, until the time T used by each camera is less than or equal to j, stop supplementing additional cameras, use p to represent the total number of cameras after supplementing additional cameras, and finally synthesize a complete slope panoramic photo by using photos of m1+m2+…+m p view angles;

[0014] f. Determine the three-dimensional coordinate range and central three-dimensional coordinates of the covered area of each photo:

[0015] The m1+m2+...+m p The photos of the m1+m2+...+m

[0016] g. According to the preliminary determination of the warning area of the slope radar online monitoring, the historical and real-time photos corresponding to the warning area are screened, and the image artificial intelligence algorithm is used to screen the interference factors in the area, and if the interference factors exist, it is a false alarm; if the interference factors do not exist, the standby camera is called to monitor the slope situation of the area in real time, and the intelligent early warning is realized.

[0017] The intelligent early warning method of the above-mentioned slope landslide, the interference factors include mine production equipment and personnel, ground stress release and rock mass relaxation caused by rock mass excavation, instantaneous loosening and deformation of rock mass caused by impact and vibration effect of blasting, rock mass changes caused by drilling, production, loading and transportation operations in the mine stope.

[0018] The present application establishes the mapping relationship between the camera perspective and the three-dimensional coordinates of the mine slope based on computer vision, automatically selects the real-time and historical high-definition numerical images of the radar warning area, uses artificial intelligence image processing algorithm to quickly verify the warning information, and eliminates the false alarm caused by interference factors, greatly improves the work efficiency, and reduces the labor intensity of personnel. BRIEF DESCRIPTION OF DRAWINGS

[0019] The present application will be further described in detail below in combination with the drawings and specific embodiments.

[0020] Figure 1 is the first camera intelligent cruise shooting schematic diagram;

[0021] Figure 2 is the flow chart of determining the camera position and quantity according to the coverage range;

[0022] Figure 3 is the flow chart of increasing the number of cameras according to the slope radar cruise cycle;

[0023] Figure 4 is the flow chart of determining the three-dimensional coordinate range and the center three-dimensional coordinate of each picture coverage area;

[0024] Figure 5 is the flow chart of the present application. DETAILED DESCRIPTION

[0025] The present application aims at the problems of high false alarm rate of slope radar single means monitoring and early warning in current open pit, low efficiency of artificial checking and provides a slope landslide intelligent early warning method combining slope radar with camera, which greatly improves the monitoring and early warning accuracy of slope radar, replaces most of the artificial checking work and improves the work efficiency, solving the problems of high false alarm rate of slope radar single means monitoring and early warning in current open pit and low efficiency of artificial checking.

[0026] In the daily monitoring and early warning process of open pit slope radar, due to the interference of complex process links such as drilling, blasting, production, dressing and transportation in the mine stope and radar system error, etc., the radar early warning false alarm rate is high according to the monitoring and early warning threshold, the present application realizes real-time data acquisition of high-definition image covering the whole area of mine radar monitoring based on the calibration of image and mine three-dimensional coordinates and the automatic cruise shooting of camera, through the combination of slope radar and multiple high-definition cameras, the image change of radar early warning area is judged in real time by using the artificial intelligence algorithm of digital image, the early warning false alarm information caused by a large number of interference factors such as daily transportation and blasting in mine is intelligently investigated, so as to improve the early warning accuracy and greatly improve the work efficiency.

[0027] The specific steps of the present application are as follows:

[0028] Step one: obtain high-definition open pit slope orthographic image and terrain raster data by unmanned aerial oblique photogrammetry, at this time, the slope orthographic image is one-to-one corresponding to the terrain raster data;

[0029] Step two: install slope radar on the opposite side of the open pit slope, so that the monitoring area of the slope radar corresponds to the shooting range of the obtained high-definition open pit slope orthographic image;

[0030] Step three: install the first high-definition telephoto intelligent cruise camera (which can be used in part of the mine that has been installed) on the opposite side of the open pit slope (which can be near the radar room) to obtain 360° view intelligent cruise shooting photos, use n to represent the number of photos, each photo corresponds to a view angle, manually select the view angles covering the radar monitoring area in n view angles, assume m1, number the m1 view angles in order, as shown in the formula: Figure 1

[0031] Step four: use the photos of m1 view angles to synthesize the panoramic photo of the early warning area, check whether there is an uncovered area, if there is an uncovered area, install the second intelligent cruise camera on the opposite side of the uncovered area, and repeat step c for the camera to obtain the photos of m2 view angles; repeat the above process until the i th camera is installed, a total of m1+m2+...+m i view angle photos are used to synthesize the complete slope, then step four stops; ​

[0032] Step 5: Assuming the slope radar monitoring cycle is j minutes, record the time taken by the intelligent cruise of cameras 1 to i for all their perspectives. Assuming the time taken by camera k is T minutes, k = 1, 2, ..., i, when T > j, an additional camera needs to be added at camera k. The additional camera and camera k each cruise for m minutes. k The photo is taken from half the viewpoint. Adding additional cameras stops when the time T used by each camera is ≤ j. Assume that after adding additional cameras, the total number of cameras is p. Finally, using m1 + m2 + ... + m... p Photos from multiple perspectives were combined to create a complete panoramic view of the slope, thereby ensuring that the intelligent cruise cycle is consistent with the radar monitoring cycle.

