A method for recognizing a stable state of a side slope
By using UAV photogrammetry and GPS RTK monitoring devices, combined with the iCSF and M3C2 methods, the cracks and deformation of the landslide can be monitored in real time. This solves the problems of mobility, time-consuming and labor-intensive, feedback lag and low accuracy of traditional measurement techniques in landslide identification, and realizes timely and accurate identification and monitoring of the stability state of the landslide.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-09-30
- Publication Date
- 2026-03-17
AI Technical Summary
Traditional measurement techniques suffer from poor mobility, time-consuming and labor-intensive processes, delayed feedback, and low accuracy in identifying the stability of landslides.
UAV photogrammetry was used to establish DOM and DSM models, which were combined with a 3D point cloud model. Deformation was monitored by iCSF filtering and M3C2 method. Combined with GPS RTK monitoring device and crack meter, the cracks and deformation of the landslide were monitored in real time to identify the landslide status.
It enables timely and accurate identification of the stability state of landslides, is applicable to complex terrain, provides all-weather 360-degree scanning and monitoring without blind spots, has strong adaptability, and can capture the development trend of cracks and deformation in a timely manner.
Smart Images
Figure CN115511850B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of geotechnical engineering technology, specifically relating to a method for identifying the stability state of a landslide. Background Technology
[0002] Landslide stability identification consists of two aspects: crack development and surface deformation. Generally, accurate judgment of landslide development stages requires support from a wealth of data, including deep deformation, surface deformation, and surface crack development. However, obtaining deep deformation data requires drilling and installing inclinometers, and the data is easily affected by the environment, while the surface reflection of landslide evolution stages is more intuitive.
[0003] Rainfall, strong earthquakes, and engineering disturbances often cause damage to mountains, resulting in numerous cracks of varying sizes on the slope. As damage to the rock mass gradually accumulates, cracks in different parts of the slope merge and connect, forming a failure boundary within the slope, thereby reducing slope stability and accelerating slope instability. Cracks also act as dominant channels for surface water infiltration during landslide formation, further accelerating the initiation of landslides. Therefore, the identification and monitoring of cracks in key locations of landslides is a crucial aspect of geological hazard investigation and monitoring.
[0004] Traditional measurement techniques for identifying the stability of the aforementioned landslides suffer from drawbacks such as limited mobility, time and labor consumption, delayed feedback, and low accuracy. Summary of the Invention
[0005] The purpose of this invention is to address the shortcomings of traditional measurement techniques in identifying the stability state of landslides, such as limitations in mobility, time and labor consumption, delayed feedback, and low accuracy associated with traditional imaging methods.
[0006] To address the aforementioned problems, this invention specifically relates to a method for identifying the stability state of a landslide, comprising the following steps:
[0007] Step 1: Conduct aerial photography of the landslide at fixed time intervals using UAV photogrammetry to establish a DOM model, a DSM model, and a 3D point cloud model. Extract surface slope information based on the DSM model and roughness information based on the 3D point cloud model. On this basis, identify the landslide boundary and identify the development and location distribution of cracks.
[0008] Step 2: Based on the 3D point cloud model obtained in Step 1, the point cloud data is filtered by the iCSF method with corrected gravity direction, ground points are separated and extracted, the relative position of the control points is checked to obtain the relative error of the monitoring, and the deformation monitoring of the multi-temporal point cloud model is carried out by the M3C2 (multi-scale model to model cloud comparison) method to obtain the visual displacement field of the landslide.
[0009] Step 3: Based on the crack location determined in Step 1, use a crack monitoring device to measure the crack development status, and periodically take and record readings of the crack monitoring device through manual / drone patrols.
[0010] Step 4: When the cracks develop to a certain stage and form a clear landslide boundary, set up monitoring points at the front and rear edges of the landslide to reflect the deformation characteristics of the landslide.
[0011] Step 5: Deploy the target prism at the monitoring point location and place the GPS receiver antenna. In addition, set up a reference point in the area outside the landslide area, and set up a GPS receiver and a GPS RTK base station receiver at the reference point. By regularly monitoring the reference point in the stable area, the deformation of the monitoring point can be inverted.
