Slope stability prediction method and device under rainfall condition

Abstract

The invention provides a slope stability prediction method and device under a rainfall condition, and relates to the technical field of slope stability prediction, and the method comprises the steps: employing a regular grid to divide a slope, and constructing a spatial-temporal feature data set containing geological, hydrological and topographic parameters; calculating a regional cooperation index based on the multi-dimensional feature vector, and dividing the slope into a plurality of communities and isolated risk areas; calculating the dynamic safety coefficient of each community by taking the mass center sub-region as a representative, and dividing the stability level according to a threshold value; and establishing a three-dimensional digital twinborn model to realize risk visualization, triggering early warning for a continuous three-period safety coefficient decreasing area, and displaying an evolution track. According to the method, accurate monitoring and early warning of slope stability are realized through multi-source data fusion and dynamic partition evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of slope stability prediction, and in particular to a method and device for predicting slope stability under rainfall conditions. Background Art

[0002] Traditional slope stability analysis methods typically rely on static data, such as the physical and mechanical properties of the soil, slope, and vegetation cover. These methods often employ empirical formulas and simple limit equilibrium analysis, failing to account for potential risks arising from dynamic changes. For example, the intensity and duration of rainfall can significantly affect soil moisture content and pore water pressure, leading to changes in soil strength. However, existing technologies often fail to monitor rainfall conditions and their impact on slope stability in real time. Consequently, static assessments often fail to accurately reflect the actual safety status of slopes, especially under extreme climatic conditions. This leads to an increased incidence of landslides, posing a serious threat to human and property safety.

[0003] In addition, traditional slope stability assessments lack a comprehensive consideration of spatiotemporal dynamics and are typically based on isolated analysis of local features, making it difficult to effectively identify and demarcate potential risk areas. In many cases, complex interactions exist between different sub-regions of the slope, and these relationships are often overlooked in traditional methods. In the process of acquiring and processing slope characteristic data, existing technologies are unable to timely integrate information from different sources, such as real-time rainfall intensity, geological parameters, and hydrological conditions, which affects the real-time assessment of the overall safety and stability of the slope. In addition, the lack of effective intelligent monitoring methods limits the response speed and accuracy of slope risk assessments, making it impossible to form an efficient early warning mechanism. Therefore, there is an urgent need for a dynamic monitoring and assessment method that integrates multidimensional data to cope with complex and changing natural environmental conditions, thereby improving the scientific nature and effectiveness of slope management.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and device for predicting slope stability under rainfall conditions to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for predicting slope stability under rainfall conditions, comprising the following steps:

[0008] Step 1: Use a regular grid to discretize the target slope into several sub-regions. Each sub-region is assigned regional attributes including geological parameters, hydrological parameters, and topographic parameters. At the same time, the real-time rainfall intensity data provided by the meteorological radar is integrated to form a slope characteristic dataset with spatiotemporal attributes.

[0009] Step 2: Based on the geological, hydrological, and topographic parameters of each sub-region, a multidimensional feature vector is used to characterize the local slope information of each sub-region. The regional synergy index between any two sub-regions is defined based on the multidimensional feature vector. The sub-regions are clustered using a spectral clustering algorithm to divide the target slope into multiple communities and isolated risk areas.

[0010] Step 3: Determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, select its centroid subregion as a representative to determine the dynamic safety factor of the community;

[0011] Step 4: Establish a three-dimensional digital twin model of the target slope, accurately match the community division results with the model through a spatial mapping algorithm, and use different colors to identify the stability level of each community and isolated risk area in the visualization interface. For communities or isolated risk areas whose dynamic safety factors show a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the three-dimensional digital twin model.

[0012] Furthermore, a regular grid is used to discretize the target slope into several sub-regions. The grid size is determined to be adjustable within a range of 5m×5m to 20m×20m based on the slope scale and monitoring accuracy requirements. Regional attributes are assigned to each sub-region, including geological parameters such as permeability, internal friction angle, and cohesion; hydrological parameters such as initial water content, saturated water content, and real-time pore water pressure; and topographic parameters such as slope and surface curvature.

[0013] The real-time rainfall intensity data provided by the meteorological radar is integrated, and its temporal and spatial resolution matches the grid division scale, ultimately forming a slope characteristic dataset with multi-dimensional characteristics of time, space and attributes. The parameters are stored in matrix form and a temporal and spatial index relationship is established.

[0014] Geological parameters, hydrological parameters, and topographic parameters are obtained through field surveys, sensor monitoring, and remote sensing measurements as regional attributes of each sub-region.

[0015] Furthermore, a multidimensional feature vector is constructed for each sub-region based on the geological parameters, hydrological parameters, and topographic parameters of each sub-region. The expression of the multidimensional feature vector is as follows:

[0016]

[0017] Where, F i is the multidimensional feature vector of subregion i, i is the index of the subregion, f geo,i 、f hydro,i and f top,i represent the geological parameters, hydrological parameters and topographic parameters of sub-region i, respectively, k i 、 and c i are the permeability coefficient, internal friction angle and cohesion of sub-region i, θ 0,i ,θ x,i and u i are the initial water content, saturated water content and real-time pore water pressure of sub-region i, α i and β i are the slope and surface curvature of sub-area i, respectively;

[0018] The regional synergy index between any two sub-regions is defined based on the multi-dimensional feature vector, and the formula is as follows:

[0019]

[0020] In the formula, SCI ij represents the regional synergy index between sub-region i and sub-region j, i and j are the indexes of sub-regions, and i≠j; Dgeo(f geo,i ,f geo,j )、Dhydro(f hydro,i ,f hydro,j ) and Dtop(f top,i ,f top,j ) represent the differences in geological characteristics, hydrological characteristics, and topographic characteristics between sub-regions i and j, respectively; f geo,j 、f hydro,j and f top,j represent the geological parameters, hydrological parameters and topographic parameters of sub-region j, respectively, k j 、 and c j are the permeability coefficient, internal friction angle and cohesion of sub-region j, θ 0,j ,θ x,j and u j are the initial water content, saturated water content and real-time pore water pressure of sub-region j, α j and β jare the slope and surface curvature of sub-region j, ω1, ω2 and ω3 are preset weights, ω1>ω2>ω3>0, and ω1+ω2+ω3=1.

