Geological disaster hidden danger point detection method and detection system

Abstract

The invention relates to the technical field of geological disaster detection, in particular to a geological disaster hidden danger point detection method and system. Comprising the following steps: S1, synchronously acquiring vegetation canopy penetration data, multispectral soil humidity spatial distribution data and lithology distribution data by adopting an unmanned aerial vehicle carried image data acquisition device and a multi-band laser radar, and generating an orthoimage and a three-dimensional point cloud data set; s2, fusing the orthoimage and the three-dimensional point cloud data through an aerial triangulation encryption algorithm to establish a three-dimensional model, loading the multispectral soil humidity spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generating a contour line topological structure through a TIN triangulation network algorithm; s3, extracting a slope factor, topographic relief, vegetation coverage, fracture distance and a lithologic factor based on the three-dimensional model to construct a geological disaster hidden danger system; according to the system, the geological disaster occurrence time can be estimated, and the rescue and protection treatment period can be effectively planned based on the geological disaster occurrence time.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster detection, and particularly to a method and a system for detecting potential geological disaster points. Background Art

[0002] Geological disasters refer to various natural disasters caused by factors such as crustal movement, groundwater level change, and climate change during the process of the Earth's surface activities. Geological disasters are characterized by suddenness, destructiveness, and unpredictability, posing a serious threat to social and economic development and the safety of people's lives and property. In order to reduce the losses caused by geological disasters, detecting and preventing geological disasters have become crucial. Potential geological disaster points are areas where disasters may occur. Detecting the changes in potential disaster points can provide early warnings and corresponding preventive measures. Currently, in the existing technologies, detection techniques for potential geological disaster points are used to detect and record potential geological disaster points, and later, the potential geological disaster points are processed or early rescue measures are taken. However, the current detection systems for potential geological disaster points can only detect potential geological disaster points, but cannot estimate the occurrence time of geological disasters, which easily leads to the occurrence of geological disasters before the processing project and rescue measures are completed.

[0003] Based on the above problems, there is an urgent need for a system that can detect potential geological disaster points and effectively estimate the occurrence time of geological disasters, and based on the occurrence time of geological disasters, effectively plan the construction period of rescue and protection management. Summary of the Invention

[0004] In view of the current inability to estimate the occurrence time of geological disasters in the detection of potential geological disaster points, which easily leads to the occurrence time of geological disasters exceeding the time of early rescue and protection construction period, this application provides a method and a system for detecting potential geological disaster points to solve the above problems.

[0005] To achieve the above object, the present invention is realized through the following technical solutions: An embodiment of this application discloses a method for detecting potential geological disaster points, including: S1: Using a drone to carry an image data acquisition device and a multi-band lidar to synchronously collect vegetation canopy penetration data, multi-spectral soil moisture spatial distribution data, and lithology distribution data, and generating an orthophoto and a three-dimensional point cloud dataset; S2: Fusing the orthophoto and the three-dimensional point cloud data through an aerial triangulation encryption algorithm to establish a three-dimensional model, loading the multi-spectral soil moisture spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generating a contour topology structure through a TIN triangulation algorithm; S3: Extract slope factors, terrain undulation degree, vegetation coverage, fracture distance, and lithology factors based on the 3D model to construct a geological hazard potential system; S4: Use the improved random forest algorithm to assign weights to potential factors, input real-time meteorological data and historical disaster case databases, and input the occurrence probabilities of boulder collapses and natural slope collapses; S5: Calculate the disaster occurrence time based on the potential probability, lithology stability coefficient, and dynamic rainfall factor, and generate the corresponding latest project start time according to the project treatment volume, equipment efficiency, and safety redundancy time, ensuring that the latest project start time is less than the disaster occurrence time minus the safety redundancy time.

[0006] Adopting the above technical solution: This solution can improve the terrain modeling accuracy to the centimeter level through the collaborative acquisition of multi-source data by drones; the dynamic DEM model realizes the fusion analysis of terrain and hydrological data and supports real-time geological state assessment; the dual-time coupling mechanism directly associates geological stability prediction with engineering response, breaking through the limitation of passive early warning and being able to assist in the timely treatment and maintenance of geological hazard points before the occurrence of geological disasters. Preferably, the calculation formula for the disaster occurrence time is: ; where P is the collapse occurrence probability of the potential point, T1 is the disaster occurrence time, is the lithology compressive strength coefficient, is the rock mass fracture density correction coefficient, and when the fracture density increases by 5 fractures / m, the value decreases by 0.1, is the dynamic rainfall influence factor, and when the rainfall exceeds 50 mm for 3 consecutive days, the value increases by 0.3.