[0012] Step 6: Conduct dynamic and continuous monitoring of key locations (GPS RTK) and crack development (crack gauge) in the landslide area, including the front edge, rear edge, and middle of the landslide, to record and identify the crack development and deformation of the landslide in real time, and to make a preliminary identification of the landslide development stage.
[0013] Step 7: Combine the landslide crack development status obtained in Step 6 with the deformation status obtained in Step 2 to further identify the state of the landslide.
[0014] Furthermore, the specific method for identifying the boundary of the landslide and the development and location distribution of cracks in step one is as follows: compared with its local neighborhood, there is an abrupt change in slope at the steep embankment or crack of the landslide. The DSM slope map identifies the larger steep embankment, and the DOM model identifies the cracks developed near the main crack at the rear edge of the landslide.
[0015] Furthermore, the method for filtering the point cloud data in step two using the iCSF method with corrected gravity direction is as follows:
[0016] The CSF algorithm first inverts the 3D point cloud data, then covers the inverted point cloud with a simulated "cloth", and generates an approximate curved surface based on the position of the "cloth" covering the point cloud under the action of gravity; by comparing the distance between the points in the original point cloud data and the generated "cloth" surface, the ground points are identified from the original point cloud data, and the distance points are separated for further feature extraction.
[0017] According to Newton's second law, the relationship between the position of the simulated cloth and the forces is determined by the following formula:
[0018]
[0019] In the formula, X represents the position of the particle at time t; F ext (X,t) represents the external force, which consists of gravity and the collision force generated by the obstacle:
[0020] F ext (X,t)=mg+f interact (X,t) (2)
[0021] When a particle encounters objects in its direction of motion; F int (X,t) represents the internal force of a particle at position X and time t, which is generated by its interaction with the point cloud (boundary). Since both the internal and external forces change with time t, the above equation is solved by numerical integration (such as the Euler method) in the cloth simulation.
[0022] The above method is suitable for gentle surfaces, but not for landslide-prone terrain. Therefore, F ext The direction of gravity (X,t) is changed from the vertical direction in the traditional CSF method to a direction perpendicular to the slope; this direction is determined by the median slope of the DSM model; the steps are as follows:
[0023] (1) Obtain the median slope α of the region from the DSM model generated by UAV photography technology, and determine the normal vector [N] based on the median slope. x N y N z ];
[0024] (2) Modified form (2) The first term on the right side of the equation, the gravity term g′=[N x N y N z 9.81;
[0025] (3) Substitute the corrected gravity term back into equation (2) to start filtering.
[0026] Furthermore, the specific method of step three is as follows: fix a steel nail at each end of the crack to ensure the synchronization of the movement of the steel nail with the soil on both sides. Fix the initial scale end and the free end of the electronic vernier caliper to the two steel nails respectively. The development of the crack can drive the steel nail to move, thereby changing the reading of the electronic vernier caliper. The vernier caliper of the crack is periodically read and recorded by manual / drone inspection.
[0027] Furthermore, in step five, the reference point is set in a stable area 30m away from the landslide body.
[0028] Furthermore, the specific method for step six is as follows:
[0029] Assume that reference point P1 is located on the relatively stable side of the crack, and monitoring point P2 is located on the unstable side of the crack. First, the world coordinates (X, Y, F) of reference point P1 and monitoring point P2 are obtained by using RTK. 10 ,Y 10 Z 10Next, dynamic and continuous monitoring will be performed on the reference point P1 and the monitoring point P2 to obtain the world coordinates (X, Y, X) of the two points. 1t ,Y 1t Z 1t ), (X 2t ,Y 2t Z 2t By solving the problem, we can obtain the data on the slip length l, crack width d, and slip height h during crack development. Assume the angle between the line connecting points P1 and P2 and the x-axis is... The included angle during the dynamic monitoring phase is The solution formula is as follows:
[0030]
[0031]
[0032] h = Z 2t -Z 1t
[0033] Based on the data of slip length l, crack width d, and slip height h during the crack development process, the development stage of the side landslide is preliminarily identified.
