Slope landslide risk early warning method and system based on artificial intelligence

Through an artificial intelligence-based method, using three-dimensional slope images and historical landslide data, a landslide risk distribution map is constructed, which solves the problems of low efficiency and strong subjectivity of slope landslide risk warning in the existing technology, and achieves high-precision risk assessment and quick warning.

CN118736800BActive Publication Date: 2024-11-12CHENGDU UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411225656.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-11-12
Estimated Expiration
2044-09-03

AI Technical Summary

Technical Problem

The existing technology has low efficiency, strong subjectivity, and lag in the early warning of slope landslide risk warning. It lacks a systematic risk assessment framework, making it difficult to meet the needs of modern engineering safety management.

Method used

Using an artificial intelligence-based method, a landslide risk distribution map is constructed by obtaining the historical trigger path and landslide linkage values ​​of historical landslide nodes in the three-dimensional image of the slope, a landslide risk thermal distribution map is generated, and risk assessment and early warning are carried out in combination with the basic distribution map.

Benefits of technology

It improves the scientificity and accuracy of landslide risk assessment, quickly generates intuitive risk distribution maps, identify high-risk areas, ensures targeted and timely warnings, and reduces the probability and losses of landslide disasters.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118736800B_ABST
    Figure CN118736800B_ABST
Patent Text Reader

Abstract

The present application provides a slope landslide risk warning method and system based on artificial intelligence. By accurately obtaining the trigger path and landslide linkage value of historical landslide nodes, it can deeply reveal the internal stress state and landslide mechanism of the slope, and then accurately evaluate the potential sliding hazard of each slope block. In addition, the risk thermal distribution map and risk distribution map generated in combination with the basic distribution map make the spatial distribution of landslide risk clear at a glance, facilitate rapid identification of high-risk areas, and provide the possibility for timely prevention measures. Ultimately, the warning information generated based on the risk distribution map ensures the pertinence and timeliness of the warning, effectively reduces the probability of slope landslide disasters and disaster losses, and protects the safety of people’s lives and property.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an artificial intelligence-based slope landslide risk early warning method and system. Background Art

[0002] As a common geological disaster, slope landslides pose a huge threat to transportation, buildings, farmland, etc., and seriously affect the safety of people's lives and property and the ecological environment. Traditional slope landslide risk warning methods mainly rely on manual inspections and qualitative analysis, which have problems such as low efficiency, strong subjectivity, and delayed warning, and are difficult to meet the needs of modern engineering safety management.

[0003] In recent years, with the rapid development of remote sensing technology and three-dimensional imaging technology, new possibilities have been provided for accurate early warning of slope landslide risks. In particular, the application of slope three-dimensional imaging technology has enabled high-precision acquisition of information such as slope topography and geological structure, providing basic data support for the quantitative assessment of landslide risks. However, how to effectively use these three-dimensional imaging data and combine the lessons learned from historical landslide events to conduct early warning of slope landslide risks is still an urgent problem to be solved.

[0004] At present, some studies have attempted to apply artificial intelligence algorithms to the prediction and early warning of slope landslide risks, but most of them focus on the consideration of a single factor and lack a systematic risk assessment framework. In addition, research on the spatial distribution characteristics of slope landslide risks, in-depth understanding of landslide triggering mechanisms, and the mutual influence of landslide risks in different regions still needs to be strengthened. Summary of the invention

[0005] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of the present application is to provide a slope landslide risk early warning method and system based on artificial intelligence.

[0006] In a first aspect, the present application provides a slope landslide risk early warning method based on artificial intelligence, the method comprising:

[0007] Obtain the historical landslide triggering path of each historical landslide node in the three-dimensional image of the slope in the target slope monitoring area in the landslide risk plane space, and obtain the landslide linkage value from each of the historical landslide nodes to the risk observation plane in the landslide risk three-dimensional space;

[0008] Obtaining a basic distribution map for constructing a slope landslide risk distribution map;

[0009] Determining the potential sliding hazard of each first slope block of the three-dimensional image of the slope according to the historical landslide triggering path and the landslide linkage value of each historical landslide node;

[0010] According to the potential sliding danger of each of the first slope blocks, construct each of the first slope blocks in the basic distribution map to generate a landslide risk thermal distribution map;

[0011] According to the potential sliding danger of each of the first slope blocks, a second slope block is constructed within the distribution range except for each of the first slope blocks in the landslide risk thermal distribution map, and a slope landslide risk distribution map of the target slope monitoring area is generated;

[0012] Corresponding warning information is generated based on the slope landslide risk distribution map of the target slope monitoring area.

[0013] In a possible implementation manner of the first aspect, obtaining a basic distribution map for constructing a slope landslide risk distribution map includes:

[0014] Obtain the target accuracy of the slope landslide risk distribution map to be constructed;

[0015] A basic distribution map with the target accuracy for constructing a slope landslide risk distribution map is generated.

[0016] In a possible implementation manner of the first aspect, the three-dimensional image of the slope is composed of a plurality of landslide risk blocks, and the two defined positions of the landslide risk blocks are respectively a first historical landslide node and a second historical landslide node in the three-dimensional image of the slope;

[0017] Determining the potential sliding hazard of each first slope block of the three-dimensional slope image according to the historical landslide triggering path and the landslide linkage value of each historical landslide node includes:

[0018] For each risk point on each landslide risk block in the three-dimensional image of the slope, the following steps are performed respectively to generate the potential sliding hazard of each first slope block in the three-dimensional image of the slope:

[0019] Obtaining spatial position parameters of the risk point in the landslide risk plane space;

[0020] Inputting the spatial position parameter and the historical landslide triggering path of the first historical landslide node into a pre-trained landslide linkage decision model to determine a first landslide linkage value between the risk point and the first historical landslide node;

[0021] Inputting the spatial position parameter and the historical landslide triggering path of the second historical landslide node into a pre-trained landslide linkage decision model to determine a second landslide linkage value between the risk point and the second historical landslide node;

[0022] using the first landslide linkage value as a first weighting coefficient of the landslide linkage value from the first historical landslide node to the risk observation plane, and using the second landslide linkage value as a second weighting coefficient of the landslide linkage value from the second historical landslide node to the risk observation plane;

[0023] According to the first weighting coefficient and the second weighting coefficient, a fusion calculation is performed on the landslide linkage value from the first historical landslide node to the risk observation plane and the landslide linkage value from the second historical landslide node to the risk observation plane to generate a fusion calculation result;

[0024] The fusion calculation result is used as the potential sliding hazard of the first slope block of the risk point.

[0025] In a possible implementation of the first aspect, the three-dimensional image of the slope is composed of a plurality of landslide risk blocks, and the three-dimensional image of the slope includes: a composite risk block generated by combining and connecting at least three landslide risk blocks associated with the historical landslide node, and the focus node of the composite risk block is the historical landslide node;

[0026] Determining the potential sliding hazard of each first slope block of the three-dimensional slope image according to the historical landslide triggering path and the landslide linkage value of each historical landslide node includes:

[0027] The composite risk block is divided into at least one multivariate risk block, and for each of the multivariate risk blocks, the following steps are performed to generate the potential slip hazard of each of the first slope blocks of the composite risk block of the slope three-dimensional image:

[0028] The historical landslide triggering paths and the landslide linkage values ​​of each of the focus nodes in the multivariate risk block are input into a pre-trained potential sliding hazard decision model to determine the potential sliding hazard of each of the first slope blocks in the multivariate risk block of the slope three-dimensional image.

[0029] In a possible implementation of the first aspect, the historical landslide triggering paths of the respective focus nodes of the multivariate risk block are characterized by historical landslide triggering path parameters; the historical landslide triggering paths and the landslide linkage values ​​of the respective focus nodes of the multivariate risk block are input into a pre-trained potential sliding hazard decision model to determine the potential sliding hazard of each of the first slope blocks of the multivariate risk block of the three-dimensional image of the slope, including:

[0030] For each of the first slope blocks of the multi-risk blocks, the following steps are performed respectively:

[0031] Obtaining a slope block distribution position of the first slope block in the basic distribution map;

[0032] According to the distribution position of the slope blocks and the historical landslide triggering path parameters of each of the focus nodes of the multi-risk block, the weighting coefficient of each of the focus nodes corresponding to the first slope block is determined, and the weighting coefficient is the weighting coefficient corresponding to the landslide linkage value from the focus node to the risk observation plane;

[0033] According to the weighted coefficients of the first slope block corresponding to the respective focus nodes, a fusion calculation is performed on the landslide linkage values ​​from the respective focus nodes to the risk observation plane to generate a fusion calculation result;

[0034] Using the fusion calculation result as the potential sliding hazard of the first slope block of the multi-risk block;

[0035] Wherein, according to the potential sliding danger of each first slope block, each first slope block is constructed in the basic distribution map to generate a landslide risk thermal distribution map, including:

[0036] For the potential sliding danger of each of the first slope blocks in the multi-risk blocks, the following steps are performed respectively to generate the landslide risk thermodynamic distribution map:

[0037] When the weighted coefficients of the first slope block corresponding to the respective focus nodes are not less than 0, the first slope block is constructed in the basic distribution map according to the potential slip hazard of the first slope block.