[0007] Adopting the above technical solution: The above solution can effectively evaluate the disaster occurrence time by analyzing data such as the lithology compressive strength coefficient, rock mass fracture density correction coefficient, dynamic rainfall influence factor, and the collapse occurrence probability of the potential point, and the prediction error is within 24 hours.

[0008] Further preferably, the calculation formula for the latest start time is: ; where: T2 is the latest start time, k is the construction machinery scheduling coefficient, and the construction machinery scheduling coefficient is dynamically adjusted according to the number of available equipment; E is the daily processing efficiency, and the daily processing efficiency is predicted by the LSTM neural network, W is the project treatment volume, is the safety redundancy time.

[0009] Adopting the above technical solution: The above solution lacks a dynamic adjustment mechanism for linking the disposal time planning of maintenance projects with geological prediction. The above solution combines the mechanical scheduling coefficient k and the LSTM prediction efficiency E, and can achieve a 30% increase in the optimization rate of the project response time. Further preferably, in S1, the calculation formula for vegetation canopy penetration data is: ; Among them, represents the number of lidar point clouds in the surface layer, characterizing the number of ground reflection points not blocked by vegetation; represents the total number of lidar scan point clouds, including all reflection points in the vegetation layer and the surface layer; represents the percentage of vegetation penetration rate, reflecting the penetration ability of lidar to the vegetation canopy. The lower the value, the denser the vegetation coverage.

[0010] Adopting the above technical solution: Based on the lack of quantitative indicators for the root soil-fixing strength in the current geological disaster assessment in vegetation-covered areas, the point cloud stratification algorithm in this design quantifies the vegetation penetration rate and accurately identifies weak areas of vegetation soil-fixing ability; the improved TCARI / OSAVI model controls the soil moisture inversion error within 5%, improving the accuracy of hydrological parameters. Further preferably, in S2, the air three encryption algorithm adopts SIFT-GPU to accelerate feature matching, and the dynamic DEM model update trigger condition is to start real-time model reconstruction when the single-day slope change is more than 0.5 degrees.

[0011] Adopting the above technical solution: Based on the lag in the update of traditional 3D modeling technology, it cannot reflect the dynamic changes of the terrain in real time; the SIFT-GPU accelerated feature matching technology increases the model reconstruction speed by more than 3 times. The dynamic DEM trigger mechanism starts when the slope change ≥ 0.5°, which can ensure the timeliness of the model, and the error ≤ 2 cm. Further preferably, the calculation method of the terrain undulation degree is: divide the DEM data into 10m x 10m grids, calculate the maximum elevation value and the minimum elevation value within the grid. The difference between the maximum elevation value and the minimum elevation value is the terrain undulation degree. When the terrain undulation degree is above 15m, a terrain mutation warning is triggered.

[0012] Adopting the above technical solution: The traditional identification of terrain mutation areas relies on manual experience judgment and lacks a standardized evaluation system; the grid elevation range difference calculation provided in this application can achieve automatic quantification of the terrain undulation degree, and the accuracy of the trigger threshold warning is increased by 40%. Further preferably, the improved random forest algorithm includes a feature importance dynamic correction module, which automatically triggers model retraining when the weight of the slope factor S changes by more than 10% in three consecutive iterations; the identification of boulder collapses uses the DBSCAN density clustering algorithm, with the clustering radius set to 0.5 m and the minimum number of points threshold set to 200 points.

[0013] Adopting the above technical solution: Based on the fixed feature weights of the traditional random forest algorithm, it cannot adapt to the dynamic geological environment; the feature importance dynamic correction module provided in this application can improve the model prediction accuracy by 15% when retraining is triggered when the weight change ≥ 10%, and the DBSCAN density clustering algorithm can achieve a boulder volume identification accuracy ≥ 95%. A detection system, applied to the geological disaster hidden danger point detection method described in any one of the above, includes: Data acquisition and processing module: The data acquisition and processing module uses an unmanned aerial vehicle to carry an image data acquisition device and a multi-band lidar to synchronously acquire vegetation canopy penetration data, multi-spectral soil moisture spatial distribution data, and lithology distribution data, and generates an orthophoto and a three-dimensional point cloud dataset; High-line topology generation module: Establish a three-dimensional model by fusing the orthophoto and the three-dimensional point cloud data through the aerial triangulation encryption algorithm, load the multi-spectral soil moisture spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generate a contour topology through the TIN triangulation algorithm; Geological disaster hidden danger system construction module: Extract slope factors, terrain undulation, vegetation coverage, fracture distance, and lithology factors based on the three-dimensional model to construct a geological disaster hidden danger system; Weight allocation module: Use the improved random forest algorithm to allocate weights to the hidden danger factors, input real-time meteorological data and a historical disaster case library, and input the occurrence probabilities of boulder collapses and natural slope collapses; Engineering time calculation module: Calculate the disaster occurrence time based on the hidden danger probability, lithology stability coefficient, and dynamic rainfall factor, and generate the corresponding latest engineering start time according to the engineering treatment volume, equipment efficiency, and safety redundancy time, ensuring that the latest engineering start time is less than the disaster occurrence time minus the safety redundancy time.