[0034] Furthermore, the method for preliminary identification of the development stage of a landslide based on the data of slip length l, crack width d, and slip height h during crack development is as follows:
[0035] (1) For a side landslide without cracks, it is determined that it is in a stable state;
[0036] (2) For areas with a small number of cracks but no obvious landslide boundary, they are judged to be in a relatively stable state;
[0037] (3) When cracks continue to develop and form a clear landslide boundary, it is judged that the crack is in a relatively unstable state;
[0038] (4) For a side landslide with bulging at the leading edge and cracks and misalignment at the trailing edge, it is determined that it is already in an unstable state. When the rate of change of the three parameters, namely the slip length l, the crack width d, and the misalignment height h, increases significantly, an early warning is required.
[0039] Furthermore, the method for further identifying the state of the landslide by combining the landslide crack development status obtained in step six and the deformation status obtained in step two in step seven is as follows:
[0040] When there are no cracks and minimal deformation in the landslide area, the landslide area is considered to be in a stable state. When some cracks appear in the area but have not yet formed a boundary and the deformation rate does not exceed 10 mm / month, the landslide area is considered to be in a relatively stable state. When cracks in the landslide area continue to develop on the surface, forming a clear landslide boundary and the deformation rate exceeds 10 mm / month, the landslide area is considered to be in a relatively unstable state. When the cracks at the rear edge of the landslide area are misaligned, and the leading edge bulges and produces feather-like cracks with a deformation rate exceeding 50 mm / month, the area is considered to be in an unstable state.
[0041] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:
[0042] (1) The method for identifying the stability of landslides based on UAV technology of the present invention utilizes UAVs to achieve all-weather 360-degree scanning and monitoring of landslide morphology without blind spots, replacing manual labor.
[0043] (2) The method for identifying the stability of landslides based on UAV technology of the present invention can capture the development trend of cracks and deformation of landslides in a timely and accurate manner, and the results given comprehensively and objectively reflect the actual situation of the landslide status.
[0044] (3) The method for identifying the stability of landslides based on UAV technology of the present invention can monitor and evaluate other landslides, railway landslides, highway landslides and other complex terrain areas, and has strong adaptability. Attached Figure Description
[0045] Figure 1 This is a schematic diagram of the logic flow of a preferred embodiment of the present invention;
[0046] Figure 2 This invention provides a preferred embodiment of generating a landslide topographic slope map based on DSM images.
[0047] Figure 3 The preferred embodiment of the present invention is a DOM image based on the identification of landslide slopes and cracks;
[0048] Figure 4 A comparison of point clouds of a landslide on December 18, 2021 and December 23, 2021, for a preferred embodiment of the present invention;
[0049] Figure 5 This is a schematic diagram of a UAV crack monitoring system according to a preferred embodiment of the present invention;
[0050] Figure 6 This is a schematic diagram of deformation monitoring at key locations of a landslide according to a preferred embodiment of the present invention.
[0051] Figure 7 This is a photograph taken from an aerial perspective in a stable state, representing a preferred embodiment of the present invention.
[0052] Figure 8 This is a photograph taken from an aerial perspective, showing a relatively stable state, according to a preferred embodiment of the present invention.
[0053] Figure 9 This is a photograph taken from an aerial perspective, showing a relatively unstable state, according to a preferred embodiment of the present invention.
[0054] Figure 10 This is a photograph taken from an aerial perspective, showing an unstable state, according to a preferred embodiment of the present invention. Detailed Implementation
[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0056] Please refer to Figure 1 A method for identifying the stability state of a landslide includes the following steps:
[0057] Step 1: Conduct aerial photography of the landslide at fixed time intervals using UAV photogrammetry to establish a DOM model, a DSM model, and a 3D point cloud model. Extract surface slope information based on the DSM model and roughness information based on the 3D point cloud model. On this basis, identify the landslide boundary and identify the development and location distribution of cracks.
