Soft rock slope three-dimensional intelligent monitoring and disaster early-stage identification and early-warning method

Abstract

The invention provides a highway soft rock slope three-dimensional intelligent monitoring and disaster early-stage identification early-warning method, and belongs to the technical field of slope monitoring, and the method comprises the following steps: shooting a highway roadbed by using remote sensing equipment, obtaining corresponding remote sensing image data, and identifying a remote sensing image; the image is input into a machine model for feature comparison learning, hidden danger point screening and disaster body recognition are carried out, a soft rock area is sampled, a soft rock mass three-dimensional space model is established, data in the soft rock mass are collected in real time by arranging a collecting device, the collected data serve as dynamic data of the soft rock mass three-dimensional space model, and the dynamic data are stored in a storage device. Remote sensing data and real-time collected data are input into an on-line early warning system, whether instability occurs in a soft rock area or not is judged, early warning information is sent to an area where instability or landslide possibly occurs, and management personnel are notified to check on site and take protective measures.

Description

Technical Field

[0001] The present invention relates to the technical field of slope monitoring, and particularly to a three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes. Background Art

[0002] The geological structure in Guangxi region is complex and the environment is fragile. Affected by the mid-south subtropical monsoon climate, geological disasters such as landslides, collapses, and debris flows occur frequently. Usually, major hidden dangers of geological disasters have complex characteristics such as concealment, suddenness, and uncertainty, which greatly increases the difficulty of active prevention and monitoring and early warning. According to the results of geological disaster investigation and evaluation in 2018, 96.65% of the land area in Guangxi belongs to geological disaster-prone areas, among which the high and medium geological disaster-prone areas reach 42.3%. Carrying out highway construction and operation in these areas poses high challenges and engineering risks.

[0003] Traditional methods for preventing and controlling geological disasters mainly rely on regular on-site investigations by personnel in the early stage of identification. For high-risk points, manual monitoring means such as GPS, level, total station, close-range photogrammetry, and displacement meters are used. Although these monitoring technologies have achieved remarkable achievements in the research of geological disaster monitoring and prediction, and have also accumulated rich practical experience, which has led to a breakthrough in disaster early warning in our region, there are still some obvious deficiencies under the special conditions in our region:

[0004] (1) The manual investigation has a large workload, low efficiency, and only relies on the naked eye to identify early geological disasters; (2) The places where disasters such as landslides and debris flows occur are generally alpine and forested areas with poor geological environments, making it difficult for humans to reach; (3) Traditional measurement methods (level, total station, close-range photogrammetry) require relevant staff to reach the site to observe and record a large amount of on-site measurement data. Therefore, there will be problems such as a large calculation workload, low work efficiency, and high costs; (4) Affected by observation costs and other factors, traditional measurement technologies cannot obtain continuous surface deformation information in the monitoring area, and the obtained results are scattered, with low spatial resolution and difficult to achieve the early warning effect; (5) A complete geological disaster prevention and control system from geological disaster identification, monitoring to treatment has not been established. Therefore, it is necessary to design a three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes. Summary of the Invention

[0005] The purpose of the present invention is to provide a three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes to solve the technical problems mentioned in the background art.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] A three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes, the method comprising the following steps:

[0008] Step 1: Use remote sensing equipment to photograph the highway subgrade and obtain corresponding remote sensing image data;

[0009] Step 2: Identify the remote sensing images, input the images into the machine model for feature contrast learning, and conduct hidden danger point screening and disaster body identification;

[0010] Step 3: Sample the soft rock area and establish a three-dimensional space model of the soft rock mass;

[0011] Step 4: Real-time collect the data inside the soft rock mass through the set collection device, and the collected data serves as the dynamic data of the three-dimensional space model of the soft rock mass;

[0012] Step 5: Input the remote sensing data and the real-time collected data into the online early warning system to judge whether there is instability in the soft rock area;

[0013] Step 6: Send early warning information to the areas where instability or landslides may occur, and notify the management personnel to conduct on-site inspections and take protective measures.