[0038] In a possible implementation of the first aspect, constructing each of the first slope blocks in the basic distribution map according to the potential sliding hazard of each of the first slope blocks to generate a landslide risk thermal distribution map includes:

[0039] For each of the first slope blocks, the following steps are performed respectively to generate the landslide risk thermal distribution map:

[0040] Determine a distribution point to be constructed of the first slope block in the basic distribution map, and determine a target slope block located at the distribution point to be constructed from the basic distribution map;

[0041] The potential sliding hazard of the target slope block is output as the potential sliding hazard of the first slope block, so as to construct the first slope block in the basic distribution map.

[0042] In a possible implementation manner of the first aspect, before outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block, the method further includes:

[0043] When no other slope blocks are constructed at the target slope block, the step of outputting the potential sliding hazard of the target slope block as the potential sliding hazard of the first slope block is performed;

[0044] When other slope blocks have been constructed at the target slope block, obtaining target potential sliding hazards of the other slope blocks constructed at the target slope block;

[0045] When the target potential slip hazard is greater than the potential slip hazard of the first slope block, the step of outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block is executed;

[0046] The method further comprises:

[0047] When the target potential slip hazard is not greater than the potential slip hazard of the first slope block, the target potential slip hazard is maintained stable.

[0048] In a possible implementation of the first aspect, based on the potential sliding hazard of each of the first slope blocks, in the landslide risk thermal distribution map except for each of the first slope blocks, constructing a second slope block to generate a slope landslide risk distribution map of the target slope monitoring area includes:

[0049] The potential sliding hazard of each of the first slope blocks is transferred to each target slope block within the distribution range other than each of the first slope blocks in the landslide risk thermal distribution map to generate the potential sliding hazard of each of the target slope blocks;

[0050] According to the potential sliding hazard of each target slope block, each target slope block is constructed in the landslide risk thermal distribution map, and the constructed target slope block is used as the second slope block to generate a slope landslide risk distribution map of the target slope monitoring area.

[0051] In a possible implementation manner of the first aspect, the risk transfer includes X rounds of risk transfer, wherein the risk transfer of the potential sliding hazard of each of the first slope blocks to each target slope block within a distribution range other than each of the first slope blocks in the landslide risk thermal distribution map is performed to generate the potential sliding hazard of each of the target slope blocks, including:

[0052] For the first round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of each of the first slope blocks is transferred to the connected slope blocks of each of the first slope blocks to generate the potential sliding hazard of the connected slope blocks of each of the first slope blocks, and the potential sliding hazard of the connected slope blocks of each of the first slope blocks is used as the potential sliding hazard of the first round target slope block;

[0053] For the mth round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of the target slope block of the m-1th round is risk-transferred to the connected slope blocks of the target slope block of the m-1th round, the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is generated, and the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is used as the potential sliding hazard of the target slope block of the m-1th round;

[0054] Wherein, X and m are integers greater than 0, and X is not less than m;

[0055] The step of transferring the potential sliding hazards of each of the first slope blocks to the connected slope blocks of each of the first slope blocks to generate the potential sliding hazards of the connected slope blocks of each of the first slope blocks includes:

[0056] For each connected slope block to which the potential slip hazard of the first slope block is transferred, at least one target first slope block to which the risk is transferred is determined, and a fusion calculation is performed on the potential slip hazard of the at least one target first slope block to generate the potential slip hazard of the connected slope block.

[0057] In the second aspect, an embodiment of the present application further provides an artificial intelligence-based slope landslide risk warning system, the artificial intelligence-based slope landslide risk warning system comprising a processor and a machine-readable storage medium, the machine-readable storage medium storing a computer program, the computer program is loaded and executed in conjunction with the processor to implement the artificial intelligence-based slope landslide risk warning method of the first aspect above.

[0058] By adopting any of the above technical solutions, through the comprehensive use of slope three-dimensional images, historical landslide data and artificial intelligence algorithms, accurate early warning of slope landslide risks is achieved. This method not only improves the scientificity and accuracy of landslide risk assessment, but also can quickly generate intuitive landslide risk thermal distribution maps and comprehensive slope landslide risk distribution maps, providing strong decision-making support for relevant departments. Specifically, by accurately obtaining the trigger path and landslide linkage value of historical landslide nodes, this method can deeply reveal the internal stress state and landslide mechanism of the slope, and then accurately evaluate the potential sliding hazard of each slope block. In addition, the risk thermal distribution map and risk distribution map generated in combination with the basic distribution map make the spatial distribution of landslide risks clear at a glance, facilitate the rapid identification of high-risk areas, and provide the possibility of taking preventive measures in a timely manner. Finally, the early warning information generated based on the risk distribution map ensures the pertinence and timeliness of the early warning, effectively reduces the probability of slope landslide disasters and disaster losses, and protects the safety of people's lives and property. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained by combining these drawings without paying creative work.

[0060] Figure 1 A schematic diagram of a process flow of a slope landslide risk early warning method based on artificial intelligence provided in an embodiment of the present application;

[0061] Figure 2 A functional structure schematic block diagram of an artificial intelligence-based landslide risk warning system for implementing the artificial intelligence-based landslide risk warning method described above is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The following description is intended to enable one of ordinary skill in the art to implement and combine the present application, and the description is provided in the context of a specific application scenario and its requirements. It will be apparent to one of ordinary skill in the art that various changes may be made to the disclosed embodiments, and that the general principles defined in the present application may be applied to other embodiments and application scenarios without departing from the principles and scope of the present application. Therefore, the present application is not limited to the described embodiments, but should be given the broadest scope consistent with the claims.

[0063] See also Figure 1 As shown, the present application provides an artificial intelligence-based slope landslide risk warning method, which includes the following steps.

[0064] Step S110, obtaining the historical landslide triggering paths of each historical landslide node in the three-dimensional image of the slope in the target slope monitoring area in the landslide risk plane space, and obtaining the landslide linkage values ​​from each of the historical landslide nodes to the risk observation plane in the landslide risk three-dimensional space.

[0065] In detail, the three-dimensional image of the slope refers to a three-dimensional digital model of the slope area obtained by remote sensing technology, drone photogrammetry, laser radar scanning (LiDAR) or other three-dimensional imaging technology. This image provides detailed three-dimensional structural information of the slope, including topography, landforms, geological structures, etc. For example, in a mountain slope monitoring project, a drone carrying a high-resolution camera is used to take multi-angle photos of the slope, and then these two-dimensional images are spliced ​​into a three-dimensional model through image processing software. This model can clearly show the slope's slope, rock stratification, soil structure and other characteristics.

[0066] The historical landslide node refers to the starting point or key point of the landslide in the historical landslide event of the slope. These historical landslide nodes are usually located in areas with unstable geological conditions and susceptible to external factors (such as rainfall and earthquakes). For example, in a historical landslide event, the landslide started from a severely weathered rock layer at the top of the mountain and gradually expanded downward. This severely weathered rock layer can be regarded as a historical landslide node.

[0067] The landslide risk plane space refers to the projection of the slope on a two-dimensional plane, which is used to analyze the extension path and range of the landslide on the horizontal plane. For example, in the three-dimensional image of the slope, a suitable observation angle is selected to project the slope onto a horizontal plane to form a landslide risk plane map. By analyzing this plane map, the possible extension direction and range of the landslide can be identified.

[0068] The historical landslide trigger path refers to the trajectory of the landslide body sliding down along a specific path when a historical landslide event occurs. This path is affected by many factors, including terrain slope, geological structure, rainfall intensity, etc. For example, in a landslide event, the landslide body slides down from the top of the mountain along a specific inclined surface, passing through broken rock layers and soft soil layers, and finally reaches the foot of the slope. This sliding trajectory is the historical landslide trigger path.

[0069] The landslide linkage value refers to the size or degree of impact of a historical landslide node on other areas of the slope when a landslide occurs. This landslide linkage value is usually obtained through complex mathematical models and geomechanical analysis, taking into account multiple factors such as geological conditions, rainfall, and seismic activity. For example, assuming that the landslide linkage value of a historical landslide node is 0.8, it means that when a landslide occurs at this node, it is likely to cause landslides or deformations in the surrounding areas with a linkage strength of 0.8.