[0021] Furthermore, the sub-regions are clustered using the spectral clustering algorithm. The specific logic is as follows:

[0022] A regional synergy index matrix is ​​constructed based on the regional synergy index between all sub-regions, and the sub-regions are divided into communities using the spectral clustering algorithm. The specific process is as follows: the Laplace matrix of the regional synergy index matrix is ​​calculated, the optimal number of communities K is determined through eigenvalue decomposition, and the sub-region is divided into K communities using the K-means algorithm; a dynamic similarity threshold τ is set so that the average regional synergy index between sub-regions within a community is less than τ, while the average regional synergy index of sub-regions between communities is greater than τ; for any sub-region, if its regional synergy index with all sub-regions is greater than or equal to the dynamic similarity threshold τ, it is determined to be an isolated risk area;

[0023] The above similarity judgment is repeated until the community affiliation of all sub-regions no longer changes, forming the final community division result.

[0024] Furthermore, the dynamic safety factor of each community and isolated risk area is determined based on the slope characteristic dataset according to the following formula:

[0025]

[0026] Where FS(t) represents the dynamic safety factor of the sub-region at time t, which is used to reflect the immediate stability state of the slope under rainfall conditions; c is the cohesion of the soil in the sub-region, γ is the bulk density of the soil in the sub-region, which is used to characterize the normal stress component generated by the weight of the sliding body. The sliding body refers to the part of the soil in the sub-region that can slide along the potential sliding surface; h is the depth of the potential sliding surface, α is the slope angle of the sub-region, u(t) is the pore water pressure of the sub-region at time t, s is the surface curvature of the sub-region, and R is the curvature radius of the slope of the sub-region. is the internal friction angle of the sub-region soil, k(θ) is the permeability coefficient related to the water content, k s is the saturated permeability coefficient, which is used to characterize the maximum permeability of the soil when it is fully saturated and is determined by indoor permeability tests; θ(t) is the current soil moisture content in the sub-region, θ r is the residual volume moisture content, which indicates the proportion of water in the soil that cannot be discharged by gravity and is determined by the soil type; θ0 and θ x are the initial water content and saturated water content, respectively; n is an empirical index used to control the rate at which the permeability coefficient changes with water content and is related to the soil pore distribution;

[0027] For each community, select its centroid sub-region as a representative to determine the dynamic safety factor of the community. The method for selecting the centroid sub-region is as follows: First, calculate the average coordinates of the geometric centers of all sub-regions within the community as the centroid position, and then select the sub-region with the shortest Euclidean distance to this centroid position as the centroid sub-region; if there are multiple, further compare the cosine similarity between their multi-dimensional feature vectors and the average feature of the community, and select the sub-region with the highest similarity as the centroid sub-region, and this centroid sub-region satisfies the following conditions:

[0028] (1) It is located inside the community and is not adjacent to the isolated risk area;

[0029] (2) The deviation of its geological parameters and hydrological parameters from the community average value does not exceed 15%.

[0030] Furthermore, compare the dynamic safety factor with the stability threshold, and based on the comparison results, divide the stability levels of each community and the isolated risk area. The specific logic is as follows:

[0031] When FS(t) ≥ FY, mark this community or isolated risk area as in a stable state, indicating that the anti-sliding force of the slope in this community or isolated risk area is significantly greater than the sliding force;

[0032] When 0.5*FY ≤ FS(t) < FY, mark this community or isolated risk area as in a warning state, indicating that monitoring needs to be strengthened and personnel entry needs to be restricted;

[0033] When 0.2*FY ≤ FS(t) < 0.5*FY, mark this community or isolated risk area as in a high-risk state, indicating that local sliding may occur and an emergency plan needs to be activated;

[0034] When FS(t) < 0.2*FY, mark this community or isolated risk area as in an unstable state, indicating that the landslide has occurred or is approaching, and immediately trigger an evacuation alarm;

[0035] In the formula, FS(t) represents the dynamic safety factor at time t, and FY is the preset stability threshold.

[0036] Furthermore, build a three-dimensional digital twin model of the target slope based on the BIM platform, and accurately match the community division results with the model grid through spatial coordinate mapping; in the visualization interface, use a four-color gradient rendering scheme to display the stability levels of each community and the isolated risk area in real time, and overlay and display the real-time rainfall intensity distribution; among them, green, yellow, orange, and red respectively represent the stable state, warning state, high-risk state, and unstable state;

[0037] For areas where the dynamic safety factor shows a monotonically decreasing trend over three consecutive monitoring cycles, an audible and visual warning signal is automatically triggered. At the same time, the area is marked in the model with a pulse flashing effect, and a warning report is generated containing the location coordinates, the rate of change of the dynamic safety factor, and the potential slip direction. The system simultaneously records the stability evolution trajectory of each area and supports the display of the landslide development process in a timeline playback manner.

[0038] The present invention further provides a device for predicting slope stability under rainfall conditions, wherein the device is used to execute the above-mentioned method for predicting slope stability under rainfall conditions, and comprises:

[0039] The slope grid modeling module is used to discretize the target slope into several sub-regions using a regular grid, assign regional attributes including geological parameters, hydrological parameters, and topographic parameters to each sub-region, and integrate the real-time rainfall intensity data provided by the meteorological radar to form a slope characteristic dataset with spatiotemporal attributes;

[0040] The intelligent community division module is used to characterize the local slope information of each sub-region using a multidimensional feature vector based on the geological parameters, hydrological parameters, and topographic parameters of each sub-region. The module also defines the regional synergy index between any two sub-regions based on the multidimensional feature vector and clusters the sub-regions using a spectral clustering algorithm. This allows the target slope to be divided into multiple communities and isolated risk areas.