[0014] Furthermore, the remote sensing equipment in Step 1 includes geological disaster general surveys based on spaceborne platforms and geological disaster detailed surveys based on airborne platforms. The specific process of the geological disaster general survey based on the spaceborne platform is as follows:

[0015] Through the satellite remote sensing platform, use the method of combining InSAR with optical remote sensing. Use optical remote sensing to extract information on landslide morphology and land cover classification, and InSAR to extract the deformation field. Combine the two to obtain factors such as surface deformation rate, land cover classification, ground inversion temperature, and vegetation cover, and construct a geological hazard knowledge map. Construct the characteristics and their correlation relationships of the three elements of the disaster-bearing environment, inducing factors, and change signs in landslide hazards. The disaster-bearing environment includes geological structure, hydrological conditions, surface undulation, and surface cover. The inducing factors include precipitation and soil freeze-thaw. The change signs include displacement and settlement. Analyze the characteristics and formation conditions of major geological disasters, screen and determine the key factors for the occurrence of geological disasters. Through the analysis and calculation of disaster-causing factors, use machine learning methods to establish an automatic geological disaster identification system to achieve automatic or semi-automatic extraction of geological disasters.

[0016] Furthermore, the specific process of detailed geological disaster investigation based on an airborne platform is as follows: An unmanned aerial vehicle (UAV) platform is used, with a laser scanner and a high-resolution optical camera set on the UAV platform to obtain ground three-dimensional coordinates and image data, and to reproduce the three-dimensional real-time changes and true morphological characteristics of ground objects. Calculate the vertical displacement of the ground surface, volume changes, and profiles before and after changes. Filter the point cloud data of steep slopes and complex terrain with dense vegetation. Obtain the digital surface model after removing vegetation through LiDAR data, and combine two-dimensional and three-dimensional interpretation to identify potential hidden hazards of geological disasters such as collapsed loose accumulation bodies, ancient landslide accumulation bodies, debris flow accumulation fans, and earthquake-cracked mountain fissures under vegetation cover, so as to achieve a detailed investigation of the geological disaster area.

[0017] Furthermore, in step 2, the preliminarily processed remote sensing image data is input into a trained machine model for feature comparison, and an image of the corresponding area with potential hazards is output. Then, the image of the area with potential hazards is re-input into another machine model trained according to the characteristics of the disaster body to output the disaster points and perform annotation.

[0018] Furthermore, the specific process of sampling in the soft rock area in step 3 is as follows: Before construction on the soft rock mass, several samples are taken by drilling. A camera device is used for identification. The camera device consists of four cameras, a ring, two sample fixing ends, a telescopic cylinder, and a connecting rod. The two sample fixing ends are located at both ends of the connecting rod respectively, used to fix both ends of the sample. The ring surrounds the outside of the sample and can slide on the connecting rod. The four cameras are installed on the side of the ring, facing the east, south, west, and north directions of the sample respectively, and are set to aim at the sample. The telescopic end of the telescopic cylinder is connected to the ring, and by driving the ring to move back and forth, the four cameras can perform peripheral identification on the sample during the movement;

[0019] Samples are extracted by means of circular drilling. If the sample breaks during drilling, it is bonded and repaired with epoxy resin. Subsequently, the sample is identified by the four cameras, and the images captured by the four cameras are stitched together to obtain the video data of the periphery of the sample. Then, image recognition processing is performed on the video data to identify the stratification of the soft rock in the sample, the materials of the soft rock layers, and the distribution and size of the pores. Finally, the characteristic data related to the soft rock layer materials is automatically retrieved through the Internet.

[0020] Furthermore, the specific process of establishing the three-dimensional space model of the soft rock mass in Step 3 is as follows: Construct a three-dimensional space model, which is constructed according to the dimensions of the slope to be monitored in accordance with the corresponding ratio. Place the three-dimensional peripheral images of all samples into the three-dimensional space model. According to the different sampling positions of each sample, place the samples at the corresponding proportional position points in the space model. In the three-dimensional space model, place the position points at the same depth of each sample at the same height. Identify the soft rock model inside the columns of each sample, and obtain the three-dimensional model of the soft rock of each sample through the finite element analysis method. In the three-dimensional space model, connect the points with the same properties or the same rock layers at the same height with lines, and use different line types or colors to distinguish the points at different heights to obtain the image color and void distribution in the sample. Render the connection lines at the same height or with the same properties into layers, and move outward from the center of the sample until the entire three-dimensional space model is rendered, thereby obtaining the three-dimensional model of the soft rock mass.