[0070] In this embodiment, it is assumed that the server is processing a target slope monitoring area located in a mountainous area. The three-dimensional image of the slope of the target slope monitoring area has been obtained through advanced geological monitoring equipment and technical means. In this three-dimensional image of the slope, the server identifies multiple historical landslide nodes, such as node A, node B and node C.

[0071] For node A, its historical landslide triggering path in the landslide risk plane space may start from a certain position on the top of the mountain, gradually extend downward along a specific inclined surface, and pass through some specific rock layers and soil structures. The characteristics of this historical landslide triggering path include factors such as slope, material composition of the slope, and geological structure.

[0072] At the same time, in the landslide risk three-dimensional space, the landslide linkage value from node A to the risk observation plane (for example, a horizontal plane set at the bottom of the slope) is obtained through complex calculations. This value comprehensively considers the impact of multiple factors such as the geological conditions, rainfall, and seismic activity at the location of node A on the possibility of landslides. For example, the landslide linkage value of node A is calculated to be 0.8 (the value range is 0-1, and the larger the value, the higher the landslide risk).

[0073] For nodes B and C, similar methods are used to obtain their historical landslide triggering paths and landslide linkage values. The historical landslide triggering path of node B may start from an unstable rock structure on the hillside and pass through some broken rock layers; its landslide linkage value is 0.6. The historical landslide triggering path of node C may start from a severely weathered area on the top of the slope and pass through some soft soil layers; its landslide linkage value is 0.7.

[0074] Step S120, obtaining a basic distribution map for constructing a slope landslide risk distribution map.

[0075] In detail, the basic distribution map refers to an initial grid map or blank map used to construct a landslide risk distribution map. This basic distribution map is divided into grids according to certain accuracy requirements, and each grid represents a part of the slope area. For example, before constructing a landslide risk distribution map, a grid map is first generated according to the target accuracy requirement (such as per square meter). Assuming that the total area of ​​the slope area is 1,000 square meters, a basic distribution map containing 1,000 grids is generated.

[0076] That is, in this embodiment, the server first obtains the target accuracy requirement of the slope landslide risk distribution map to be constructed. Assume that the target accuracy requirement is to be able to accurately assess the risk per square meter. Then, the server generates a corresponding basic distribution map based on this target accuracy. This basic distribution map can be a blank grid map, and each grid represents one square meter of the target slope monitoring area. For example, if the area of ​​the target slope monitoring area is 1,000 square meters, then the generated basic distribution map is a blank map of 1,000 grids. Each grid has its corresponding coordinates and identifiers, so that slope blocks can be constructed and risk information can be annotated on it later.

[0077] Step S130, determining the potential sliding hazard of each first slope block in the three-dimensional image of the slope based on the historical landslide trigger path and the landslide linkage value of each historical landslide node, and constructing each first slope block in the basic distribution map based on the potential sliding hazard of each first slope block to generate a landslide risk thermal distribution map.

[0078] In detail, the first slope block refers to the slope area preliminarily divided according to geological conditions, historical landslide events and other factors in the three-dimensional image of the slope. These first slope blocks are the basic units for further analysis of potential sliding hazards. For example, in the three-dimensional image of the slope, the slope is divided into several blocks according to factors such as terrain changes, geological structures, and historical landslide nodes. Each block contains specific geological conditions and potential landslide risks. The potential sliding hazard refers to the possibility of landslides in the slope block under specific conditions and the possible consequences. This value is calculated by comprehensively considering multiple factors (such as geological conditions, rainfall, seismic activity, etc.). For example, through the landslide linkage decision model and other geomechanical analysis methods, the potential sliding hazard of a certain slope block is calculated to be 0.72. This means that under certain conditions, there is a high probability of landslides in the block, which may cause certain damage and losses.

[0079] That is, in this embodiment, it is assumed that the three-dimensional image of the slope is composed of multiple landslide risk blocks, such as block 1, block 2 and block 3. For block 1, the server first obtains the spatial position parameters of each risk point inside it. Assume that the risk point P is located in block 1. The spatial position parameters of the risk point P and the historical landslide trigger path of node A are input into the pre-trained landslide linkage decision model. After analysis and calculation, the model determines that the first landslide linkage value between the risk point P and the node A is 0.7. Similarly, the spatial position parameters of the risk point P and the historical landslide trigger path of node B are input into the model to determine that the second landslide linkage value between the risk point P and the node B is 0.4. Then, the first landslide linkage value 0.7 is used as the first weighting coefficient of the landslide linkage value 0.8 from node A to the risk observation plane, and the second landslide linkage value 0.4 is used as the second weighting coefficient of the landslide linkage value 0.6 from node B to the risk observation plane.

[0080] Through weighted calculation, i.e., 0.8×0.7+0.6×0.4=0.72, a fusion calculation result of 0.72 is generated. This result 0.72 is used as the potential slip hazard of the first slope block 1 where the risk point P is located.

[0081] The same method is used to calculate the potential slip hazard of Block 2 and Block 3. Assume that the calculation result of the potential slip hazard of Block 2 is 0.65, and the calculation result of the potential slip hazard of Block 3 is 0.85.

[0082] Next, each first slope block is constructed in the basic distribution map according to the potential slip hazard of each first slope block. For example, for block 1, the server determines its distribution point to be constructed in the basic distribution map, assuming that it is the area with coordinates (10,20) to (20,30). From the basic distribution map, the target slope block located in this area is determined. If no other slope blocks are constructed at the target slope block, the server outputs the calculated potential slip hazard of 0.72 to the target slope block, thereby constructing block 1 in the basic distribution map.

[0083] If other slope blocks have been constructed at the target slope block, such as block 4, and the target potential slip hazard of block 4 is 0.6, since 0.72 is greater than 0.6, the server outputs 0.72 as the potential slip hazard of block 1, overwriting the original 0.6.

[0084] Block 2 and Block 3 are constructed in the same way in the basic distribution map.

[0085] After all the first slope blocks are constructed, a landslide risk thermodynamic distribution map is generated. In this map, different potential sliding hazards are represented by different colors or brightness. For example, blocks with high potential sliding hazards are displayed in red with high brightness, and blocks with low potential sliding hazards are displayed in blue with low brightness.

[0086] Step S140, constructing second slope blocks within the distribution range of the landslide risk thermal distribution map except for the first slope blocks according to the potential sliding danger of each of the first slope blocks, and generating a slope landslide risk distribution map of the target slope monitoring area.

[0087] In this embodiment, it is assumed that the first slope blocks include blocks 1, 2, and 3, and their potential sliding hazards are 0.72, 0.65, and 0.85, respectively. The server transfers the potential sliding hazards of each first slope block to each target slope block within the distribution range other than each first slope block in the landslide risk thermal distribution map.

[0088] Assume that three rounds of risk transfer are performed. In the first round of risk transfer, the potential slip risk of block 1 of 0.72 is transferred to its connected slope blocks, such as blocks 4 and 5. For block 4, the first target slope block to which the risk is transferred is determined to be block 1. The potential slip risk of block 1 of 0.72 is appropriately attenuated (for example, multiplied by an attenuation coefficient of 0.8), and the potential slip risk of block 4 is 0.576. Similarly, the potential slip risk of block 5 is calculated to be 0.6. After the first round of risk transfer is completed, the potential slip risk of the first round target slope blocks is determined to be 0.576 for block 4 and 0.6 for block 5. In the second round of risk transfer, the potential slip risk of the first round target slope block 4 of 0.576 is transferred to its connected slope blocks, such as block 6. After calculation, the potential slip risk of block 6 is 0.46. In the third round of risk transfer, no new target slope blocks are generated and the risk transfer ends.

[0089] According to the potential sliding hazard of each target slope block (such as block 4, block 5, block 6, etc.), each target slope block is constructed in the landslide risk thermal distribution map, and the constructed target slope block is used as the second slope block to finally generate a slope landslide risk distribution map of the target slope monitoring area.

[0090] Step S150: generating corresponding warning information based on the slope landslide risk distribution map of the target slope monitoring area.

[0091] In this embodiment, the server performs analysis based on the generated landslide risk distribution map. If the landslide risk distribution map shows that the potential slip risk of some areas is extremely high (for example, greater than 0.9), the server generates a high-level warning message to remind relevant departments to take emergency measures immediately, such as evacuating nearby residents and setting up warning signs.