[0014] Further preferably, it further includes: Priority intelligent sorting module: Divide the hidden danger points into red warnings, orange warnings, and yellow warnings. When the disaster occurrence time minus the latest start time is less than 15 days, a red warning is issued. An orange warning is issued when the disaster occurrence time minus the latest start time is between 15 - 30 days. A yellow warning is issued when the disaster occurrence time minus the latest start time is more than 30 days; 3D Visualization Module: Overlay and display the heat map of potential hazard points, the engineering treatment progress area, and the simulated line of predicted collapse trajectory in the real-scene model.

[0015] Further preferably, it further includes: Edge Computing Node Module: Equipped with NVIDIA Jetson AGX Xavier processor to process point cloud data in real time; Cloud Analysis Platform: Adopting a microservices architecture, including model construction service, potential hazard analysis service, and early warning push service; IoT Engineering Interface: Supporting direct connection to the CAN bus of excavators and anchoring equipment. Description of the Drawings

[0016] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the following described drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0017] Figure 1 It is a flowchart of a method for detecting potential geological hazard points in the present application. Detailed Embodiments

[0018] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0019] It should be understood that when used in the specification and appended claims of the present application, the term "including" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0020] Please refer to Figure 1 , for example, the detection of the original potential geological hazard points can realize the corresponding calculation, analysis, and positioning of potential geological hazard points by combining various data. However, it is impossible to effectively evaluate the time of geological disasters occurring at locations with potential geological hazards, and it is impossible to effectively estimate the construction period for subsequent protection and treatment of potential geological hazard points, which may lead to geological disasters occurring before maintenance. Based on the above problems, the embodiments of the present application disclose a method for detecting potential geological hazard points, including: S1: Use a drone to carry an image data acquisition device and a multi-band lidar to synchronously collect vegetation canopy penetration data, multi-spectral soil moisture spatial distribution data, and lithology distribution data, and generate an orthophoto and a three-dimensional point cloud dataset; the image data acquisition device in the above can use an inclined camera equipped with five lenses, where the vertical resolution of the oblique photography camera ≤ 3 cm, the lidar point cloud density ≥ 200 points / ㎡, and the multi-spectral soil moisture inversion error ≤ 5%‌ S2: Establish a three-dimensional model by fusing the orthophoto and the three-dimensional point cloud data through an aerial triangulation encryption algorithm, load the multi-spectral soil moisture spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generate a contour topology structure through the TIN triangulation algorithm; S3: Extract slope factors, terrain undulation, vegetation coverage, fracture distance, and lithology factors based on the three-dimensional model to construct a geological disaster hazard system; S4: Use an improved random forest algorithm to assign weights to the hazard factors, input real-time meteorological data and a historical disaster case library, and input the occurrence probabilities of boulder collapse and natural slope collapse; S5: Calculate the disaster occurrence time based on the hazard probability, lithology stability coefficient, and dynamic rainfall factor, and generate the corresponding latest project start time according to the project treatment volume, equipment efficiency, and safety redundancy time, ensuring that the latest project start time is less than the disaster occurrence time minus the safety redundancy time.

[0021] It is worth mentioning that: This solution can improve the terrain modeling accuracy to the centimeter level through the collaborative acquisition of multi-source data by drones; the dynamic DEM model realizes the fusion analysis of terrain and hydrological data and supports real-time geological state assessment; the dual-time coupling mechanism directly associates geological stability prediction with engineering response, breaks through the limitation of passive early warning, and can assist in timely handling and maintenance of geological disaster points before the occurrence of geological disasters.‌‌‌ The calculation formula for the disaster occurrence time is: ; where P is the collapse occurrence probability of the hazard point. It should be noted that the collapse occurrence probability of the hazard point is the collapse probability output by the improved random forest algorithm and is dynamically updated based on multi-source data. The multi-source data includes: slope, soil moisture content, and vegetation coverage. T1 is the disaster occurrence time, is the lithology compressive strength coefficient, which is assigned values according to the rock type classification, is the rock mass fracture density correction coefficient, which is quantified by the lidar point cloud fracture identification algorithm. When the fracture density increases by 5 fractures / m, the value decreases by 0.1, is the dynamic rainfall influence factor, which is jointly calculated based on the real-time rainfall intensity (mm / h) and the historical infiltration rate. When the continuous 3-day rainfall is above 50 mm The value increases by 0.3.