[0041] The dynamic safety assessment module is used to determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, its centroid sub-region is selected as a representative to determine the dynamic safety factor of the community;

[0042] The three-dimensional early warning visualization module is used to establish a three-dimensional digital twin model of the target slope, accurately match the community division results with the model through a spatial mapping algorithm, and identify the stability level of each community and isolated risk area with different colors in the visualization interface. For communities or isolated risk areas whose dynamic safety factors show a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the three-dimensional digital twin model.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] This solution discretizes the slope using a regular grid, integrating geological parameters, hydrological parameters, topographic parameters, and real-time rainfall intensity data to form a slope characteristic dataset with spatiotemporal attributes, enabling dynamic monitoring and real-time assessment of slope status. The calculation of the regional synergy index based on multidimensional feature vectors, combined with the precise assessment of the dynamic safety factor, can effectively identify potential risk areas on the slope and promptly implement corresponding early warning measures. In addition, the application of a three-dimensional digital twin model not only enables visual monitoring, but also dynamically records the evolution of slope stability, thus providing a scientific basis for decision makers and significantly improving the intelligent level of slope management and emergency response capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Schematic diagram of the overall method flow of the present invention;

[0046] Figure 2 This is a schematic diagram of the overall device module of the present invention;

[0047] Figure 3-5 These are the functional relationship diagrams of cohesion, internal friction angle, pore water pressure and dynamic safety factor. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0049] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meaning understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and other similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and other similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0050] "Up", "down", "left", "right", etc. are only used to express relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0051] Example:

[0052] See also Figure 1 , the present invention provides a technical solution:

[0053] A method for predicting slope stability under rainfall conditions, comprising the following steps:

[0054] Step 1: Use a regular grid to discretize the target slope into several sub-regions. Each sub-region is assigned regional attributes including geological parameters, hydrological parameters, and topographic parameters. At the same time, the real-time rainfall intensity data provided by the meteorological radar is integrated to form a slope characteristic dataset with spatiotemporal attributes.

[0055] In this embodiment, a regular grid is used to discretize the target slope into several sub-regions. The grid size is determined to be adjustable within a range of 5m×5m to 20m×20m based on the slope scale and monitoring accuracy requirements. Regional attributes, including geological parameters, hydrological parameters, and topographic parameters, are assigned to each sub-region. The geological parameters include permeability, internal friction angle, and cohesion; the hydrological parameters include initial water content, saturated water content, and real-time pore water pressure; and the topographic parameters include slope and surface curvature.

[0056] The real-time rainfall intensity data provided by the meteorological radar is integrated, and its temporal and spatial resolution matches the grid division scale, ultimately forming a slope characteristic dataset with multi-dimensional characteristics of time, space and attributes. The parameters are stored in matrix form and a temporal and spatial index relationship is established.

[0057] Geological parameters, hydrological parameters, and topographic parameters are obtained through field surveys, sensor monitoring, and remote sensing measurements as regional attributes of each sub-region.

[0058] Step 1 discretizes the target slope into multiple subregions using a regular grid. Each subregion is assigned detailed geological, hydrological, and topographic parameters. This data is then integrated with real-time rainfall intensity data to form a slope characteristic dataset with spatiotemporal attributes. The advantage of this approach lies in its systematic integration of multiple influencing factors, ensuring comprehensive monitoring and assessment of slope conditions. Compared with existing technologies, this overcomes the limitations of traditional static analysis methods, enabling the timely capture and processing of real-time dynamic changes, thereby improving the accuracy and timeliness of risk identification.

[0059] By implementing Step 1, the overall solution establishes a dynamic, comprehensive slope monitoring system, promoting the scientific and accurate prediction of slope stability. Compared to previous approaches that isolated local features, this solution divides the slope into multiple sub-regions, clarifying the interrelationships between these regions and enabling more effective identification of potential risk areas. This approach not only improves the intelligence of data processing but also provides a solid foundation for subsequent dynamic safety factor calculations and risk warnings, further enhancing the overall efficiency and safety of slope management.

[0060] Step 2: Based on the geological, hydrological, and topographic parameters of each sub-region, a multidimensional feature vector is used to characterize the local slope information of each sub-region. The regional synergy index between any two sub-regions is defined based on the multidimensional feature vector. The sub-regions are clustered using a spectral clustering algorithm, and the remaining sub-regions are marked as isolated risk areas. In this way, the target slope is divided into multiple communities and isolated risk areas.

[0061] In this embodiment, a multidimensional feature vector is constructed for each sub-region based on the geological parameters, hydrological parameters, and topographic parameters of each sub-region. The expression of the multidimensional feature vector is as follows:

[0062]

[0063] Where, F i is the multidimensional feature vector of subregion i, i is the index of the subregion, f geo,i 、f hydro,i and f top,i represent the geological parameters, hydrological parameters and topographic parameters of sub-region i, respectively, k i 、 and c i are the permeability coefficient, internal friction angle and cohesion of sub-region i, θ 0,i ,θ x,i and u i are the initial water content, saturated water content and real-time pore water pressure of sub-region i, α i and β i are the slope and surface curvature of sub-area i, respectively;

[0064] The regional synergy index between any two sub-regions is defined based on the multi-dimensional feature vector, and the formula is as follows:

[0065]