[0033] Step 6: Calculate m1 + m2 + ... + m p Photos from each perspective were compared with high-definition orthophotos of the open slope. A deep learning image matching algorithm was used for pixel matching, and combined with terrain raster data, to determine the three-dimensional coordinate range and center three-dimensional coordinates of the area covered by each photo. In the later stage, artificial intelligence algorithms were used to capture the coordinate changes before and after the images to check whether displacement had occurred, so as to verify the slope landslide warning situation.

[0034] Step 7: Based on the preliminary early warning area determined by the slope radar online monitoring, timely filter historical and real-time photos covering the area, and use image artificial intelligence algorithms to screen for interference factors such as personnel, equipment, and blasting excavation in the area (collect data sets of photos of mining production equipment, personnel, blasting, etc. in advance for training) to eliminate false alarms; for images without interference factors after screening, call the backup camera to monitor the slope condition of the area in real time to achieve intelligent early warning.

[0035] Unless otherwise specified, the terms used in this invention have the meanings commonly understood by those skilled in the art.

[0036] The embodiments described in this invention are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Those skilled in the art can make various other substitutions, changes and improvements within the scope of this invention. Therefore, this invention is not limited to the above embodiments, but is only defined by the claims.

Claims

1. A method for intelligent early warning of slope landslide, characterized in that, The method comprises the following steps: a. Obtain high-definition open-pit slope orthographic images and terrain raster data through unmanned aerial oblique photogrammetry, and one-to-one correspondence between any pixel of the slope orthographic images and the terrain raster data; b. Install a slope radar on the opposite side of the open-pit slope, so that the monitoring area of the slope radar corresponds to the shooting range of the obtained high-definition open-pit slope orthographic images; c. Install the first camera that can intelligently cruise: Install the first camera that can intelligently cruise on the opposite side of the open-pit slope, obtain 360°-angle intelligent cruise shooting photos of the open-pit slope, use n to represent the number of photos, each photo corresponds to an angle, manually select angles covering the radar monitoring area from n angles, assume m1, and sequentially number the m1 angles; d. When the first camera cannot cover the early warning area, increase the number of cameras, and use photos taken by i cameras to synthesize panoramic photos of the early warning area: The panorama photo of the early warning area is synthesized by using m1 photos of different perspectives, it is checked whether there is an uncovered area, if there is an uncovered area, a second intelligent cruise camera is installed on the slope of the uncovered area, and the step c is repeated for the second camera to obtain m2 photos of different perspectives; the above process is repeated until the ith camera is installed, a total of m1+m2+...+m i perspectives are used to synthesize a complete panorama photo of the slope, and the step d stops. e. Increase the number of cameras according to the monitoring period of the slope radar: Let j represent the slope radar monitoring cycle. Record the time taken by the intelligent cruise of cameras 1 through i for all their perspectives. Assume the time taken by the kth camera is... T Minutes, k=1,2,…,i, when… T When >j, an additional camera is added at the k-th camera location. This additional camera and the k-th camera each cruise m. k Half the perspective in the photo, up to the time used by each camera. T If the sum of the numbers is less than or equal to j, stop adding additional cameras. Let p represent the total number of cameras after adding additional cameras. Finally, use m1 + m2 + ... + m p Photos from multiple perspectives were combined to create a complete panoramic photo of the slope. f. Determine the three-dimensional coordinate range and central three-dimensional coordinates of the covered area of each photo: m1+ m2+...+ m p The photos of the m1+ m2+...+ m p perspective views are respectively matched with high-definition open-pit slope orthographic images by using a deep learning image matching algorithm, and three-dimensional coordinate ranges and central three-dimensional coordinates of the coverage areas of the photos are determined in combination with terrain grid data. g. According to the preliminary determination of the early warning area by the online monitoring of the slope radar, screen the historical and real-time photos corresponding to the angles of the early warning area, screen the interference factors in the area based on image artificial intelligence algorithm, if there are interference factors, it is a false alarm; if there are no interference factors, call the standby camera to monitor the slope condition of the area in real time, and realize intelligent early warning; The interference factors include mine production equipment and personnel, ground stress release and rock mass relaxation caused by rock mass excavation, instantaneous loosening and deformation of rock mass caused by impact and vibration effect of blasting, and rock mass changes caused by drilling, production, loading and transportation operations in the mine stope.

Images