[0022] It is worth mentioning that by integrating data such as the lithology compressive strength coefficient, the corrected coefficient of rock fracture density, the dynamic rainfall influence factor, and the probability of collapse at potential hazard points, the above scheme can effectively evaluate the time of disaster occurrence, and the prediction error is within 24 hours.

[0023] The calculation formula for the latest start time is: ; Where: T2 is the latest start time, k is the construction machinery scheduling coefficient, which is dynamically adjusted according to the equipment availability rate. For example, when the equipment failure rate ≥ 10%, k = 0.8. The construction machinery scheduling coefficient is dynamically adjusted according to the number of available equipment; E is the daily processing efficiency, which is predicted by the LSTM neural network, W is the engineering processing volume automatically measured by the three-dimensional point cloud model, ‌ is the safety redundancy time. It is worth mentioning that the above scheme lacks a dynamic adjustment mechanism for linking the maintenance engineering disposal time plan with geological prediction. By combining the mechanical scheduling coefficient k and the LSTM prediction efficiency E, the above scheme can achieve a 30% improvement in the optimization rate of the engineering response time. In the above S1, the calculation formula for the vegetation canopy penetration data is: ; Where, represents the number of lidar point clouds in the surface layer, which characterizes the number of ground reflection points not blocked by vegetation; represents the total number of lidar scan point clouds, including all reflection points in the vegetation layer and the surface layer; represents the vegetation penetration rate percentage, which reflects the penetration ability of lidar to the vegetation canopy. The lower the value, the denser the vegetation coverage. It is worth mentioning that based on the lack of a quantitative index for the root soil reinforcement strength in the current geological disaster assessment in vegetated areas, the point cloud layering algorithm in this design quantifies the vegetation penetration rate and accurately identifies areas with weak vegetation soil reinforcement ability; the improved TCARI / OSAVI model controls the soil moisture inversion error within 5%, improving the accuracy of hydrological parameters. In the above S2, the space resection and densification algorithm uses SIFT-GPU to accelerate feature matching, and the trigger condition for updating the dynamic DEM model is to start real-time model reconstruction when the daily slope change is more than 0.5 degrees. It is worth mentioning that based on the lag of traditional three-dimensional modeling technology in updating and being unable to reflect the dynamic changes of the terrain in real time; the SIFT-GPU accelerated feature matching technology improves the model reconstruction speed by more than 3 times, and the dynamic DEM trigger mechanism starts when the slope change ≥ 0.5°, which can ensure the timeliness of the model, with an error ≤ 2 cm. The method for calculating the terrain undulation is as follows: divide the DEM data into 10mx10m grids, calculate the maximum and minimum elevation values within the grid, and the difference between the maximum and minimum elevation values is the terrain undulation. When the terrain undulation is above 15m, a terrain mutation warning is triggered. It is worth mentioning that: traditional identification of terrain mutation areas relies on manual experience and judgment, and lacks a standardized evaluation system; the gridded elevation range calculation provided in this application can realize the automatic quantification of terrain undulation, and the accuracy of triggering threshold warnings is increased by 40%. ‌‌ The improved random forest algorithm includes a dynamic correction module for feature importance, which automatically triggers model retraining when the weight of the slope factor S changes by more than 10% in three consecutive iterations; the identification of boulder collapse adopts the DBSCAN density clustering algorithm, with the clustering radius set to 0.5m and the minimum point threshold to 200 points. It is worth mentioning that: based on the solidified feature weights of the traditional random forest algorithm, it cannot adapt to the dynamic geological environment; the dynamic correction module for feature importance provided in this application can increase the model prediction accuracy by 15% when the weight change is ≥10% and triggers retraining, while the DBSCAN density clustering algorithm can achieve a boulder volume recognition accuracy of ≥95%. ‌‌‌ A detection system, applied to any one of the above-mentioned methods for detecting geological hazard points, comprising: Data acquisition and processing module: The data acquisition and processing module uses an image data acquisition device mounted on a drone and a multi-band lidar to synchronously collect vegetation canopy penetration data, multispectral soil moisture spatial distribution data, and lithology distribution data to generate orthophotos and three-dimensional point cloud data sets; Elevation topology generation module: This module uses an aerial triangulation algorithm to fuse orthophotos and 3D point cloud data to build a 3D model. It then loads multispectral soil moisture spatial distribution data into the 3D model to generate a dynamic digital elevation model. It then uses a TIN (Tin triangulation) algorithm to generate contour topology. A geological disaster hazard system construction module: extracts slope factors, terrain relief, vegetation coverage, fracture distance and lithology factors based on the three-dimensional model to construct a geological disaster hazard system; Weight allocation module: uses an improved random forest algorithm to assign weights to hidden danger factors, inputs real-time meteorological data and a historical disaster case database, and inputs the probability of rock collapse and natural slope collapse; Engineering time calculation module: Calculates the disaster occurrence time based on the hidden danger probability, rock stability coefficient and dynamic rainfall factor, and generates the corresponding latest project start time according to the project processing volume, equipment efficiency and safety redundancy time, ensuring that the latest project start time is less than the disaster occurrence time minus the safety redundancy time.