[0021] Furthermore, in Step 4, install detection devices on all the taken samples, put the samples back into the soft rock mass for real-time detection. Install multiple humidity sensors, stress sensors, and displacement sensors on the side of the sampled samples, and connect the transmission lines to the outside. There is a collection device outside to collect data in real time and transmit it back to the data processing system. In the initial collection stage, the collected data represents the initial state of the soft rock mass, that is, its stable state data;

[0022] Among them, during the collection, if there is no hidden danger area that does not appear in the remote sensing census, and the collected data does not exceed the set value within the set time period, reduce the collection rate, adopt the adaptive variable-frequency intelligent collection method, and use the low-frequency collection mode to avoid the large accumulation of redundant data;

[0023] Collect the internal data of the soft rock mass before construction, and calculate the self-gravity of the soft rock according to the material properties of the soft rock. The material properties include hardness and density. Detect the humidity data through the humidity sensor, and identify the pore structure to determine the structural stress of the soft rock in the initial state. Superimpose the self-gravity and the structural stress to obtain the initial stress state of the soft rock mass. Subsequently, perform fusion processing on the data related to the initial stress state and the three-dimensional model of the soft rock mass, so that the stress data of the three-dimensional model of the soft rock mass can be updated in real time according to the changes in the detection data, and the user can observe the stress condition in the three-dimensional model of the soft rock mass in real time through the display screen.

[0024] Further, in step 4, according to the material properties of different depths of soft rock identified, the strength reduction method is adopted to gradually reduce the shear strength parameters of the material, which include cohesion and internal friction angle, until the slope reaches the critical state, that is, failure occurs. At this time, it is the instability condition of the slope. According to this instability condition, the instability failure boundary condition is applied, and the three-dimensional model of the soft rock mass is backed up as the second three-dimensional model of the soft rock mass. The failure condition is simulated in the second three-dimensional model of the soft rock mass to make the soft rock mass gradually tend to the critical state, so as to determine the area where instability occurs fastest and mark the key area of instability. Finally, the simulated key area of instability is integrated into the original three-dimensional model of the soft rock mass to form a three-dimensional model of the soft rock mass with area annotation.

[0025] Further, in step 5, both remote sensing data and collected data need to be taken into account. When there is a hidden danger point in one of them, the other needs to conduct a detailed verification of the corresponding area. When the remote sensing data judges that there is a hidden danger, and then when there is a data change detected in the field soft body area, it is judged that there will be a landslide or instability hidden danger in the corresponding area, and the corresponding personnel are notified to supervise the corresponding area. When there is no hidden danger in the remote sensing and there is a hidden danger detected in the field soft rock mass, it is determined that the construction soft rock mass area is unstable;

[0026] Among them, the online early warning system is used to display the disaster early warning supervision data of the whole region, zoom in on each detected area at any time, and at the same time, a corresponding three-dimensional dynamic model is generated for a specific soft rock area, so that the management personnel can intuitively see the change of the three-dimensional model from normal data to the whole instability dynamic, and conduct real-time analysis according to the change of the three-dimensional model.

[0027] Further, in step 6, in the instability data, if it is due to the increase in humidity, then drilling and pumping are carried out. The bottom hole is drilled for vacuum pumping, and nitrogen is introduced into the upper hole. After the humidity reaches the original humidity, both the pumping hole and the nitrogen through hole are sealed. If there is movement or stress increase, it is judged that it is caused by vibration or internal oxidation and loosening, then anchor rods need to be used for reinforcement and cement slurry is injected into the anchor rods for reinforcement until the data collected again returns to the error range set by the initial data;

[0028] If there are hidden dangers in both the remote sensing data and the real-time monitoring three-dimensional dynamic model in the collected data, the corresponding area is isolated, and regular observation and artificial intervention blasting are used to intervene in the occurrence time of the hidden danger.