[0058] The specific method for identifying the boundary of the landslide and the development and location distribution of cracks in step one is as follows: the steep slope or crack of the landslide has an abrupt change in slope compared with its local neighborhood. The DSM slope map identifies the larger steep slope, and the DOM model identifies the cracks developed near the main crack at the rear edge of the landslide.
[0059] Figure 2 A slope map of a landslide boundary generated from DSM images is presented (slope represents the gradient in the image). The slope distribution in the study area ranges from 0° to 90°. The slope calculation results show that, compared to their local neighborhoods, cracks and steep slopes at the perimeter of active landslides generally have slopes exceeding 50°, which is related to abrupt changes in surface cracks and steep slopes. At steep slopes or cracks, there are abrupt changes in slope (greater than 50°) compared to their local neighborhoods. By combining DOM and DSM data to collaboratively identify trailing cracks and determine the landslide boundary, the DSM slope map can identify larger steep slopes, while the DOM model can identify small cracks near the main crack at the trailing edge of the landslide, arranged in approximately concentric arcs on the plane. The final landslide boundary identification results are shown below. Figure 3 As shown.
[0060] Step 2: Based on the 3D point cloud model obtained in Step 1, the point cloud data is filtered by the iCSF method with corrected gravity direction, ground points are separated and extracted, the relative position of the image control points is checked to obtain the relative error of the monitoring, and the deformation monitoring of the multi-temporal point cloud model is performed by the M3C2 method to obtain the visual displacement field of the landslide.
[0061] The iCSF method, which corrects for the direction of gravity, is used to filter point cloud data as follows:
[0062] The CSF algorithm first inverts the 3D point cloud data, then covers the inverted point cloud with a simulated "cloth", and generates an approximate curved surface based on the position of the "cloth" covering the point cloud under the action of gravity; by comparing the distance between the points in the original point cloud data and the generated "cloth" surface, the ground points are identified from the original point cloud data, and the distance points are separated for further feature extraction.
[0063] According to Newton's second law, the relationship between the position of the simulated cloth and the forces is determined by the following formula:
[0064]
[0065] In the formula, X represents the position of the particle at time t; F ext (X,t) represents the external force, which consists of gravity and the collision force generated by the obstacle:
[0066] F ext (X,t)=mg+f interact (X,t) (2)
[0067] When a particle encounters objects in its direction of motion; F int (X,t) represents the internal force of a particle at position X and time t, which is generated by its interaction with the point cloud (boundary). Since both internal and external forces change with time t, the above equation is usually solved by numerical integration (such as Euler's method) in cloth simulation implementation.
[0068] The above method is suitable for gentle surfaces but not for landslide-prone terrain. Therefore, this proposal suggests a modified CSF method, which uses F... ext The direction of gravity (X,t) is changed from the vertical direction in the traditional CSF method to a direction perpendicular to the slope; this direction is determined by the median slope of the DSM model; the steps are as follows:
[0069] (1) Obtain the median slope α of the region from the DSM model generated by UAV photography technology, and determine the normal vector [N] based on the median slope. x N y N z ];
[0070] (2) Modified form (2) The first term on the right side of the equation, the gravity term g′=[N x N y N z 9.81;
[0071] (3) Substitute the corrected gravity term back into equation (2) to start filtering.
[0072] For deformation monitoring Figure 4 A comparison of point cloud data from December 18, 2021, and December 23, 2021, for a landslide is presented. Point cloud-based landslide deformation monitoring provides a visualized view of material loss (collapse) or gain (accumulation) in the area of surface change. Under continuous blasting operations, a collapse occurred at the fault exposure site, leading to a landslide of the strongly weathered soil overlying the fault, ultimately accumulating into three fan-shaped accumulations (#1, #2, #3) on the bottom platform.
[0073] Step 3, please refer to Figures 5-10 Based on the crack locations determined in step one, a crack monitoring device (a combination of electronic vernier calipers and steel nails) is used to measure the crack development. The crack monitoring device is periodically read and recorded through manual / drone patrols. The specific method is as follows: a steel nail is fixed at each end of the crack to ensure the synchronization of the steel nail with the movement of the soil on both sides. The initial scale end and free end of the electronic vernier caliper are fixed to the two steel nails respectively. The development of the crack can drive the steel nail to move, thereby changing the reading of the electronic vernier caliper. The crack monitoring vernier caliper is periodically read and recorded through manual / drone patrols.