[0024] Also includes:‌‌‌‌ Priority intelligent sorting module: Classify potential hazard points into red warning, orange warning, and yellow warning. When the time of disaster occurrence minus the latest start time is less than 15 days, a red warning is issued; when the time of disaster occurrence minus the latest start time is between 15 and 30 days, an orange warning is issued; when the time of disaster occurrence minus the latest start time is more than 30 days, a yellow warning is issued. 3D visualization module: Overlay and display the heat map of potential hazard points, the engineering treatment progress area, and the simulated line of predicted collapse trajectory in the real scene model.

[0025] It is worth mentioning that the above classification warning method can effectively remind the construction team to speed up the construction period and timely handle the potential hazard points of geological disasters before the arrival of geological disasters.

[0026] It also includes: Edge computing node module: Equipped with an NVIDIA Jetson AGX Xavier processor to process point cloud data in real time. Cloud analysis platform: Adopts a microservices architecture, including model construction services, potential hazard analysis services, and warning push services. IoT engineering interface: Supports direct CAN bus connection with excavators and anchoring equipment.

[0027] In the above embodiments, the device components involved are all conventional device components unless otherwise specified. The connection methods and control methods involved are all conventional connection methods and control methods unless otherwise specified.

[0028] The above has described the present invention in detail in combination with the embodiments. However, those skilled in the art can understand that without departing from the purpose of the present invention, various specific parameters in the above embodiments can be changed to form multiple specific embodiments, which are all within the common change range of the present invention and will not be elaborated here one by one.

Claims

1. A method for detecting potential geological hazard points, characterized in that, Including: S1: Using an unmanned aerial vehicle (UAV) equipped with an image data acquisition device and a multi-band lidar to synchronously collect vegetation canopy penetration data, multi-spectral soil moisture spatial distribution data, and lithology distribution data, and generating an orthophoto image and a three-dimensional point cloud dataset; S2: Establishing a three-dimensional model by fusing the orthophoto image and the three-dimensional point cloud data through an aerial triangulation encryption algorithm, loading the multi-spectral soil moisture spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generating a contour topology structure through the TIN triangulation algorithm; S3: Extracting slope factors, terrain undulation, vegetation coverage, fracture distance, and lithology factors based on the three-dimensional model to construct a geological disaster hidden danger system; S4: Using an improved random forest algorithm to assign weights to the hidden danger factors, inputting real-time meteorological data and a historical disaster case database, and inputting the occurrence probabilities of boulder collapses and natural slope collapses; S5: Calculating the disaster occurrence time based on the hidden danger probability, lithology stability coefficient, and dynamic rainfall factor, and generating the corresponding latest project start time according to the project treatment volume, equipment efficiency, and safety redundancy time, ensuring that the latest project start time is less than the disaster occurrence time minus the safety redundancy time.

2. The geological hazard potential point detection method according to claim 1, characterized in that The calculation formula for the disaster occurrence time is: ； Among them, P is the probability of collapse at the hidden danger point, and T1 is the disaster occurrence time. is the lithology compressive strength coefficient. is the correction coefficient of the rock mass fracture density. When the fracture density increases by 5 fractures / m the value decreases by 0.

1. is the dynamic rainfall influence factor. When the rainfall is more than 50 mm for 3 consecutive days the value increases by 0.