[0066] In the formula, SCI ij represents the regional synergy index between sub-region i and sub-region j, i and j are the indexes of sub-regions, and i≠j; Dgeo(f geo,i ,f geo,j )、Dhydro(f hydro,i ,f hydro,j ) and Dtop(f top,i ,f top,j ) represent the differences in geological characteristics, hydrological characteristics, and topographic characteristics between sub-regions i and j, respectively; f geo,j 、f hydro,j and f top,j represent the geological parameters, hydrological parameters and topographic parameters of sub-region j, respectively, k j 、 and cj are the permeability coefficient, internal friction angle and cohesion of sub-region j, θ 0,j ,θ x,j and u j are the initial water content, saturated water content and real-time pore water pressure of sub-region j, α j and β j are the slope and surface curvature of subregion j, respectively. ω1, ω2, and ω3 are preset weights: ω1 = 0.5, ω2 = 0.3, and ω3 = 0.2. The weighting reflects the importance of different characteristics in contributing to the regional synergy index. In slope stability analysis, geological characteristics directly affect the physical properties and anti-sliding capacity of the soil. Therefore, their differences have the greatest impact on regional stability and are therefore assigned the highest weight. Hydrological characteristics, while equally important, have a relatively minor influence and are therefore assigned a medium weight. Topographic characteristics can affect landslide risk in some cases, but their impact is generally smaller than that of geological and hydrological characteristics and are therefore assigned the lowest weight. This weighting helps more accurately reflect the relative contributions of different characteristics to regional synergy.

[0067] In the first formula, the dependent variable, the regional synergy index, specifically reflects the similarity or synergy between any two sub-regions. A larger regional synergy index value indicates more similar characteristics between the two sub-regions, meaning they are closer in geological, hydrological, and topographical parameters, and vice versa. The regional synergy index can help identify areas with similar characteristics, enabling effective community delineation in slope stability prediction, thereby improving the ability to identify and manage potential risks. The independent variables include geological characteristic difference, hydrological characteristic difference, and topographic characteristic difference. These independent variables represent the differences in geological, hydrological, and topographical parameters between the two sub-regions, respectively. The relationship between the independent variables and the dependent variable is that these differences are calculated to reflect the degree of similarity between the sub-regions. Specifically, geological characteristic difference reflects variations in geological parameters such as permeability, internal friction angle, and cohesion. These parameters directly affect the physical properties and stability of soil, and therefore their differences can have a significant impact on slope stability. Hydrological characteristic difference reflects variations in hydrological parameters such as initial moisture content, saturated moisture content, and real-time pore water pressure. These parameters are related to the soil's moisture state and, in turn, affect the slope's ability to resist sliding. The difference in terrain characteristics reflects changes in slope and surface curvature. These terrain factors affect the potential sliding surface and stress state of the landslide. Therefore, the difference in terrain characteristics will also directly affect the stability of the slope. In this formula, the relationship between the dependent variable and the independent variable is an inverse correlation. Specifically: when the value of any of the independent variables, such as the difference in geological characteristics, the difference in hydrological characteristics, or the difference in terrain characteristics, increases, it means that the difference between sub-region i and sub-region j increases, that is, their characteristics are no longer similar. This will cause the value of the regional synergy index to decrease, reflecting that the synergy relationship between the two regions is weakened. When the value of the difference in geological characteristics, the difference in hydrological characteristics, or the difference in terrain characteristics decreases, it means that the characteristics between the sub-regions are closer, and the value of the regional synergy index increases, reflecting that the similarity between the sub-regions is enhanced.

[0068] Formulas 2, 3, and 4 calculate the differences in geological, hydrological, and topographic characteristics between subregions, respectively. The formal rationality of these formulas lies in their use of Euclidean distance, which allows for quantification and comprehensive consideration of differences between different characteristics. By taking the square root of the sum of squared differences, the relative differences between the two subregions in their respective characteristic parameters can be effectively reflected. Furthermore, the introduction of weights allows each difference to have a different degree of influence on the overall regional synergy index. This weighting approach allows the model to adapt to the impact of different characteristics on regional synergy, enhancing the flexibility and accuracy of the regional synergy index. Therefore, these formulas are not only mathematically sound but also effectively support the assessment of characteristic similarity in slope stability prediction.

[0069] The spectral clustering algorithm is used to cluster the sub-regions. The specific logic is as follows:

[0070] A regional synergy index matrix is ​​constructed based on the regional synergy index between all sub-regions, and the sub-regions are divided into communities using the spectral clustering algorithm. The specific process is as follows: the Laplace matrix of the regional synergy index matrix is ​​calculated, the optimal number of communities K is determined through eigenvalue decomposition, and the sub-region is divided into K communities using the K-means algorithm; a dynamic similarity threshold τ is set so that the average regional synergy index between sub-regions within a community is less than τ, while the average regional synergy index of sub-regions between communities is greater than τ; for any sub-region, if its regional synergy index with all sub-regions is greater than or equal to the dynamic similarity threshold τ, it is determined to be an isolated risk area;

[0071] The optimal number of communities K is determined by eigenvalue decomposition, and the sub-region is divided into K communities using the K-means algorithm. Specifically, a symmetrical SCI matrix is ​​constructed based on the regional synergy index among all sub-regions, and the matrix element SCI ij This matrix reflects the combined geological, hydrological, and topographic similarity between subregions i and j. Next, the normalized Laplace matrix of this matrix is ​​calculated, and the eigenvalue decomposition (Eigenvalue Decomposition) is used to extract the eigenvectors corresponding to the first K largest eigenvalues. The value of K is automatically determined using the Eigenvalue Gap Method and represents the optimal number of communities. The eigenvectors are then organized into a new matrix row by row, and the row vectors are clustered using the K-means algorithm to form K initial communities. During this process, a similarity threshold τ is dynamically set, requiring that the average SCI within a community is less than τ and the average SCI between communities is greater than τ. This threshold τ is optimized using the silhouette coefficient maximization principle. For each subregion, if its SCI with all adjacent subregions is ≥ τ, it is considered an isolated risk area. The community divisions and the τ value are iteratively adjusted until the affiliation of all subregions remains stable, ultimately outputting stable community divisions and isolated risk area labels. This process uses a spectral clustering algorithm to transform multidimensional feature similarities into spatial clustering relationships, ensuring homogeneity across similar regions while identifying isolated risk units with geological anomalies or hydrological abrupt changes.