[0092] If there are some areas with high potential slip risk (for example, between 0.7-0.9), a medium-level warning message is generated, and it is recommended to increase the monitoring frequency and prepare relevant emergency supplies. For areas with low potential slip risk (for example, less than 0.7), a low-level warning message is generated and normal monitoring is maintained. The warning information includes detailed information such as the specific location of the dangerous area, the value of the potential slip risk, and the recommended measures, so that relevant departments and personnel can respond in a timely and accurate manner.

[0093] Based on the above steps, the embodiment of the present application realizes accurate early warning of slope landslide risk by comprehensively utilizing slope three-dimensional images, historical landslide data and artificial intelligence algorithms. This method not only improves the scientificity and accuracy of landslide risk assessment, but also can quickly generate intuitive landslide risk thermal distribution maps and comprehensive slope landslide risk distribution maps, providing strong decision-making support for relevant departments. Specifically, by accurately obtaining the trigger path and landslide linkage value of historical landslide nodes, this method can deeply reveal the internal stress state and landslide mechanism of the slope, and then accurately evaluate the potential slip hazard of each slope block. In addition, the risk thermal distribution map and risk distribution map generated in combination with the basic distribution map make the spatial distribution of landslide risk clear at a glance, facilitate rapid identification of high-risk areas, and provide the possibility for timely prevention measures. Finally, the early warning information generated based on the risk distribution map ensures the pertinence and timeliness of the early warning, effectively reduces the probability of occurrence and disaster losses of slope landslide disasters, and protects the safety of people's lives and property.

[0094] In a possible implementation, step S120 includes:

[0095] Step S121, obtaining the target accuracy of the slope landslide risk distribution map to be constructed.

[0096] Step S122, generating a basic distribution map with the target accuracy for constructing a slope landslide risk distribution map.

[0097] In this embodiment, the server receives a task to construct a landslide risk distribution map for a specific slope area. First, the server obtains the target accuracy of the slope landslide risk distribution map to be constructed. For example, the relevant department requires that the target accuracy of the slope landslide risk distribution map to be constructed this time is to be able to accurately evaluate the area every 5 square meters. After obtaining this target accuracy requirement, the server begins to generate a basic distribution map. The server divides the slope area into grids of corresponding sizes according to the target accuracy. For this example, the server divides the slope area into a grid every 5 square meters. Assuming that the total area of ​​this slope area is 500 square meters, the basic distribution map generated by the server is a chart composed of 100 grids of 5 square meters in size. Each grid has its unique identifier and coordinates for subsequent data filling and processing. In the process of generating the basic distribution map, the server also establishes a corresponding data storage structure for each grid to record information such as potential slip hazards obtained by subsequent calculations and analysis.

[0098] Through the above steps, a basic distribution map with specified target accuracy for constructing the slope landslide risk distribution map is generated, which is ready for subsequent risk assessment and distribution map construction.

[0099] In a possible implementation manner, the three-dimensional image of the slope is composed of a plurality of landslide risk blocks, and two defined positions of the landslide risk blocks are respectively a first historical landslide node and a second historical landslide node in the three-dimensional image of the slope.

[0100] Step S130 includes:

[0101] For each risk point on each landslide risk block in the three-dimensional image of the slope, the following steps are performed respectively to generate the potential sliding hazard of each first slope block in the three-dimensional image of the slope:

[0102] Step S131, obtaining the spatial position parameters of the risk point in the landslide risk plane space.

[0103] Step S132: input the spatial position parameter and the historical landslide triggering path of the first historical landslide node into a pre-trained landslide linkage decision model to determine a first landslide linkage value between the risk point and the first historical landslide node.

[0104] Step S133: input the spatial position parameter and the historical landslide triggering path of the second historical landslide node into a pre-trained landslide linkage decision model to determine a second landslide linkage value between the risk point and the second historical landslide node.

[0105] Step S134: using the first landslide linkage value as a first weighting coefficient of the landslide linkage value from the first historical landslide node to the risk observation plane, and using the second landslide linkage value as a second weighting coefficient of the landslide linkage value from the second historical landslide node to the risk observation plane.

[0106] Step S135, performing a fusion calculation on the landslide linkage value from the first historical landslide node to the risk observation plane and the landslide linkage value from the second historical landslide node to the risk observation plane according to the first weighting coefficient and the second weighting coefficient, and generating a fusion calculation result.

[0107] Step S136: Using the fusion calculation result as the potential sliding hazard of the first slope block of the risk point.

[0108] In this embodiment, it is assumed that the server is processing a complex target slope monitoring area in a mountainous area, and the three-dimensional image of the slope in this area has been accurately acquired through high-precision geological monitoring equipment and advanced imaging technology. In this three-dimensional image of the slope, multiple landslide risk blocks, such as block Z, are determined.

[0109] The two defined locations of block Z are the first historical landslide node D and the second historical landslide node E.

[0110] There are multiple risk points in block Z, such as risk point Q.

[0111] The server first obtains the spatial position parameters of the risk point Q in the landslide risk plane space. These parameters describe in detail the coordinates of the risk point Q, the slope where it is located, the inclination direction of the slope, and other information.

[0112] Next, the server inputs the spatial location parameters of the risk point Q and the historical landslide trigger path of the first historical landslide node D into the pre-trained landslide linkage decision model. This model contains a large amount of data on landslide characteristics and laws. After complex calculations and analysis, the first landslide linkage value between the risk point Q and the first historical landslide node D is determined. Assume that this value is 0.6.

[0113] Then, the server inputs the spatial position parameters of the risk point Q and the historical landslide triggering path of the second historical landslide node E into the same landslide linkage decision model, and determines the second landslide linkage value between the risk point Q and the second historical landslide node E. Assume that this value is 0.4.

[0114] The server uses the first landslide linkage value 0.6 as the first weighting coefficient of the landslide linkage value 0.7 from the first historical landslide node D to the risk observation plane, and uses the second landslide linkage value 0.4 as the second weighting coefficient of the landslide linkage value 0.5 from the second historical landslide node E to the risk observation plane.

[0115] Next, the fusion calculation is performed based on these weighted coefficients. The specific calculation method is: 0.7×0.6+0.5×0.4=0.58.

[0116] The server uses the fusion calculation result 0.58 as the potential slip hazard of the first slope block Z where the risk point Q is located.

[0117] For other risk points in block Z, such as risk point R and risk point S, the server also follows the same steps.

[0118] For the risk point R, after obtaining its spatial position parameters, the historical landslide trigger paths of the first historical landslide node D and the second historical landslide node E are input into the landslide linkage decision model respectively, and the first landslide linkage value with node D is 0.7, and the second landslide linkage value with node E is 0.3.

[0119] The weighting coefficients are determined in the same way: the weighting coefficient of the first historical landslide node D is 0.7, and the weighting coefficient of the second historical landslide node E is 0.3.

[0120] Fusion calculation: 0.7×0.7+0.5×0.3=0.64, 0.64 is taken as the potential slip hazard of the first slope block Z where the risk point R is located.

[0121] For risk point S, it is assumed that the first landslide linkage value with node D is 0.5, and the second landslide linkage value with node E is 0.5.

[0122] The weighted coefficients are 0.5 respectively, and the fusion calculation is: 0.7×0.5+0.5×0.5=0.6, and 0.6 is taken as the potential slip hazard of the first slope block Z where the risk point S is located.

[0123] By performing such processing on each risk point within block Z, the server can determine the potential slip risk of the entire block Z.

[0124] For other landslide risk blocks, such as block X and block Y, the server also repeats the above steps.

[0125] In block X, assume there are risk points T, U, and V.

[0126] For the risk point T, the linkage values ​​calculated with the first historical landslide node F and the second historical landslide node G are 0.8 and 0.2 respectively. After weighted fusion calculation, the potential slip hazard is 0.7.

[0127] For the risk point U, the linkage values ​​are 0.4 and 0.6 respectively, and the potential slip hazard is 0.5 after weighted fusion calculation.

[0128] For the risk point V, the linkage values ​​are 0.7 and 0.3 respectively, and the potential slip hazard is 0.6 after weighted fusion calculation.

[0129] In block Y, it is assumed that the potential slip hazards calculated for risk points W, X, and Y are 0.8, 0.4, and 0.7, respectively.

[0130] The server performs such detailed and precise calculations on the risk points in all landslide risk blocks, and finally determines the potential sliding hazard of each first slope block in the three-dimensional image of the slope.

[0131] This calculation process fully considers the relationship between each risk point and different historical landslide nodes, as well as the landslide linkage value of each historical landslide node itself, so that the potential sliding hazard of each first slope block can be evaluated more accurately and comprehensively, providing an important data basis for subsequent slope landslide risk assessment and early warning.