[0074] Step 4: When the cracks develop to a certain stage and form a clear landslide boundary, set up monitoring points at the front and rear edges of the landslide to reflect the deformation characteristics of the landslide.
[0075] Step 5: Deploy the target prism at the monitoring point location and place the GPS receiver antenna. In addition, set up a reference point in the area outside the landslide (the reference point in Step 5 is set up in the stable area 30m away from the landslide body). Set up the GPS receiver and GPS RTK base station receiver at the reference point. By regularly monitoring the reference point in the stable area, the deformation of the monitoring point can be inverted.
[0076] Step Six: Conduct dynamic and continuous monitoring of key locations (GPS RTK) and crack development (crack gauges) in the landslide area, including the leading edge, trailing edge, and middle section of the landslide. Record and identify the crack development and deformation status of the landslide in real time, and make a preliminary identification of the landslide development stage.
[0077] The specific method for step six is as follows:
[0078] Assume that reference point P1 is located on the relatively stable side of the crack, and monitoring point P2 is located on the unstable side of the crack. First, the world coordinates (X, Y, F) of reference point P1 and monitoring point P2 are obtained by using RTK. 10 ,Y 10 Z 10 Next, dynamic and continuous monitoring will be performed on the reference point P1 and the monitoring point P2 to obtain the world coordinates (X, Y, X) of the two points. 1t ,Y 1t Z 1t ), (X 2t ,Y 2t Z 2t By solving the problem, we can obtain the data on the slip length l, crack width d, and slip height h during crack development. Assume the angle between the line connecting points P1 and P2 and the x-axis is... The included angle during the dynamic monitoring phase is The solution formula is as follows:
[0079]
[0080]
[0081] h = Z 2t -Z 1t
[0082] Then, based on the data of slip length l, crack width d, and slip height h during crack development, the development stage of the side landslide is preliminarily identified. The method for preliminarily identifying the development stage of the side landslide based on the data of slip length l, crack width d, and slip height h during crack development is as follows:
[0083] (1) For a side landslide without cracks, it is determined that it is in a stable state;
[0084] (2) For areas with a small number of cracks but no obvious landslide boundary, they are judged to be in a relatively stable state;
[0085] (3) When cracks continue to develop and form a clear landslide boundary, it is judged that the crack is in a relatively unstable state;
[0086] (4) For a side landslide with bulging at the leading edge and cracks and misalignment at the trailing edge, it is determined that it is already in an unstable state. When the rate of change of the three parameters, namely the slip length l, the crack width d, and the misalignment height h, increases significantly, an early warning is required.
[0087] Step 7: Further identify the state of the landslide by combining the crack development data obtained in Step 6 and the deformation data obtained in Step 2; the specific method is as follows:
[0088] When there are no cracks and minimal deformation in the landslide area, the landslide area is considered to be in a stable state. When some cracks appear in the area but have not yet formed a boundary and the deformation rate does not exceed 10 mm / month, the landslide area is considered to be in a relatively stable state. When cracks in the landslide area continue to develop on the surface, forming a clear landslide boundary and the deformation rate exceeds 10 mm / month, the landslide area is considered to be in a relatively unstable state. When the cracks at the rear edge of the landslide area are misaligned, and the leading edge bulges and produces feather-like cracks with a deformation rate exceeding 50 mm / month, the area is considered to be in an unstable state.
[0089] In summary, the methods for identifying different stages of the stability state of a landslide are summarized as follows:
[0090] 1. Slope without cracks: On the one hand, regular joint inspections using manpower and drones are conducted to check for cracks in the landslide. On the other hand, for landslides with cracks, benchmark control stakes are set up in relatively stable areas, and static measurements are performed using RTK to obtain the coordinates of the control stakes. The landslide boundary is determined by the location of the cracks, and monitoring points are selected at the front, middle, and rear edges of the landslide. Dynamic measurements are performed using RTK to obtain the deformation of the monitoring points at the front, middle, and rear edges.