3.

3. The geological hazard potential point detection method according to claim 2, characterized in that The calculation formula for the latest start time is: ； Wherein: T2 is the latest start time, k is the construction machinery scheduling coefficient, and the construction machinery scheduling coefficient is dynamically adjusted according to the number of available equipment; E is the daily processing efficiency, and the daily processing efficiency is predicted by an LSTM neural network, and W is the engineering processing volume, is the safety redundancy time.

4. The geological hazard potential point detection method according to claim 1, characterized in that In S1, the calculation formula for the vegetation canopy penetration data is: ； Among them, represents the number of lidar point clouds in the surface layer, characterizing the number of ground reflection points not blocked by vegetation; represents the total number of lidar scan point clouds, including all reflection points in the vegetation layer and the surface layer; represents the vegetation penetration rate percentage, reflecting the penetration ability of lidar through the vegetation canopy. The lower the value, the denser the vegetation coverage.

5. A method for detecting potential geological hazard points according to claim 1, characterized in that, In S2, the aerial triangulation encryption algorithm uses SIFT-GPU to accelerate feature matching, and the dynamic DEM model update trigger condition is to start real-time model reconstruction when the single-day slope change is more than 0.5 degrees.

6. The geological hazard potential point detection method according to claim 1, characterized in that The calculation method for the terrain undulation is: Dividing the DEM data into 10m x 10m grids, calculating the maximum elevation value and the minimum elevation value within the grid, and the difference between the maximum elevation value and the minimum elevation value is the terrain undulation. When the terrain undulation is more than 15m, a terrain mutation warning is triggered.

7. A geological hazard potential point detection method according to claim 1, characterized in that, The improved random forest algorithm includes a feature importance dynamic correction module. When the weight of the slope factor S changes by more than 10% in three consecutive iterations, the model is automatically triggered for retraining; the identification of boulder collapses uses the DBSCAN density clustering algorithm, with the clustering radius set to 0.5m and the minimum number of points threshold set to 200 points.

8. A detection system is applied to a geological hazard potential point detection method as described in any one of claims 1-7, characterized in that, Including: Data acquisition and processing module: The data acquisition and processing module uses an unmanned aerial vehicle (UAV) equipped with an image data acquisition device and a multi-band lidar to synchronously collect vegetation canopy penetration data, multi-spectral soil moisture spatial distribution data, and lithology distribution data, and generating an orthophoto image and a three-dimensional point cloud dataset; High-line topology structure generation module: Establishing a three-dimensional model by fusing the orthophoto image and the three-dimensional point cloud data through an aerial triangulation encryption algorithm, loading the multi-spectral soil moisture spatial distribution data into the three-dimensional model to generate a dynamic digital elevation model, and generating a contour topology structure through the TIN triangulation algorithm; Geological disaster hidden danger system construction module: Extracting slope factors, terrain undulation, vegetation coverage, fracture distance, and lithology factors based on the three-dimensional model to construct a geological disaster hidden danger system; Weight assignment module: Using an improved random forest algorithm to assign weights to the hidden danger factors, inputting real-time meteorological data and a historical disaster case database, and inputting the occurrence probabilities of boulder collapses and natural slope collapses; Engineering time calculation module: calculates the disaster occurrence time based on the hidden danger probability, lithological stability coefficient, and dynamic rainfall factor, and generates the corresponding latest project start time according to the engineering treatment volume, equipment efficiency, and safety redundancy time, ensuring that the latest project start time is less than the disaster occurrence time minus the safety redundancy time.

9. A detection system according to claim 8, characterized in that, It also includes: ‌‌‌‌ Priority intelligent sorting module: divides the hidden danger points into red warnings, orange warnings, and yellow warnings. When the disaster occurrence time minus the latest start time is less than 15 days, a red warning is issued. An orange warning is issued when the disaster occurrence time minus the latest start time is between 15 and 30 days. A yellow warning is issued when the disaster occurrence time minus the latest start time is more than 30 days. Three-dimensional visualization module: superimposes and displays the hidden danger point heat map, engineering treatment progress area, and predicted collapse trajectory simulation line in the real scene model.

10. A detection system according to claim 9, characterized in that, It also includes: Edge computing node module: equipped with an NVIDIA Jetson AGX Xavier processor to process point cloud data in real time. Cloud analysis platform: adopts a microservices architecture, including model building services, hidden danger analysis services, and warning push services. IoT engineering interface: supports direct CAN bus connection with excavators and anchoring equipment.

Images