[0072] Step 2 effectively characterizes the local slope information by constructing a multidimensional feature vector for each subregion and defining a regional synergy index based on these vectors. This method offers the advantage of comprehensively considering the relationships between geological, hydrological, and topographical parameters across subregions, thereby revealing the mutual influences between regions. By applying the regional synergy index to community delineation, areas with similar characteristics can be more precisely identified, ensuring a more accurate and scientific identification of risk areas.

[0073] Compared with existing technologies, the implementation of Step 2 significantly improves the complexity and accuracy of slope stability prediction. Traditional methods often use single features or static data for analysis, failing to fully consider interrelated multidimensional factors. By introducing a regional synergy index, Step 2 can dynamically assess the similarities between different sub-regions and their potential risk associations. This design quantifies the interdependencies under different geological and hydrological conditions, thereby promoting information integration and scientific decision-making within the overall plan. Ultimately, this will provide a more reliable data foundation for the dynamic safety factor calculation and risk warning in subsequent steps, improving the overall effectiveness of slope management.

[0074] Step 3: Determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, select its centroid subregion as a representative to determine the dynamic safety factor of the community;

[0075] In this embodiment, the dynamic safety factor of each community and isolated risk area is determined based on the slope characteristic dataset, according to the following formula:

[0076]

[0077] In the formula, FS(t) represents the dynamic safety factor of the sub-region at time t, which is used to reflect the immediate stability state of the slope under rainfall conditions; c is the cohesion of the soil in the sub-region, which reflects the shear strength generated by the chemical bonding between soil particles and is determined by indoor direct shear test or triaxial test. γ is the bulk density of the soil in the sub-region, which is used to characterize the normal stress component generated by the deadweight of the sliding body. The sliding body refers to the part of the soil in the sub-region that can slide along the potential sliding surface. h is the depth of the potential sliding surface, which determines the volume and deadweight load of the sliding body. α is the slope angle of the sub-region, u(t) is the pore water pressure of the sub-region at time t, which is solved by coupling the Richards equation with the real-time rainfall intensity. s is the surface curvature of the sub-region, R is the curvature radius of the slope of the sub-region, and their ratio Used to correct local pore water pressure distribution. is the internal friction angle of the soil in the sub-region, which is used to reflect the friction between soil particles. k(θ) is the permeability coefficient related to water content, k s is the saturated permeability coefficient, which is used to characterize the maximum permeability of the soil when it is fully saturated and is determined by indoor permeability tests; θ(t) is the current soil moisture content in the sub-region, θ r is the residual volume moisture content, which indicates the proportion of water in the soil that cannot be discharged by gravity and is determined by the soil type; θ0 and θ xare the initial moisture content and saturated moisture content, respectively. n is an empirical exponent used to control the rate at which the permeability coefficient changes with moisture content. It is related to the pore distribution of the soil, and n = 2. Specifically, k(θ) reflects the permeability of the soil under different moisture content conditions. The larger its value, the stronger the permeability of the soil. The γ*h*sinα term represents the component of the self-weight sliding down the slope. The term represents the rate of change of pore water pressure over time, characterizing the transient seepage force caused by rainfall infiltration.

[0078] The dynamic safety factor of an isolated risk area is the dynamic safety factor of the sub-area corresponding to the isolated risk area, and the dynamic safety factor of a community is the dynamic safety factor of its centroid sub-area.

[0079] In the first given formula, the dependent variable FS(t) directly represents the real-time stability of the slope under rainfall conditions. Its physical meaning is the real-time ratio of the anti-sliding force to the sliding force. The larger the value, the more stable the slope and the stronger the anti-sliding ability. On the contrary, the smaller the value, the more unstable the slope may be, and therefore there is a risk of landslide. The dynamic safety factor provides a quantitative assessment of the slope stability, which helps to monitor and warn potential landslide risks in real time. FS(t) is coupled with the dynamic changes of pore water pressure. The time-varying characteristics of the permeability coefficient k(θ) solve the problem that the traditional safety factor cannot reflect the transient impact of rainfall infiltration. The technical effects are reflected in: (1) capturing the impact of the sudden increase in pore water pressure caused by short-term heavy rainfall on stability; (2) through the curvature correction term Quantify the local amplification or weakening effect of terrain convexity and concavity on water pressure distribution to improve the accuracy of shallow landslide warning. Independent variables include cohesion, soil bulk density, potential sliding surface depth, slope angle, pore water pressure, surface curvature, slope curvature radius, internal friction angle, and permeability coefficient related to water content. The relationship between these independent variables and FS(t) is as follows: cohesion, soil bulk density and slope angle jointly determine the normal stress component and anti-sliding force generated by the sliding body's own weight, thereby affecting the FS(t) value. The pore water pressure and the internal friction angle of the soil jointly affect the sliding force. The increase in pore water pressure will reduce the effective stress of the slope, thereby reducing the anti-sliding ability and causing a decrease in FS(t). The permeability coefficient k(θ) related to water content directly affects the moisture state and anti-sliding ability of the soil. When the soil moisture content changes, the change in permeability will affect the pore water pressure, thereby indirectly affecting FS(t). The ratio of surface curvature to slope curvature radius It will also affect the sliding force. When its ratio increases, FS(t) will decrease.