[0091] 2. Crack formation stage: A simple crack monitoring device is used, which involves fixing a steel nail at each end of the crack, and then fixing the initial scale end and free end of an electronic vernier caliper to the two steel nails respectively. As the crack develops, it causes the steel nails to move, thereby changing the reading of the electronic vernier caliper. On the other hand, the crack is located by using the slope change model of the DSM model and the DOM image model.
[0092] 3. Crack Development Stage: Periodically, a drone equipped with a zoom camera is used to photograph a vernier caliper installed at the crack location, and the electronic reading of the caliper is used to obtain the crack width. When the crack develops to a certain extent, a fault occurs at the rear edge of the landslide. The fault height and crack width can be directly obtained by calculating the coordinates of the RTK monitoring points. On the other hand, the crack development is quantitatively extracted using the DSM model of slope abrupt changes and the DOM image model to determine the approximate extent of the landslide.
[0093] 4. Front edge bulging and rear edge misalignment stage: When the crack develops to a certain extent, bulging occurs at the front edge of the slope and misalignment occurs at the rear edge of the slope. The horizontal displacement of the bulging and the height of the misalignment can be directly read by drone photography. In terms of deformation, the deformation is extracted by GPS RTK monitoring and multi-temporal point cloud model change detection.
[0094] 5. Identification of landslide development stages: By combining crack gauge readings obtained by UAVs, deformation of 3D point cloud models, and deformation trends of key locations of landslides obtained by RTK, a visualized trend of landslide deformation and failure is formed, allowing for a preliminary judgment on the development stage of the landslide.
[0095] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for identifying a stable state of a side slope, characterized by, The method comprises the following steps: Step one: taking photos of the side landslide at fixed time intervals by unmanned aerial photography, establishing a DOM model, a DSM model and a three-dimensional point cloud model, and extracting surface slope information based on the DSM model; on this basis, side landslide boundary identification is carried out, and crack development and position distribution are identified; Step two: based on the three-dimensional point cloud model obtained in step one, the point cloud data is filtered by the iCSF method of correcting the gravity direction, ground points are separated and extracted, the relative position of the image control point is checked to obtain the relative error of the monitoring, and the M3C2 method is used to monitor the deformation of the multi-time point cloud model to obtain the visual displacement field of the side landslide; Step three: based on the crack position determined in step one, the crack development status is measured by using a crack monitoring device, and the crack monitoring device is periodically read and recorded by manual / unmanned aerial vehicle patrol; Step four: when the crack development forms a clear landslide boundary, monitoring points are arranged at positions reflecting the deformation characteristics of the side landslide at the front and rear edges of the side landslide; Step five: target prisms are arranged at the monitoring point positions, and GPS receiver antennas are placed, and in addition, reference points are arranged outside the side landslide, and GPS receivers and GPS RTK reference station receivers are erected on the reference points, and the deformation of the crack is inversed by periodically monitoring the stable area reference points; Step six: the side landslide area, the front and rear edges of the side landslide, and the key positions of the monitoring points and the crack development are dynamically and continuously monitored, the crack development and deformation status of the side landslide are recorded and identified in real time, and the development stage of the side landslide is preliminarily identified; The specific method of step six is: The reference point P1 is arranged on the relatively stable side of the crack, and the monitoring point P2 is arranged on the unstable side of the crack; firstly, the RTK is used to position the reference point P1 and the monitoring point P2 to obtain the world coordinates of the points P1 and P2; , , ); next, the reference point P1 and the monitoring point P2 are continuously monitored to obtain the world coordinates of the two points , , ), , , ); the displacement length l, the crack width d and the fault height h data in the crack development process are obtained by solving, the angle between the line connecting the points P1 and P2 and the x axis is , the angle in the dynamic monitoring stage is , and the solving formula is as follows: ； ； ； Then, the development stage of the side landslide is preliminarily identified according to the data of the crack development length l, crack width d and fault height h; Step seven: the state of the side landslide is further identified by comprehensively considering the crack development status obtained in step six and the deformation status obtained in step two.