[0029] Due to the adoption of the above technical solutions, the present invention has the following beneficial effects:

[0030] Through remote sensing data and real-time detection inside the soft rock mass, the present invention can better detect the slope area, without missing any potential hazard point. At the same time, it first conducts a screening from a large-scale space to provide better detection points for subsequent precise identification. According to the previous screening, it reduces the real-time collection of each detection point and avoids the large accumulation of redundant data. Samples of different points of the soft rock are collected, and the samples are identified through pattern recognition and analyzed by the finite element method to obtain a three-dimensional solid model of the soft rock of the samples. All the samples are combined, and a model of the entire soft rock mass is generated through identification and rendering. The samples are put back into the soft rock mass, and the data of the soft rock mass are collected in real time. The collected data are put into the model of the soft rock mass, and the change of the data in the model of the soft rock mass can be seen in real time, and the change of the internal structure of the soft rock mass can be seen more directly, and the data is more accurate. Corresponding protective measures are given according to the change of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is the flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the following preferred embodiments are cited with reference to the accompanying drawings for a further detailed description of the present invention. However, it should be noted that many details listed in the specification are only for enabling the reader to have a thorough understanding of one or more aspects of the present invention, and these aspects of the present invention can be implemented even without these specific details.

[0033] As Figure 1 shown, for the three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes, the method includes the following steps:

[0034] Step 1: Use remote sensing equipment to photograph the highway subgrade to obtain corresponding remote sensing image data. The remote sensing equipment includes general surveys of geological disasters based on spaceborne platforms and detailed surveys of geological disasters based on airborne platforms. Make full use of the advantages of each platform of the sky, space and ground, and establish a geological disaster identification and risk assessment system that combines the three and complements each other.

[0035] The specific process of geological disaster census based on the spaceborne platform is as follows: By using the InSAR combined with optical remote sensing method through the satellite remote sensing platform, the information of landslide morphology and ground object classification is extracted by optical remote sensing, and the deformation field is extracted by InSAR. By combining the two, factors such as surface deformation rate, ground object classification, ground inversion temperature, and vegetation coverage are obtained, and a geological hazard knowledge graph is constructed. The characteristics and their correlation relationships of the three elements of the disaster-bearing environment, inducing factors, and change signs in landslide hazards are constructed. The disaster-bearing environment includes geological structure, hydrological conditions, surface undulation, and surface coverage. The inducing factors include precipitation and soil freeze-thaw. The change signs include displacement and settlement. Analyze the characteristics and formation conditions of major geological disasters, screen and determine the key factors for the occurrence of geological disasters. Through the analysis and calculation of disaster-causing factors, use machine learning methods to establish an automatic geological disaster identification system to achieve automatic or semi-automatic extraction of geological disasters.

[0036] The specific process of detailed investigation of geological disasters based on the airborne platform is as follows: Use the unmanned aerial vehicle (UAV) platform, set a laser scanner and a high-resolution optical camera on the UAV platform to obtain ground three-dimensional coordinates and image data, and realize the real-time change of the three-dimensional ground objects and the reproduction of the real morphological characteristics. Calculate the surface vertical displacement, volume change, and cross-section before and after the change. Filter the point cloud data of steep slopes and complex terrain with dense vegetation. Obtain the digital surface model by removing vegetation from LiDAR data, and combine two-dimensional and three-dimensional interpretation to identify the potential hazards of geological disasters such as collapsed loose deposits, ancient landslide deposits, debris flow fans, and earthquake-cracked mountain cracks under vegetation coverage, so as to achieve a detailed investigation of high-risk areas of geological disasters.

[0037] Step 2: Identify the remote sensing image and input the image into the machine model for feature contrast learning to carry out screening of potential hazard points and identification of disaster bodies. Input the preliminarily processed remote sensing image data into the trained machine model for feature contrast, and output the image of the corresponding area with potential hazards. Then input the image of the area with potential hazards into another machine model trained according to the characteristics of the disaster body to output the disaster points and mark them. The machine model is a convolutional neural network, which is suitable for processing image data. Through two identifications, the processing data of the image can be reduced, and the feature recognition is more targeted.