[0132] Subsequently, the server can construct corresponding blocks in the basic distribution map according to these potential sliding hazards, generate a landslide risk thermal distribution map, and further transfer risks and construct a slope landslide risk distribution map, ultimately generating accurate and effective early warning information to protect the lives and property of residents in mountainous areas and the normal operation of related facilities.

[0133] In a possible implementation, the three-dimensional image of the slope is composed of multiple landslide risk blocks, and the three-dimensional image of the slope includes: a composite risk block generated by combining and connecting at least three landslide risk blocks associated with the historical landslide node, and the focus node of the composite risk block is the historical landslide node.

[0134] Step S130 includes:

[0135] Step A110, dividing the composite risk block into at least one multivariate risk block, and performing the following steps for each of the multivariate risk blocks to generate the potential slip hazard of each of the first slope blocks of the composite risk block of the slope three-dimensional image:

[0136] Step A120, inputting the historical landslide triggering paths and the landslide linkage values ​​of each of the focus nodes of the multivariate risk block into a pre-trained potential sliding hazard decision model to determine the potential sliding hazard of each of the first slope blocks of the multivariate risk block of the slope three-dimensional image.

[0137] Assume that the server is processing a large mountain slope monitoring area. The 3D image of the slope in the mountain slope monitoring area shows that there are multiple interrelated landslide risk blocks. Among them, there is a composite risk block formed by three landslide risk blocks (labeled as blocks A, B and C) connected to each other through historical landslide nodes. The nodes of interest of this composite risk block are historical landslide nodes H, I and J.

[0138] The server first divides the composite risk block into two multivariate risk blocks, which are marked as multivariate risk block 1 and multivariate risk block 2. For multivariate risk block 1, the nodes of interest it contains are historical landslide nodes H and I.

[0139] The server obtains the historical landslide triggering path of the historical landslide node H (for example, starting from a steep slope at the top of the mountain and passing through a specific rock crack) and its landslide linkage value (assuming it is 0.7), and also obtains the historical landslide triggering path of the historical landslide node I (for example, starting from a weathered rock layer on the mountainside and extending to the foot of the mountain) and its landslide linkage value (assuming it is 0.6).

[0140] Then, these historical landslide trigger paths and landslide linkage values ​​are input into the pre-trained potential slip hazard decision model. After calculation and analysis, the potential slip hazard decision model comprehensively considers multiple factors such as the location of node H and node I, the characteristics of the trigger path, the landslide linkage value, and the geological conditions in the multivariate risk block 1, and finally determines that the potential slip hazard of the first slope block corresponding to the multivariate risk block 1 is 0.65.

[0141] For multivariate risk block 2, the nodes of interest it contains are historical landslide nodes I and J.

[0142] The server also obtains the historical landslide triggering path and landslide linkage value of historical landslide node I (obtained previously), as well as the historical landslide triggering path of historical landslide node J (for example, starting from a soft soil area at the top of the slope and sliding down the slope) and its landslide linkage value (assumed to be 0.8).

[0143] After inputting these data into the potential sliding hazard decision model, through calculation and analysis of the model, it is determined that the potential sliding hazard of the first slope block corresponding to the multivariate risk block 2 is 0.72.

[0144] By dividing and analyzing the composite risk blocks in this way, the server can accurately determine the potential sliding hazard of the first slope block corresponding to each multi-risk block, providing important data support for the subsequent generation of accurate slope landslide risk distribution maps and early warning information.

[0145] The input data of the landslide linkage decision-making model mainly include:

[0146] Spatial location parameters of risk points: describe the three-dimensional coordinates or other location information of specific risk points on the slope.

[0147] Historical landslide triggering path of historical landslide node: This is the trajectory of the landslide body sliding down along a specific path when the landslide occurs, usually including the starting point of the landslide, path length, direction, slope change and other information.

[0148] The training process of the landslide linkage decision-making model includes:

[0149] First, the landslide trigger path data and the corresponding risk point location data are collected from historical landslide events. Then, useful features such as path length, slope change rate, geological condition changes, etc. are extracted from the landslide trigger path data. At the same time, the spatial location parameters of the risk point are converted into a format acceptable to the model. A suitable machine learning or deep learning model is selected according to the characteristics of the problem. Since the landslide linkage value is usually closely related to the characteristics of the landslide trigger path and the location of the risk point, a neural network (such as a fully connected network, a convolutional neural network, etc.) can be used to capture this relationship. The extracted features and the corresponding landslide linkage values ​​are input into the model as training data, and the model parameters are adjusted by the back propagation algorithm to minimize the error between the predicted landslide linkage value and the actual landslide linkage value. Finally, the model performance is evaluated using the validation set, and the model structure and hyperparameters are adjusted according to the evaluation results until satisfactory performance is achieved.

[0150] Furthermore, the input data of the potential slip risk decision model mainly includes:

[0151] Historical landslide triggering paths of focus nodes in the multivariate risk block: This is the landslide triggering path corresponding to each focus node in the multivariate risk block when a landslide occurs.

[0152] Landslide linkage value: describes the impact of each focus node on other areas when a landslide occurs.

[0153] The training process of the potential slip hazard decision model includes:

[0154] Collect the landslide triggering paths and corresponding landslide linkage values ​​of the nodes of interest in the multivariate risk block from historical landslide events. Perform preprocessing operations such as cleaning and standardization on the data. Extract useful features from the landslide triggering paths and consider the landslide linkage value as one of the input features. At the same time, it may be necessary to divide the multivariate risk block into smaller units (such as grids or blocks) so that each unit can be evaluated for potential sliding hazard separately. Select a suitable machine learning or deep learning model to capture the relationship between landslide triggering paths, landslide linkage values ​​and potential sliding hazard. Since this problem involves spatial features and multiple input sources (such as landslide triggering paths and linkage values ​​of multiple nodes of interest), it may be necessary to use more complex model structures, such as graph neural networks (GNNs) or hybrid models combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The processed features and corresponding potential sliding hazards are input into the model as training data, and the model parameters are adjusted by back-propagation algorithms or other optimization algorithms. Use the validation set to evaluate the model performance, and adjust the model structure and hyperparameters based on the evaluation results. Since this problem involves multiple input sources and complex spatial relationships, multiple iterations of tuning may be required to achieve satisfactory performance.

[0155] In a possible implementation, the historical landslide triggering paths of the respective focus nodes of the multi-risk block are characterized by historical landslide triggering path parameters. Step A120 includes:

[0156] For each of the first slope blocks of the multi-risk blocks, the following steps are performed respectively:

[0157] Step A121, obtaining the slope block distribution position of the first slope block in the basic distribution map.

[0158] Step A122, based on the distribution position of the slope blocks and the historical landslide trigger path parameters of each of the focus nodes of the multi-risk block, determine the weighting coefficient of each of the focus nodes corresponding to the first slope block, and the weighting coefficient is the weighting coefficient corresponding to the landslide linkage value from the focus node to the risk observation plane.

[0159] Step A123, based on the weighted coefficients of the first slope blocks corresponding to the respective focus nodes, a fusion calculation is performed on the landslide linkage values ​​from the respective focus nodes to the risk observation plane to generate a fusion calculation result.

[0160] Step A124, using the fusion calculation result as the potential sliding hazard of the first slope block of the multi-risk block.

[0161] Assume that the server is processing a slope monitoring area containing multiple multivariate risk blocks. One of the multivariate risk blocks is denoted as multivariate risk block M, which contains three focus nodes, namely focus nodes K, L and M.

[0162] The server first obtains a first slope block in the multi-risk blocks M, such as the first slope block N.

[0163] The server obtains the slope block distribution position of the first slope block N in the basic distribution map, assuming that its coordinate range is from (x1, y1) to (x2, y2).

[0164] Then, the server determines the weighting coefficients of the first slope block N corresponding to each focus node according to the distribution position of the slope block and the historical landslide triggering path parameters of each focus node in the multi-risk block M.

[0165] The historical landslide triggering path parameters of the focus node K show that it has a greater impact on the first slope block N at the current position. After comprehensive consideration, the weighted coefficient of the landslide linkage value from the focus node K to the risk observation plane of the first slope block N is determined to be 0.4.

[0166] The historical landslide triggering path parameters of the focus node L show that its impact on the first slope block N is relatively small, and the weighting coefficient of the first slope block N corresponding to the focus node L is determined to be 0.3.

[0167] The historical landslide triggering path parameters of the focus node M show that it has a certain impact on the first slope block N. The weighting coefficient of the first slope block N corresponding to the focus node M is determined to be 0.3.

[0168] The landslide linkage value from the focus node K to the risk observation plane is 0.7, the landslide linkage value from the focus node L is 0.6, and the landslide linkage value from the focus node M is 0.8.