[0080] Table 1: Dynamic safety factor statistics

[0081] Group slope angle Cohesion internal friction angle pore water pressure Surface curvature Permeability coefficient Dynamic safety factor 1 12 30 40 3 -0.05 <![CDATA[5.0×10 -5 ]]> 2.78 2 15 22 36 5 -0.04 <![CDATA[3.5×10 -5 ]]> 2.41 3 18 25 38 8 -0.03 <![CDATA[2.5×10 -5 ]]> 2.15 4 20 12 32 10 -0.01 <![CDATA[8.0×10 -6 ]]> 1.67 5 22 18 33 12 -0.02 <![CDATA[1.5×10 -5 ]]> 1.88 6 24 16 31 22 0.02 <![CDATA[6.0×10 -6 ]]> 1.27 7 25 10 30 15 0.02 <![CDATA[1.0×10 -5 ]]> 1.42 8 26 14 29 18 0.01 <![CDATA[9.0×10 -6 ]]> 1.33 9 28 20 35 30 0.04 <![CDATA[3.0×10 -6 ]]> 1.24 10 30 15 28 20 0.03 <![CDATA[5.0×10 -6 ]]> 1.18 11 32 7 26 28 0.03 <![CDATA[4.0×10 -6 ]]> 0.89 12 35 8 25 25 0.01 <![CDATA[2.0×10 -5 ]]> 0.95 13 38 9 24 40 0.06 <![CDATA[7.0×10 -7 ]]> 0.65 14 40 5 22 35 0.05 <![CDATA[1.0×10 -6 ]]> 0.72 15 42 6 20 45 0.07 <![CDATA[5.0×10 -7 ]]> 0.53

[0082] See also Figure 3-Figure 5 In this data analysis, the soil bulk density is taken as 18kN / m 3 , the sliding surface depth is 5m, the curvature radius R = 100m, and the pore water pressure change rate From the overall trend, the dynamic safety factor shows a clear negative correlation with the slope angle. When the slope angle increases from 12° to 42°, FS(t) decreases accordingly, from 2.78 to 0.53, a decrease of 81%. This variation pattern is completely consistent with the basic mechanical principles of slope stability, namely, as the slope angle increases, the sliding force component increases significantly, while the anti-sliding force component decreases relatively, resulting in a decrease in overall stability. It is particularly noteworthy that when the slope angle exceeds 30°, FS(t) generally drops below 1.3, entering the stability range requiring early warning. This finding provides an important reference for determining the critical slope angle in engineering practice.

[0083] Secondly, cohesion and internal friction angle, as key parameters of soil strength, have a significant impact on the dynamic safety factor. Data show that under similar slope angle conditions (e.g., in the range of 20-26°), a 5kPa increase in cohesion increases the dynamic safety factor by an average of 0.25-0.35, while a 5° increase in the internal friction angle increases the dynamic safety factor by an average of 0.15-0.2. For example, comparing the fourth and fifth data sets, at similar slope angles, the dynamic safety factor increases from 1.67 to 1.88, demonstrating that improving soil strength parameters can effectively enhance slope stability.

[0084] The influence of hydrogeological parameters cannot be ignored either. Pore water pressure shows a significant negative correlation with the dynamic safety factor. When the pore water pressure increases from 3kPa to 45kPa, the dynamic safety factor decreases by 81%. The influence of the permeability coefficient is dual: on the one hand, low permeability will aggravate the accumulation of pore water pressure, such as the dynamic safety factor of Group 10 is only 0.65; on the other hand, high permeability promotes drainage. For example, in the first group of data, even if the pore water pressure is 3kPa, the dynamic safety factor value is still maintained at a high stability of 2.78. The influence of surface curvature is manifested as follows: convex slopes (i.e., when the surface curvature is greater than zero) reduce the dynamic safety factor by an average of 12-18%, while concave slopes (i.e., when the surface curvature is less than zero) can increase the dynamic safety factor by about 8-15%.

[0085] It is particularly noteworthy that the interaction between parameters has a significant impact on the dynamic safety factor. On steep slopes (when the slope toe is greater than 35°), the impact of changes in hydrological parameters on the dynamic safety factor is amplified. For example, in Groups 13-15, the decrease in the dynamic safety factor (approximately 0.08-0.1) for each 5 kPa increase in pore water pressure is significantly greater than the decrease on gentle slopes (approximately 0.03-0.05). This phenomenon indicates that on steep slopes, changes in water pressure caused by rainfall infiltration pose a more severe threat to stability.

[0086] Finally, critical state analysis revealed that when FS(t) < 1.0, the corresponding parameter combinations all met the following characteristics: slope angle > 35°, cohesion < 10 kPa, and u(t) > 25 kPa. This result provides a clear parameter threshold reference for determining slope hazardous conditions. The significant difference between the maximum FS(t) value of 2.78 (Group 1) and the minimum value of 0.53 (Group 15) across all data sets clearly demonstrates the necessity of considering the influence of multi-parameter coupling.

[0087] For each community, its centroid sub-region is selected as a representative to determine the dynamic safety factor of the community. The centroid sub-region is selected as follows: first, the average coordinates of the geometric centers of all sub-regions in the community are calculated as the centroid position, and then the sub-region with the closest Euclidean distance to the centroid position is selected as the centroid sub-region; if there are multiple sub-regions, the cosine similarity of their multi-dimensional feature vectors and the community average feature is further compared, and the sub-region with the highest similarity is selected as the centroid sub-region, and the centroid sub-region meets the following conditions:

[0088] (1) Located within the community and not adjacent to isolated risk areas;

[0089] (2) The deviation of its geological and hydrological parameters from the community average does not exceed 15%.

[0090] When determining the dynamic safety factor for each community, selecting a centroid subregion as a representative is both scientifically and practically important. First, by calculating the average coordinates of the geometric centers of all subregions within the community, the community's central location can be accurately reflected, providing a basis for subsequent selection. Selecting the subregion with the closest Euclidean distance to the centroid ensures that the selected region is spatially closest to the community core and reflects the overall characteristics of the area. If multiple equally spaced subregions exist, the cosine similarity of their multidimensional feature vectors with the community average is further compared. This method effectively assesses the consistency of the geological and hydrological characteristics of each subregion, ensuring that the selected centroid subregion better represents the overall characteristics of the community. Finally, to ensure the geographical and environmental rationality of the selected centroid subregion, it is required to be located within the community and not adjacent to isolated risk areas, effectively reducing the potential risk of the selected region. Furthermore, the requirement that the deviation of its geological and hydrological parameters from the community average should not exceed 15% further ensures that the selected region's environmental characteristics are consistent with the community's, thereby improving the accuracy and reliability of the dynamic safety factor calculation. Through the above methods, the centroid sub-region can more accurately reflect the actual situation of the community and provide a reliable basis for subsequent safety assessments.