[0038] Step 3: Sample the soft rock area and establish a three-dimensional spatial model of the soft rock mass. Before construction of the soft rock mass, drill several samples, and use a camera device for identification. The camera device consists of four cameras, a ring, two sample fixing ends, a telescopic cylinder, and a connecting rod. The two sample fixing ends are located at both ends of the connecting rod respectively, and are used to fix both ends of the sample. The ring surrounds the outside of the sample and can slide on the connecting rod. The four cameras are installed on the side of the ring, facing the east, south, west, and north directions of the sample respectively, and are set to aim at the sample. The telescopic end of the telescopic cylinder is connected to the ring, and by driving the ring to move reciprocally, the four cameras can perform peripheral identification on the sample during the movement.

[0039] Extract samples using a circular drilling method. If the sample breaks during drilling, use epoxy resin for bonding and repair. Subsequently, identify the sample through the four cameras, splice the images captured by the four cameras to obtain video data of the sample periphery. Then, perform image recognition processing on the video data to identify the stratification of the soft rock in the sample, the materials of the soft rock layers, and the distribution and size of the pores. Finally, automatically retrieve the characteristic data related to the soft rock layer materials through the Internet.

[0040] Construct a three-dimensional spatial model, which is constructed according to the required slope size at a corresponding ratio. Place the three-dimensional peripheral images of all samples into the three-dimensional spatial model, and place the samples at the corresponding position points in the spatial model according to the different sampling positions of each sample. In the three-dimensional spatial model, place the position points at the same depth of each sample at the same height, identify the soft rock model inside the columns of each sample, and obtain the three-dimensional model of the soft rock of each sample through the finite element analysis method. In the three-dimensional spatial model, connect the points with the same properties or the same rock layers at the same height with lines, and use different line types or colors to distinguish the points at different heights to obtain the image color and void distribution in the sample. Render the connection lines at the same height or with the same properties into layers, and move outward from the center of the sample until the entire three-dimensional spatial model is rendered, thereby obtaining the three-dimensional model of the soft rock mass.

[0041] Step 4: Real-time collect the data inside the soft rock mass by setting up a collection device, and the collected data serves as the dynamic data of the three-dimensional spatial model of the soft rock mass. Install detection devices on all the taken samples, put the samples back into the soft rock mass for real-time detection. Install multiple humidity sensors, stress sensors, and displacement sensors on the side of the sampled samples, and connect the transmission lines to the outside. There is a collection device outside to collect data in real time and transmit it back to the data processing system. In the initial collection stage, the collected data represents the initial state of the soft rock mass, that is, its stable state data.

[0042] Among them, during the acquisition, if the data collected in the hidden danger area that did not appear in the remote sensing census do not exceed the set value within the set time period, the acquisition rate will be reduced. For the adaptive variable-frequency intelligent acquisition method, the low-frequency acquisition mode is adopted to avoid the large accumulation of redundant data;

[0043] Collect the internal data of the soft rock mass before construction, and calculate the self-gravity of the soft rock according to the material properties of the soft rock. The material properties include hardness and density. Detect the humidity data through a humidity sensor and identify the pore structure to determine the structural stress of the soft rock in the initial state. Superimpose the self-gravity and the structural stress to obtain the initial stress state of the soft rock mass. Subsequently, fuse the data related to the initial stress state with the three-dimensional model of the soft rock mass, so that the stress data of the three-dimensional model of the soft rock mass can be updated in real time according to the changes in the detection data, and users can observe the stress conditions in the three-dimensional model of the soft rock mass in real time through the display screen.

[0044] According to the identified material properties of the soft rock at different depths, use the strength reduction method to gradually reduce the shear strength parameters of the material, which include cohesion and internal friction angle, until the slope reaches the critical state, that is, failure occurs. At this time, it is the instability condition of the slope. According to this instability condition, apply the instability failure boundary condition, back up the three-dimensional model of the soft rock mass as the second three-dimensional model of the soft rock mass, simulate the failure condition in the second three-dimensional model of the soft rock mass, make the soft rock mass gradually tend to the critical state, so as to determine the area where instability occurs fastest, and mark the key area of instability. Finally, integrate the simulated key area of instability into the original three-dimensional model of the soft rock mass to form a three-dimensional model of the soft rock mass with area annotation.