[0169] The server performs fusion calculation on the landslide linkage value from each focus node to the risk observation plane according to the weighted coefficient of each focus node corresponding to the first slope block N.

[0170] The calculation process is: 0.7×0.4+0.6×0.3+0.8×0.3=0.68.

[0171] The server uses the fusion calculation result 0.68 as the potential slip hazard of the first slope block N in the multi-risk block M.

[0172] For other first side slope blocks in the multi-risk block M, such as the first side slope block P and the first side slope block Q, the server also processes according to the same steps.

[0173] For the first slope block P, after determining its slope block distribution position in the basic distribution map, the weighting coefficients of the corresponding focus nodes K, L and M are determined to be 0.3, 0.4 and 0.3 respectively according to the historical landslide triggering path parameters of the focus nodes.

[0174] The landslide linkage value of each focus node remains unchanged, and a fusion calculation is performed: 0.7×0.3+0.6×0.4+0.8×0.3=0.69, and 0.69 is taken as the potential sliding hazard of the first slope block P.

[0175] For the first slope block Q, assuming that the corresponding weighted coefficients are 0.5, 0.2 and 0.3, the fusion calculation is: 0.7×0.5+0.6×0.2+0.8×0.3=0.71, and 0.71 is taken as the potential slip hazard of the first slope block Q.

[0176] Therefore, the potential sliding hazard of each first slope block in the multi-risk block M can be accurately determined, providing strong support for the subsequent generation of accurate and reliable slope landslide risk distribution maps and early warning information.

[0177] Wherein, according to the potential sliding danger of each first slope block, each first slope block is constructed in the basic distribution map to generate a landslide risk thermal distribution map, including:

[0178] For the potential sliding hazard of each of the first slope blocks in the multi-risk block, the following steps are performed respectively to generate the landslide risk thermal distribution map: when the weighted coefficients of the first slope blocks corresponding to each of the focus nodes are not less than 0, the first slope blocks are constructed in the basic distribution map according to the potential sliding hazard of the first slope blocks.

[0179] Assume that the server is processing a slope monitoring area including multiple multivariate risk blocks, one of which is denoted as multivariate risk block X.

[0180] The multi-risk block X includes multiple first slope blocks, such as first slope block Y and first slope block Z.

[0181] The server first determines the weight coefficients of each focus node corresponding to the first slope block Y. Assuming that the focus nodes are node A, node B, and node C, the corresponding weight coefficients are 0.3, 0.4, and 0.3, respectively. Since these weight coefficients are not less than 0, the server will construct the first slope block Y in the basic distribution map according to its potential slip hazard.

[0182] Assume that the potential slip risk of the first slope block Y is calculated to be 0.7. The server finds the corresponding position in the basic distribution map and fills the first slope block Y with a specific mark or color to indicate that its potential slip risk is 0.7.

[0183] Next, the first slope block Z is processed. The weight coefficients of the corresponding focus nodes (such as node D, node E and node F) are determined to be 0.2, 0.3 and 0.5 respectively, which are also not less than 0.

[0184] Assuming that the potential slip risk of the first slope block Z is 0.6, the server constructs the first slope block Z at the corresponding position in the basic distribution map, and uses a different mark or color from the first slope block Y to indicate that its potential slip risk is 0.6.

[0185] The server also processes the other first slope blocks in the multivariate risk block X in the same way. Through such operations, all the first slope blocks that meet the conditions (weighted coefficients are not less than 0) are gradually constructed in the basic distribution map, and finally a landslide risk thermal distribution map that can intuitively reflect the distribution of the potential sliding hazard of the slope is generated.

[0186] In a possible implementation, constructing each of the first slope blocks in the basic distribution map according to the potential sliding hazard of each of the first slope blocks to generate a landslide risk thermal distribution map includes:

[0187] For each of the first slope blocks, the following steps are performed to generate the landslide risk thermal distribution map: determine the distribution point to be constructed of the first slope block in the basic distribution map, and determine the target slope block located at the distribution point to be constructed from the basic distribution map. Output the potential sliding hazard of the target slope block as the potential sliding hazard of the first slope block to construct the first slope block in the basic distribution map.

[0188] In this embodiment, it is assumed that the server is processing a complex slope monitoring area and has calculated the potential sliding hazards of multiple first slope blocks.

[0189] For example, there is a first slope block A. The server first determines the distribution points to be constructed in the basic distribution map of the first slope block A. Assume that through calculation and positioning, it is determined that the distribution points to be constructed are the area with coordinates ranging from (100, 200) to (150, 250) in the basic distribution map.

[0190] Then, the server searches the basic distribution map for the target slope block located at the distribution point to be constructed. After searching and comparing, it is found that the target slope block B is located in the area.

[0191] Next, the server outputs the potential slip hazard of the target slope block B as the potential slip hazard of the first slope block A. Assuming that the potential slip hazard of the target slope block B is 0.8, then this 0.8 is determined as the potential slip hazard of the first slope block A.

[0192] The server processes other first slope blocks in this manner.

[0193] For example, for the first slope block C, its distribution points to be constructed in the basic distribution map are determined to be the area with coordinates ranging from (300, 400) to (350, 450), and the target slope block D located in this area is found. Assuming that the potential slip hazard of the target slope block D is 0.6, 0.6 is output as the potential slip hazard of the first slope block C, and the first slope block C is constructed in the basic distribution map.

[0194] By performing such processing on each first slope block, the server gradually completes the construction of all first slope blocks in the basic distribution map, and finally generates a landslide risk thermal distribution map that can clearly show the potential sliding hazard of each slope block.

[0195] In a possible implementation manner, before outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block, the method further includes:

[0196] Step B110, when no other slope blocks are constructed at the target slope block, the step of outputting the potential sliding hazard of the target slope block as the potential sliding hazard of the first slope block is executed.

[0197] Step B120, when other slope blocks have been constructed at the target slope block, obtaining the target potential sliding hazard of the other slope blocks constructed at the target slope block.

[0198] Step B130, when the target potential slip hazard is greater than the potential slip hazard of the first slope block, the step of outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block is executed.

[0199] Wherein, when the target potential slip hazard is not greater than the potential slip hazard of the first slope block, the target potential slip hazard is maintained stable.

[0200] In this embodiment, it is assumed that the server is processing a plurality of first slope blocks in a slope monitoring area.

[0201] For a specific first side slope block E, the server first determines the distribution point to be constructed in the basic distribution map, assuming that the distribution point to be constructed corresponds to the target side slope block F.

[0202] The server checks the situation at the target slope block F. If no other slope blocks are constructed at the target slope block F, the server directly outputs the potential slip hazard of the target slope block F as the potential slip hazard of the first slope block E, and constructs the first slope block E in the basic distribution map.

[0203] If other slope blocks have been constructed at the target slope block F, for example, slope block G already exists. The server obtains the target potential slip hazard of slope block G. Assume that the potential slip hazard of the first slope block E is 0.7, and the target potential slip hazard of slope block G is 0.8. Since 0.8 is greater than 0.7, the server executes the step of outputting the potential slip hazard of the target slope block F (i.e., 0.8) as the potential slip hazard of the first slope block E, and constructs the first slope block E in the basic distribution map.

[0204] For another first slope block H, a slope block J has been constructed at its corresponding target slope block I. Assume that the potential slip hazard of the first slope block H is 0.9, and the target potential slip hazard of the slope block J is 0.8. Since 0.9 is greater than 0.8, the server does not change the target potential slip hazard of the slope block J, maintains its stability, and still constructs 0.9 as the potential slip hazard of the first slope block H in the basic distribution map.

[0205] The server processes each first slope block according to such rules to ensure that each slope block is accurately constructed in the basic distribution map and their potential sliding danger is reasonably determined.

[0206] In a possible implementation, step S140 includes:

[0207] Step S141, transferring the potential sliding hazard of each of the first slope blocks to each target slope block within the distribution range other than each of the first slope blocks in the landslide risk thermal distribution map, and generating the potential sliding hazard of each of the target slope blocks.

[0208] Step S142, constructing each of the target slope blocks in the landslide risk thermal distribution map according to the potential sliding hazard of each of the target slope blocks, and using the constructed target slope blocks as the second slope blocks to generate a slope landslide risk distribution map of the target slope monitoring area.

[0209] In this embodiment, it is assumed that the server has generated a landslide risk thermodynamic distribution map including multiple first slope blocks, for example, the potential sliding hazard of first slope block A is 0.8, the potential sliding hazard of first slope block B is 0.7, and the potential sliding hazard of first slope block C is 0.9.