[0091] Compare the dynamic safety factor with the stability threshold. Based on the comparison results, classify the stability levels of each community and isolated risk area. The specific logic is as follows:

[0092] When FS(t) ≥ FY, mark the community or isolated risk area as in a stable state, indicating that the anti-sliding force of the slope at this community or isolated risk area is significantly greater than the sliding force.

[0093] When 0.5*FY ≤ FS(t) < FY, mark the community or isolated risk area as in a warning state, indicating that monitoring needs to be strengthened and personnel entry is restricted.

[0094] When 0.2*FY ≤ FS(t) < 0.5*FY, mark the community or isolated risk area as in a high-risk state, indicating that local sliding may occur and an emergency plan needs to be activated.

[0095] When FS(t) < 0.2*FY, mark the community or isolated risk area as in an unstable state, indicating that a landslide has occurred or is approaching, and immediately trigger an evacuation alarm.

[0096] In the formula, FS(t) represents the dynamic safety factor at time t, and FY is the preset stability threshold.

[0097] In step 3, determine the dynamic safety factor of each community and isolated risk area according to the slope feature dataset. The advantage of this process is that it can reflect the stability state of the slope under rainfall conditions in real time. The introduction of the dynamic safety factor not only considers the influence of various environmental factors but also realizes the dynamic monitoring of each sub-region or community, enabling the risk assessment to respond to environmental changes in a timely manner. By comparing with the stability threshold, the risk states of each region can be effectively classified, providing a clear management basis.

[0098] Compared with the existing technology, the implementation of step 3 improves the real-time performance and accuracy of slope stability prediction. Traditional methods often rely on static or historical data for evaluation and tend to ignore the impact of instantaneous events such as rainfall on slope stability. Through the calculation of the dynamic safety factor in step 3, the evaluation results have higher timeliness and can accurately reflect the changes in the safety factor caused by factors such as rainfall. This mechanism provides an important basis for subsequent risk warning and decision-making, ensuring the scientific nature and effectiveness of slope management, thus playing a key promoting role in the overall plan and promoting the establishment of a more refined and intelligent slope monitoring and management system.

[0099] Step 4: Build a 3D digital twin model of the target slope. Use a spatial mapping algorithm to accurately match the community division results with the model. Use different colors to identify the stability level of each community and isolated risk area in the visualization interface. For communities or isolated risk areas where the dynamic safety factor shows a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the 3D digital twin model.

[0100] In this embodiment, a three-dimensional digital twin model of the target slope is constructed based on the BIM platform. The community division results are accurately matched with the model grid through spatial coordinate mapping. In the visualization interface, a four-color gradient rendering scheme is used to display the stability level of each community and isolated risk area in real time, and the real-time rainfall intensity distribution is superimposed. Among them, green, yellow, orange, and red represent stable state, warning state, high-risk state, and unstable state, respectively.

[0101] For areas where the dynamic safety factor shows a monotonically decreasing trend over three consecutive monitoring cycles, an audible and visual warning signal is automatically triggered. At the same time, the area is marked in the model with a pulse flashing effect, and a warning report is generated containing the location coordinates, the rate of change of the dynamic safety factor, and the potential slip direction. The system simultaneously records the stability evolution trajectory of each area and supports the display of the landslide development process in a timeline playback manner.

[0102] Step 4 creates a 3D digital twin model of the target slope and accurately matches the community division results with the model, enabling a visual display of the slope's status. This process offers the advantage of not only visually displaying the stability levels of different communities and isolated risk areas, but also overlaying rainfall intensity distribution in real time, providing comprehensive information support. The multi-color visual interface allows decision makers to quickly identify potential risk areas and implement appropriate management measures, improving decision-making efficiency and response speed.

[0103] Compared with existing technologies, the implementation of Step 4 significantly improves the intelligence and operability of slope monitoring. Traditional methods often rely on static reports or single data views, which make it difficult to fully reflect the dynamic changes of slopes. The introduction of a three-dimensional digital twin model, combined with real-time data, enables dynamic monitoring and visualization of slope conditions, making risk assessment more scientific and accurate. This mechanism not only improves managers' understanding of slope stability but also provides an intuitive basis for early warning of potential risks and emergency response, thus playing a key role in the overall plan and promoting the modernization of slope monitoring and management.

[0104] See also Figure 2 , a device for predicting slope stability under rainfall conditions, comprising:

[0105] The slope grid modeling module is used to discretize the target slope into several sub-regions using a regular grid, assign regional attributes including geological parameters, hydrological parameters, and topographic parameters to each sub-region, and integrate the real-time rainfall intensity data provided by the meteorological radar to form a slope characteristic dataset with spatiotemporal attributes;

[0106] The intelligent community division module is used to characterize the local slope information of each sub-region using a multidimensional feature vector based on the geological parameters, hydrological parameters, and topographic parameters of each sub-region. The regional synergy index between any two sub-regions is defined based on the multidimensional feature vector, and the sub-regions are clustered using a spectral clustering algorithm. The remaining sub-regions are marked as isolated risk areas, thereby dividing the target slope into multiple communities and isolated risk areas.

[0107] The dynamic safety assessment module is used to determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, its centroid sub-region is selected as a representative to determine the dynamic safety factor of the community;

[0108] The three-dimensional early warning visualization module is used to establish a three-dimensional digital twin model of the target slope, accurately match the community division results with the model through a spatial mapping algorithm, and identify the stability level of each community and isolated risk area with different colors in the visualization interface. For communities or isolated risk areas whose dynamic safety factors show a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the three-dimensional digital twin model.