[0045] Step 5: Input the remote sensing data and the real-time collected data into the online early warning system to judge whether there is instability in the soft rock area. Both the remote sensing data and the collected data need to be considered. When there is a hidden danger point in one of them, the other one needs to conduct a detailed verification of the corresponding area. When the remote sensing data judges that there is a hidden danger and there is a data change detected in the field soft body area, it is judged that there is a landslide hidden danger in the corresponding area, and the corresponding personnel are notified in time for area supervision. When there is no hidden danger in the remote sensing and there is a hidden danger detected in the field soft rock mass, it is determined that the soft rock mass area under construction is unstable;

[0046] Among them, the online early warning system is used to display the disaster early warning supervision data of the entire region, zoom in on each detected area at any time, and at the same time, a corresponding three-dimensional dynamic model is generated for a specific soft rock area. Managers can intuitively see the change of the three-dimensional model from normal data to the entire instability dynamic, and conduct real-time analysis according to the change of the three-dimensional model.

[0047] Step 6: Issue early warning information for areas where instability or landslides may occur, and notify the management personnel to conduct on-site inspections and take protective measures. According to the instability data that may be caused as queried by the management personnel, if it is due to increased humidity, drilling and pumping are carried out. The bottom is drilled for vacuum pumping, and nitrogen is introduced through the upper hole. After the humidity reaches the original humidity, both the pumping holes and the nitrogen through-holes are sealed. If there is movement or increased stress, and it is judged that it is caused by vibration or loosening due to internal oxidation, then anchor rods need to be used for reinforcement and cement slurry is injected into the anchor rods for reinforcement until the data is collected again and returns to the error range set by the initial data.

[0048] If potential hazards appear in the remote sensing data and also in the real-time monitoring 3D dynamic model, the corresponding area is isolated, and regular observations are made and artificial intervention blasting is used to intervene in the occurrence time of the hazards.

[0049] Effectively identify geological hazard instability or landslide areas, propose treatment plans for geological hazards at different landslide stages, and improve the response ability of geological hazards to complex working condition changes. For deformation monitoring data, pre-reinforcement protection is carried out for soft rock geological hazards that may exceed the disaster early warning value, and emergency rescue and treatment are carried out for geological hazards that have exceeded the disaster early warning value. In the early stage, small and fast rescue methods with high cost performance such as anchor bar piles, steel pipe piles or steel light rails are used for treatment, and in the later stage, it is combined with anchor rod retaining walls, anti-slide piles, etc. for treatment, forming an idea of prevention first, combination of prevention and control, emergency rescue and comprehensive treatment. Optimize the design of the geological hazard retaining structure, combine the stress and deformation data of the retaining structures such as anchor rods and anti-slide piles obtained from on-site monitoring, carry out numerical simulation analysis, and optimize the design of the retaining structure type, size, reinforcement, etc. under different working conditions, making the geological hazard treatment more economical, safe, green and environmentally friendly

[0050] Matters not covered by this invention are well-known technologies.

[0051] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes, characterized in that: The method includes the following steps: Step 1: Use a remote sensing device to photograph the highway subgrade and obtain corresponding remote sensing image data; Step 2: Identify the remote sensing image, input the image into the machine model for feature contrast learning, and conduct hidden danger point screening and disaster body identification; Step 3: Sample the soft rock area and establish a three-dimensional space model of the soft rock mass; Step 4: Set up a collection device to collect the data inside the soft rock mass in real time, and the collected data is used as the dynamic data of the three-dimensional space model of the soft rock mass; Step 5: Input the remote sensing data and the real-time collected data into the online early warning system to judge whether there is instability in the soft rock area; Step 6: Send out early warning information for the areas where instability or landslides may occur, and notify the management personnel to conduct on-site inspections and take protective measures.

2. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: In Step 1, the remote sensing device includes geological disaster general surveys based on satellite-borne platforms and geological disaster detailed surveys based on airborne platforms. The specific process of the geological disaster general survey based on the satellite-borne platform is as follows: Using the method of combining InSAR with optical remote sensing through the satellite remote sensing platform, using optical remote sensing to extract information on landslide morphology and land cover classification, and InSAR to extract the deformation field. By combining the two, factors such as the surface deformation rate, land cover classification, ground inversion temperature, and vegetation coverage are obtained, and a geological hazard knowledge graph is constructed. The characteristics and their correlation relationships of the three elements of the disaster-bearing environment, inducing factors, and change signs in the landslide hazard are constructed. The disaster-bearing environment includes geological structure, hydrological conditions, surface undulation, and surface coverage. The inducing factors include precipitation and soil freeze-thaw. The change signs include displacement and settlement. Analyze the characteristics and formation conditions of major geological disasters, screen and determine the key factors for the occurrence of geological disasters, and through disaster-causing factor analysis and calculation, use machine learning methods to establish an automatic geological disaster recognition system to achieve automatic or semi-automatic extraction of geological disasters.

3. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 2, characterized in that: The specific process of the geological disaster detailed survey based on the airborne platform is as follows: Use an unmanned aerial vehicle (UAV) platform, set up a laser scanner and a high-resolution optical camera on the UAV platform to obtain ground three-dimensional coordinates and image data, and reproduce the three-dimensional real-time changes and true morphological characteristics of the ground objects. Calculate the vertical displacement, volume change, and cross-section before and after the change of the surface, and filter the point cloud data of steep slopes and complex terrain with dense vegetation. Obtain the digital surface model by removing vegetation from LiDAR data, and combine two-dimensional and three-dimensional interpretation to identify the potential hidden dangers of geological disasters such as collapsed loose accumulation bodies, ancient landslide accumulation bodies, debris flow accumulation fans, and earthquake-cracked mountain cracks under vegetation coverage, so as to conduct a detailed survey of the geological disaster area.

4. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: In Step 2, input the preliminarily processed remote sensing image data into the trained machine model for feature contrast, output the image of the corresponding hidden danger area, then input the image of the hidden danger area into another machine model trained according to the characteristics of the disaster body, output the disaster points, and conduct annotation.

5. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: The specific process of sampling the soft rock area in Step 3 is as follows: Before constructing the soft rock mass, drill several samples, and use a camera device for identification. The camera device consists of four cameras, a ring, two sample fixing ends, a telescopic cylinder, and a connecting rod. The two sample fixing ends are located at both ends of the connecting rod respectively, and are used to fix both ends of the sample. The ring surrounds the outside of the sample and can slide on the connecting rod. The four cameras are installed on the side of the ring, facing the east, south, west, and north directions of the sample respectively, and are set to aim at the sample. The telescopic end of the telescopic cylinder is connected to the ring, and by driving the ring to move back and forth, the four cameras can perform peripheral identification on the sample during the movement; Adopt a circular drilling method to extract samples. If the sample breaks during the drilling process, use epoxy resin for bonding and repair. Subsequently, identify the sample through four cameras, splice the images captured by the four cameras to obtain video data of the sample periphery. Then, perform image recognition processing on the video data to identify the soft rock stratification, soft rock layer materials, and the distribution and size of pores in the sample. Finally, automatically retrieve the characteristic data related to the soft rock layer materials through the Internet.

6. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 5, characterized in that: The specific process of establishing the three-dimensional space model of the soft rock mass in Step 3 is as follows: Construct a three-dimensional space model, which is constructed according to the required slope size at a corresponding ratio. Place the three-dimensional peripheral images of all samples into the three-dimensional space model. According to the different sampling positions of each sample, place the sample at the corresponding position point in the space model at the corresponding ratio. In the three-dimensional space model, place the position points at the same depth of each sample at the same height, identify the soft rock model inside the column of each sample, and obtain the three-dimensional model of the soft rock of each sample through the finite element analysis method. In the three-dimensional space model, connect the points with the same nature or the same rock layer at the same height with lines, and use different line types or colors to distinguish the points at different heights to obtain the image color and void distribution in the sample. Render the connection lines at the same height or with the same nature into layers, and move from the center of the sample outward until the entire three-dimensional space model is rendered, thereby obtaining the three-dimensional model of the soft rock mass.

7. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: In Step 4, install detection devices on all the taken samples, put the samples back into the soft rock mass for real-time detection. Install multiple humidity sensors, stress sensors, and displacement sensors on the side of the sampled samples, and connect the transmission lines to the outside. There is a collection device outside to collect data in real time and transmit it back to the data processing system. In the initial collection stage, the collected data represents the initial state of the soft rock mass, that is, its stable state data; Among them, during the collection, if there is no hidden danger area that does not appear in the remote sensing census, and the collected data does not exceed the set value within the set time period, reduce the collection rate, adopt an adaptive variable-frequency intelligent collection method, and use a low-frequency collection mode to avoid a large accumulation of redundant data; Collect the internal data of the soft rock before construction, and calculate the self-gravity of the soft rock based on the material properties of the soft rock. The material properties include hardness and density. Detect the humidity data through a humidity sensor and identify the pore structure to determine the structural stress of the soft rock in the initial state. Superimpose the self-gravity and the structural stress to obtain the initial stress state of the soft rock mass. Subsequently, fuse the data related to the initial stress state with the three-dimensional model of the soft rock mass, so that the stress data of the three-dimensional model of the soft rock mass can be updated in real time according to the changes in the detection data. The user can observe the stress condition in the three-dimensional model of the soft rock mass in real time through the display screen.

8. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: In step 4, according to the identified material properties of the soft rock at different depths, use the strength reduction method to gradually reduce the shear strength parameters of the material. These parameters include cohesion and internal friction angle until the slope reaches the critical state, that is, failure occurs. At this time, it is the instability condition of the slope. According to this instability condition, apply the instability failure boundary condition, back up the three-dimensional model of the soft rock mass as the second three-dimensional model of the soft rock mass, simulate the failure condition in the second three-dimensional model of the soft rock mass, make the soft rock mass gradually tend to the critical state, so as to determine the area where instability occurs fastest and mark the key area of instability. Finally, integrate the simulated key area of instability into the original three-dimensional model of the soft rock mass to form a three-dimensional model of the soft rock mass with area markings.

9. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, characterized in that: In step 5, both remote sensing data and collected data need to be taken into account. When there is a hidden danger point in one of them, the other needs to conduct a detailed verification of the corresponding area. When the remote sensing data judges that there is a hidden danger, and then when there is a data change detected in the field soft body area, it is judged that a landslide or instability hidden danger will occur in the corresponding area, and the corresponding personnel will be notified to supervise the corresponding area. When there is no hidden danger in remote sensing and there is a hidden danger detected in the field soft rock mass, it is determined that the construction soft rock mass area is unstable; Among them, the online early warning system is used to display the disaster early warning supervision data of the entire region, zoom in on each detected area at any time, and at the same time, a corresponding three-dimensional dynamic model is generated for a specific soft rock area. The management personnel can intuitively see the change of the three-dimensional model from normal data to the entire instability dynamic, and conduct real-time analysis according to the change of the three-dimensional model.

10. The three-dimensional intelligent monitoring and early disaster identification and warning method for soft rock slopes according to claim 1, wherein: In step 6, in the instability data, if it is due to an increase in humidity, then drill holes for pumping water. Drill holes at the bottom for vacuum pumping, and drill holes at the upper end for introducing nitrogen. After the humidity reaches the original humidity, seal both the pumping holes and the nitrogen through holes. If there is movement or an increase in stress, it is judged that it is caused by vibration or internal oxidation and loosening, then it is necessary to use anchor rods for reinforcement and inject cement slurry inside the anchor rods for reinforcement until the data collected again returns to the error range set by the initial data; If there are hidden dangers in both the remote sensing data and the real-time monitoring three-dimensional dynamic model in the collected data, isolate the corresponding area and observe it regularly and use artificial intervention blasting to intervene in the occurrence time of the hidden danger.

Images