[0210] The server first transfers the potential sliding risk of the first slope block A, 0.8, to each target slope block in the distribution range except the first slope block A in the landslide risk thermal distribution map. Assume that the target slope blocks D and E adjacent to the first slope block A are subject to risk transfer. The server appropriately attenuates 0.8 according to the preset risk transfer algorithm, for example, attenuates it to 0.6 and transfers it to the target slope block D, and attenuates it to 0.5 and transfers it to the target slope block E.

[0211] Similarly, the potential sliding hazard of the first slope block B, 0.7, is transferred to the adjacent target slope blocks F and G, attenuating to 0.4 and 0.3 respectively.

[0212] The potential sliding risk of the first slope block C, 0.9, is transferred to the adjacent target slope blocks H and I, attenuating to 0.7 and 0.6.

[0213] This generates the potential sliding hazards of the target slope blocks D, E, F, G, H, and I.

[0214] Next, the server constructs these target slope blocks in the landslide risk thermal distribution map according to the potential sliding hazard of each target slope block.

[0215] For the target slope block D, whose potential sliding hazard is 0.6, the server constructs the block at the corresponding position in the figure and marks it as the second slope block.

[0216] For the target slope block E, the potential sliding hazard is 0.5, and it is also constructed and marked as the second slope block at the corresponding position.

[0217] The target slope blocks F, G, H, and I are also constructed and marked according to their respective potential sliding hazards.

[0218] Through the above steps, a complete slope landslide risk distribution map of the target slope monitoring area is finally generated, which clearly shows the potential landslide risk distribution of the entire area.

[0219] In a possible implementation, the risk transfer includes X rounds of risk transfer, and step S141 includes:

[0220] Step S1411, for the first round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of each of the first slope blocks is risk-transferred to the connected slope blocks of each of the first slope blocks, the potential sliding hazard of the connected slope blocks of each of the first slope blocks is generated, and the potential sliding hazard of the connected slope blocks of each of the first slope blocks is used as the potential sliding hazard of the first round target slope block.

[0221] Step S1412, for the mth round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of the target slope block of the m-1th round is risk-transferred to the connected slope blocks of the target slope block of the m-1th round, the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is generated, and the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is used as the potential sliding hazard of the target slope block of the m-1th round.

[0222] Wherein, X and m are integers greater than 0, and X is not less than m.

[0223] In this embodiment, it is assumed that the server is processing a slope monitoring area and has determined that there are three first slope blocks, namely, the first slope block P has a potential sliding risk of 0.8, the first slope block Q has a potential sliding risk of 0.7, and the first slope block R has a potential sliding risk of 0.9. It is assumed that risk transfer is performed for a total of 3 rounds, that is, X=3.

[0224] First round of risk transfer:

[0225] The server transfers the potential sliding risk of the first slope block P, 0.8, to its connected slope blocks, such as connected slope blocks S and T. Assume that 0.8 is attenuated by a certain proportion, attenuated to 0.6 and transferred to the connected slope block S, and attenuated to 0.5 and transferred to the connected slope block T.

[0226] The potential sliding risk of the first slope block Q, 0.7, is transferred to its connected slope blocks U and V, attenuated to 0.4 and 0.3 respectively.

[0227] The potential sliding risk of the first slope block R, 0.9, is transferred to its connected slope blocks W and X, attenuated to 0.7 and 0.6 respectively.

[0228] The potential sliding hazards of connected slope blocks S, T, U, V, W, and X are generated, and these connected slope blocks become the first round of target slope blocks, and their potential sliding hazards are 0.6, 0.5, 0.4, 0.3, 0.7, and 0.6, respectively.

[0229] Second round of risk transfer (m=2):

[0230] The server transfers the potential sliding risk of 0.6 of the first round target slope block S to its connected slope blocks, such as connected slope blocks Y and Z, and attenuates it to 0.4 and 0.3 respectively.

[0231] The potential sliding hazard of the first round target slope block T is transferred to its connected slope blocks A and B, attenuating to 0.3 and 0.2 respectively.

[0232] By analogy, the same processing is performed on the first round of target slope blocks U, V, W, and X.

[0233] The potential sliding hazards of connected slope blocks Y, Z, A, B, etc. are generated. These connected slope blocks become the second round of target slope blocks, and their potential sliding hazards are 0.4, 0.3, 0.3, 0.2, etc. respectively.

[0234] The third round of risk transfer (m=3):

[0235] The server continues to perform risk transfer and generation of potential sliding hazards in the above manner until there are no new connected slope blocks or the preset number of transfer rounds is reached.

[0236] Through such multiple rounds of risk transfer, the server can more comprehensively assess the potential landslide risk of the entire slope monitoring area, providing richer and more accurate data support for generating accurate slope landslide risk distribution maps.

[0237] Among them, step S1411 includes: for each connected slope block to which the potential slip hazard of the first slope block is transmitted, determining at least one target first slope block to which the risk is transmitted to the connected slope block, performing a fusion calculation on the potential slip hazard of the at least one target first slope block, and generating the potential slip hazard of the connected slope block.

[0238] In this embodiment, it is assumed that there are three first slope blocks in a slope monitoring area, namely, the potential sliding risk of the first slope block M is 0.8, the potential sliding risk of the first slope block N is 0.7, and the potential sliding risk of the first slope block O is 0.9. There are some slope blocks connected to these three first slope blocks. For example, the connected slope block P is simultaneously affected by the potential sliding risk of the first slope blocks M and N.

[0239] The server determines that the target first slope blocks whose risks are transferred to the connected slope block P are M and N. Then, the potential slip hazards of these two target first slope blocks are fused and calculated. Assuming a simple average calculation method is used, the potential slip hazards of M and N are weighted first. Since the potential slip hazard of M is higher, the weight is set to 0.6, and the weight of N is set to 0.4. The fusion calculation is: 0.8×0.6+0.7×0.4=0.76, and the potential slip hazard of the connected slope block P is generated to be 0.76.

[0240] For another example, the connected slope block Q is subject to the transmission of the potential slip risk of the first slope blocks N and O. The server determines that the target first slope blocks for risk transmission to the connected slope block Q are N and O. The weighted fusion calculation is also performed, assuming that the weight of N is 0.4 and the weight of O is 0.6. The calculation is: 0.7×0.4+0.9×0.6=0.82, and the potential slip risk of the connected slope block Q is generated to be 0.82.

[0241] For other connected slope blocks to which the potential sliding hazard of the first slope block is transmitted, the server determines the target first slope block in a similar manner and performs fusion calculations to generate accurate potential sliding hazard of each connected slope block.

[0242] Further, Figure 2 FIG. 1 is a schematic diagram of the hardware structure of a slope landslide risk warning system 100 based on artificial intelligence for implementing the method provided in the embodiment of the present application. Figure 2 As shown, the artificial intelligence-based slope landslide risk warning system 100 includes a memory 111, a storage controller 112, a processor 113, a peripheral interface 114, an input and output unit 115, an audio unit 116, a display unit 117 and a radio frequency unit 118.

[0243] The memory 111, the storage controller 112, the processor 113, the peripheral interface 114, the input / output unit 115, the audio unit 116, the display unit 117, and the radio frequency unit 118 are electrically connected to each other directly or indirectly to achieve data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines.

[0244] The memory 111 may be, but not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electric erasable programmable read-only memory (EEPROM), etc. The memory 111 is used to store the application 210, and the processor 113 executes the application 210 after receiving the execution instruction. The processor 113 and other possible components may access the memory 111 under the control of the storage controller 112.

[0245] The processor 113 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0246] The peripheral interface 114 couples various input / output devices to the processor 113 and the memory 111. In some embodiments, the peripheral interface 114, the processor 113 and the memory controller 112 can be implemented in a single chip. In other embodiments, they can be implemented by separate chips.

[0247] The input-output unit 115 is used to provide input data to the user to realize the interaction between the user and the artificial intelligence-based landslide risk warning system 100. The input-output unit 115 can be, but is not limited to, a mouse and a keyboard.

[0248] The audio unit 116 provides an audio interface to the user and may include one or more microphones, one or more speakers, and an audio circuit.

[0249] The display unit 117 provides an interactive interface (e.g., a user operation interface) between the artificial intelligence-based landslide risk warning system 100 and the user or is used to display image data. In this embodiment, the display unit 117 can be a liquid crystal display or a touch display. If it is a touch display, it can be a capacitive touch screen or a resistive touch screen that supports single-point and multi-point touch operations. Supporting single-point and multi-point touch operations means that the touch display can sense touch operations generated from one or more positions on the touch display, and the sensed touch operations are handed over to the processor for calculation and processing.