2. The method for identifying a stable state of a side slope landslide according to claim 1, wherein The specific method of step one for carrying out side landslide boundary identification and identifying crack development and position distribution is that the scarp or crack of the side landslide has a sudden change in slope compared with its local neighborhood, the DSM slope graph identifies the scarp, and the DOM model identifies the cracks developed near the main crack of the landslide rear edge.
3. The method for identifying a stable state of a side slope landslide according to claim 1, wherein The method of step two for filtering the point cloud data by the iCSF method of correcting the gravity direction is: The CSF algorithm first inverts the 3D point cloud data, then covers the inverted point cloud by simulating a "cloth", and generates an approximate surface according to the position of the "cloth" covering the point cloud under the action of gravity; by comparing the distance between the points in the original point cloud data and the "cloth" surface, the ground points are identified from the original point cloud data, and the ground points are separated for further feature extraction; The relationship between the cloth position and the force is determined according to Newton's second law: ； where X represents the position of the particle at time t; represents the external force, which consists of the gravitational force and the collision force due to the obstacle: ； When the particle encounters some object in its direction of motion; represents the internal force of the particle at position X and time t, the internal force resulting from the interaction with the boundary of the point cloud; since both the internal and external forces vary with time t, the above equation is solved by numerical integration in the implementation of cloth simulation; The above method is suitable for flat surface, and is not suitable for side landslide topographic features, so the The gravity direction is changed from the vertical direction of the traditional CSF method to the direction perpendicular to the slope surface; it is determined by the median value of the slope of the DSM model; the steps are as follows: (1) obtaining a median of regional slope based on a DSM model generated by a UAV photography technology , determining a normal vector according to the median slope ; (2) Replacing the first term, the gravity term, on the right side of the equation in (2) ; (3) Return the corrected gravity term to formula (2) to start filtering.
4. The method for identifying a stable state of a side slope landslide according to Claim 1, wherein The specific method of step three is: fixing a steel nail at each end of the crack, ensuring the synchronization of the steel nail and the movement of the soil on both sides, fixing the initial scale end and the free end of the electronic vernier caliper on the two steel nails respectively, and driving the steel nails to move with the development of the crack, thereby changing the reading of the electronic vernier caliper, and periodically reading the crack meter vernier caliper by manual / unmanned aerial vehicle patrol and recording.
5. The method for identifying a stable state of a side slope landslide according to Claim 1, wherein The reference point in step five is arranged in a stable area 30 m away from the side landslide body.
6. The method of claim 1, wherein The method for preliminarily identifying the development stage of the side landslide according to the dislocation length l, the crack width d and the dislocation height h data in the crack development process is: (1) For the side landslide without cracks, it is judged to be in a stable state; (2) For the side landslide with a small amount of cracks but without forming a clear side landslide boundary, it is judged to be in a relatively stable state; (3) When the cracks continue to develop and form a clear side landslide boundary, it is judged to be in a relatively unstable state; (4) For the side landslide with bulging front edge and dislocation rear edge, it is judged to be in an unstable state; when the change rates of the three parameters of dislocation length l, crack width d and dislocation height h significantly increase, early warning is needed.
7. The method for identifying the stable state of a side slope landslide according to any one of claims 1 to 6, characterized in that, The method for further identifying the state of the side landslide by comprehensively considering the crack development condition of the side landslide obtained in step six and the deformation condition obtained in step two in step seven is: When there is no crack in the side landslide area and the deformation is small, it is determined that the side landslide area is in a stable state; when part of the cracks are generated in the area, but the boundary is not formed and the deformation rate is not more than 10 mm / month, it is determined that the side landslide area is in a relatively stable state; when the cracks continue to develop in the side landslide area and form a clear landslide boundary on the surface, and the deformation rate is more than 10 mm / month, it is determined that it is in a relatively unstable state; when the cracks on the rear edge of the side landslide area dislocate on the upper and lower plates, the front edge bulges and appears a feather-shaped crack, and the deformation rate is more than 50 mm / month, it is determined that the area is in an unstable state.