[0250] The radio frequency unit 118 is used to receive and send radio wave signals (such as electromagnetic waves) to achieve mutual conversion between radio waves and electrical signals, thereby realizing communication between the artificial intelligence-based landslide risk warning system 100 and the network or other communication devices.

[0251] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of the embodiments of the present application are described. Other embodiments are within the scope of the attached claims. In some cases, the actions or steps recorded in the claims can be performed according to an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0252] Each embodiment in the embodiments of the present application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the above different embodiments, since they are basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0253] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program. The above program may be stored in a computer-readable storage medium, and the above storage medium may be a read-only memory, a disk or an optical disk, etc.

Claims

1. A slope landslide risk early warning method based on artificial intelligence, characterized in that: The method comprises: Obtain the historical landslide triggering path of each historical landslide node in the three-dimensional image of the slope in the target slope monitoring area in the landslide risk plane space, and obtain the landslide linkage value from each historical landslide node to the risk observation plane in the landslide risk three-dimensional space. The landslide linkage value refers to the size or degree of the impact of a single historical landslide node on other areas of the slope when a landslide occurs. The landslide linkage value is obtained through mathematical models and geomechanical analysis, taking into account the factors of geological conditions, rainfall, and seismic activity; Obtaining a basic distribution map for constructing a slope landslide risk distribution map; Determining the potential sliding hazard of each first slope block in the three-dimensional slope image according to the historical landslide trigger path and the landslide linkage value of each historical landslide node, wherein the first slope block refers to the slope area preliminarily divided according to geological conditions and historical landslide event factors in the three-dimensional slope image; According to the potential sliding danger of each of the first slope blocks, construct each of the first slope blocks in the basic distribution map to generate a landslide risk thermal distribution map; According to the potential sliding danger of each of the first slope blocks, a second slope block is constructed within the distribution range except for each of the first slope blocks in the landslide risk thermal distribution map, and a slope landslide risk distribution map of the target slope monitoring area is generated; Generate corresponding warning information based on the slope landslide risk distribution map of the target slope monitoring area; The three-dimensional image of the slope is composed of a plurality of landslide risk blocks, and the three-dimensional image of the slope includes: a composite risk block generated by combining and connecting at least three landslide risk blocks associated with the historical landslide node, and the focus node of the composite risk block is the historical landslide node; Determining the potential sliding hazard of each first slope block of the three-dimensional slope image according to the historical landslide triggering path and the landslide linkage value of each historical landslide node includes: The composite risk block is divided into at least one multivariate risk block, and for each of the multivariate risk blocks, the following steps are performed to generate the potential slip hazard of each of the first slope blocks of the composite risk block of the slope three-dimensional image: Inputting the historical landslide triggering paths and the landslide linkage values ​​of each of the focus nodes of the multivariate risk block into a pre-trained potential sliding hazard decision model to determine the potential sliding hazard of each of the first slope blocks of the multivariate risk block of the slope three-dimensional image; The historical landslide triggering paths of each of the focus nodes of the multivariate risk block are characterized by historical landslide triggering path parameters; the historical landslide triggering paths and the landslide linkage values ​​of each of the focus nodes of the multivariate risk block are input into a pre-trained potential sliding hazard decision model to determine the potential sliding hazard of each of the first slope blocks of the multivariate risk block of the slope three-dimensional image, including: For each of the first slope blocks of the multi-risk blocks, the following steps are performed respectively: Obtaining a slope block distribution position of the first slope block in the basic distribution map; According to the distribution position of the slope blocks and the historical landslide triggering path parameters of each of the focus nodes of the multi-risk block, the weighting coefficient of each of the focus nodes corresponding to the first slope block is determined, and the weighting coefficient is the weighting coefficient corresponding to the landslide linkage value from the focus node to the risk observation plane; According to the weighted coefficients of the first slope block corresponding to the respective focus nodes, a fusion calculation is performed on the landslide linkage values ​​from the respective focus nodes to the risk observation plane to generate a fusion calculation result; Using the fusion calculation result as the potential sliding hazard of the first slope block of the multi-risk block; Wherein, according to the potential sliding danger of each first slope block, each first slope block is constructed in the basic distribution map to generate a landslide risk thermal distribution map, including: For the potential sliding danger of each of the first slope blocks in the multi-risk blocks, the following steps are performed respectively to generate the landslide risk thermodynamic distribution map: When the weighted coefficients of the first slope block corresponding to the respective focus nodes are not less than 0, constructing the first slope block in the basic distribution map according to the potential slip hazard of the first slope block; The method of constructing a second slope block within the distribution range other than the first slope blocks in the landslide risk thermal distribution map according to the potential sliding danger of each of the first slope blocks, and generating a slope landslide risk distribution map of the target slope monitoring area includes: The potential sliding hazard of each of the first slope blocks is transferred to each target slope block within the distribution range other than each of the first slope blocks in the landslide risk thermal distribution map to generate the potential sliding hazard of each of the target slope blocks; According to the potential sliding hazard of each target slope block, each target slope block is constructed in the landslide risk thermal distribution map, and the constructed target slope block is used as the second slope block to generate a slope landslide risk distribution map of the target slope monitoring area.

2. The artificial intelligence-based landslide risk early warning method according to claim 1 is characterized in that: The step of obtaining a basic distribution map for constructing a slope landslide risk distribution map includes: Obtain the target accuracy of the slope landslide risk distribution map to be constructed; A basic distribution map with the target accuracy for constructing a slope landslide risk distribution map is generated.

3. The artificial intelligence-based landslide risk early warning method according to claim 1 is characterized in that: The step of constructing each of the first slope blocks in the basic distribution map according to the potential sliding danger of each of the first slope blocks to generate a landslide risk thermal distribution map includes: For each of the first slope blocks, the following steps are performed respectively to generate the landslide risk thermal distribution map: Determine a distribution point to be constructed of the first slope block in the basic distribution map, and determine a target slope block located at the distribution point to be constructed from the basic distribution map; The potential sliding hazard of the target slope block is output as the potential sliding hazard of the first slope block, so as to construct the first slope block in the basic distribution map.

4. The artificial intelligence-based landslide risk early warning method according to claim 3 is characterized in that: Before outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block, the method further includes: When no other slope blocks are constructed at the target slope block, the step of outputting the potential sliding hazard of the target slope block as the potential sliding hazard of the first slope block is performed; When other slope blocks have been constructed at the target slope block, obtaining target potential sliding hazards of the other slope blocks constructed at the target slope block; When the target potential slip hazard is greater than the potential slip hazard of the first slope block, the step of outputting the potential slip hazard of the target slope block as the potential slip hazard of the first slope block is executed; The method further comprises: When the target potential slip hazard is not greater than the potential slip hazard of the first slope block, the target potential slip hazard is maintained stable.

5. The artificial intelligence-based landslide risk early warning method according to claim 1 is characterized in that: The risk transfer includes X rounds of risk transfer, wherein the potential sliding hazard of each first slope block is transferred to each target slope block within the distribution range other than each first slope block in the landslide risk thermal distribution map to generate the potential sliding hazard of each target slope block, including: For the first round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of each of the first slope blocks is transferred to the connected slope blocks of each of the first slope blocks to generate the potential sliding hazard of the connected slope blocks of each of the first slope blocks, and the potential sliding hazard of the connected slope blocks of each of the first slope blocks is used as the potential sliding hazard of the first round target slope block; For the mth round of risk transfer in the X rounds of risk transfer, the potential sliding hazard of the target slope block of the m-1th round is risk-transferred to the connected slope blocks of the target slope block of the m-1th round, the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is generated, and the potential sliding hazard of the connected slope blocks of the target slope block of the m-1th round is used as the potential sliding hazard of the target slope block of the m-1th round; Wherein, X and m are integers greater than 0, and X is not less than m; The step of transferring the potential sliding hazards of each of the first slope blocks to the connected slope blocks of each of the first slope blocks to generate the potential sliding hazards of the connected slope blocks of each of the first slope blocks includes: For each connected slope block to which the potential slip hazard of the first slope block is transferred, at least one target first slope block to which the risk is transferred is determined, and a fusion calculation is performed on the potential slip hazard of the at least one target first slope block to generate the potential slip hazard of the connected slope block.

6. A slope landslide risk early warning system based on artificial intelligence, characterized in that: The artificial intelligence-based slope landslide risk warning system includes a processor and a machine-readable storage medium, in which machine-executable instructions are stored. The machine-executable instructions are loaded and executed by the processor to implement the artificial intelligence-based slope landslide risk warning method described in any one of claims 1 to 5.

Citation Information

Cited By

  • Rock slope risk assessment method and system based on artificial intelligence

    CN120634242A

  • Ecological slope early warning method and system based on artificial intelligence

    CN122369230A