[0109] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0110] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0111] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0112] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for predicting slope stability under rainfall conditions, characterized in that: The specific steps include: Step 1: Use a regular grid to discretize the target slope into several sub-regions. Each sub-region is assigned regional attributes including geological parameters, hydrological parameters, and topographic parameters. At the same time, the real-time rainfall intensity data provided by the meteorological radar is integrated to form a slope characteristic dataset with spatiotemporal attributes. Step 2: Based on the geological, hydrological, and topographic parameters of each sub-region, a multidimensional feature vector is used to characterize the local slope information of each sub-region. The regional synergy index between any two sub-regions is defined based on the multidimensional feature vector. The sub-regions are clustered using a spectral clustering algorithm to divide the target slope into multiple communities and isolated risk areas. Step 3: Determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, select its centroid subregion as a representative to determine the dynamic safety factor of the community; Step 4: Establish a three-dimensional digital twin model of the target slope, accurately match the community division results with the model through a spatial mapping algorithm, and use different colors to identify the stability level of each community and isolated risk area in the visualization interface. For communities or isolated risk areas whose dynamic safety factors show a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the three-dimensional digital twin model.

2. The method for predicting slope stability under rainfall conditions according to claim 1, characterized in that: The target slope is discretized into several sub-areas using a regular grid. The grid size is adjustable from 5m×5m to 20m×20m based on the slope scale and monitoring accuracy requirements. Each sub-area is assigned regional attributes including geological parameters, hydrological parameters, and topographic parameters. The geological parameters include permeability, internal friction angle, and cohesion; the hydrological parameters include initial water content, saturated water content, and real-time pore water pressure; and the topographic parameters include slope and surface curvature. The real-time rainfall intensity data provided by the meteorological radar is integrated, and its temporal and spatial resolution matches the grid division scale, ultimately forming a slope characteristic dataset with multi-dimensional characteristics of time, space and attributes. The parameters are stored in matrix form and a temporal and spatial index relationship is established. Geological parameters, hydrological parameters, and topographic parameters are obtained through field surveys, sensor monitoring, and remote sensing measurements as regional attributes of each sub-region.

3. The method for predicting slope stability under rainfall conditions according to claim 2, characterized in that: According to the geological parameters, hydrological parameters and topographic parameters of each sub-region, a multidimensional feature vector is constructed for each sub-region. The expression of the multidimensional feature vector is as follows: Where, F i is the multidimensional feature vector of subregion i, i is the index of the subregion, f geo,i 、f hydro,i and f top,i represent the geological parameters, hydrological parameters and topographic parameters of sub-region i, respectively, k i 、 and c i are the permeability coefficient, internal friction angle and cohesion of sub-region i, θ 0,i ,θ x,i and u i are the initial water content, saturated water content and real-time pore water pressure of sub-region i, α i and β i are the slope and surface curvature of sub-area i, respectively; The regional synergy index between any two sub-regions is defined based on the multi-dimensional feature vector, and the formula is as follows: In the formula, SCI ij represents the regional synergy index between sub-region i and sub-region j, i and j are the indexes of sub-regions, and i≠j; Dgeo(f geo,i ,f geo,j )、Dhydro(f hydro,i ,f hydro,j ) and Dtop(f top,i ,f top,j ) represent the differences in geological characteristics, hydrological characteristics, and topographic characteristics between sub-regions i and j, respectively; f geo,j 、f hydro,j and f top,j represent the geological parameters, hydrological parameters and topographic parameters of sub-region j, respectively, k j 、 and c h are the permeability coefficient, internal friction angle and cohesion of sub-region j, θ 0,j ,θ x,j and u j are the initial water content, saturated water content and real-time pore water pressure of sub-region j, α j and β j are the slope and surface curvature of sub-region j, ω1, ω2 and ω3 are preset weights, ω1>ω2>ω3>0, and ω1+ω2+ω3=1.

4. The method for predicting slope stability under rainfall conditions according to claim 1, characterized in that: The spectral clustering algorithm is used to cluster the sub-regions. The specific logic is as follows: Construct a regional collaboration index matrix based on the regional collaboration indices between all sub-regions, and use the spectral clustering algorithm to divide the sub-regions into communities. The specific process is as follows: Calculate the Laplacian matrix of the regional collaboration index matrix, determine the optimal number of communities K through eigenvalue decomposition, and use the K-means algorithm to divide the sub-regions into K communities; Set a dynamic similarity threshold τ such that the average regional collaboration index between sub-regions within a community is less than τ, while the average regional collaboration index between sub-regions of different communities is greater than τ; For any sub-region, if its regional collaboration indices with all sub-regions are greater than or equal to the dynamic similarity threshold τ, it is determined as an isolated risk area; Repeat the above similarity judgment until the community affiliations of all sub-regions no longer change, forming the final community division result.

5. The method for predicting slope stability under rainfall conditions according to claim 1, characterized in that: Determine the dynamic safety factors of each community and isolated risk area based on the slope feature dataset, according to the following formula: Where FS(t) represents the dynamic safety factor of the sub-region at time t, which is used to reflect the immediate stability state of the slope under rainfall conditions; c is the cohesion of the soil in the sub-region, γ is the bulk density of the soil in the sub-region, which is used to characterize the normal stress component generated by the weight of the sliding body. The sliding body refers to the part of the soil in the sub-region that can slide along the potential sliding surface; h is the depth of the potential sliding surface, α is the slope angle of the sub-region, u(t) is the pore water pressure of the sub-region at time t, s is the surface curvature of the sub-region, and R is the curvature radius of the slope of the sub-region. is the internal friction angle of the sub-region soil, k(θ) is the permeability coefficient related to the water content, k s is the saturated permeability coefficient, which is used to characterize the maximum permeability of the soil when it is fully saturated and is determined by indoor permeability tests; θ(t) is the current soil moisture content in the sub-region, θ r is the residual volume moisture content, which indicates the proportion of water in the soil that cannot be discharged by gravity and is determined by the soil type; θ0 and θ x are the initial water content and saturated water content, respectively; n is an empirical index used to control the rate at which the permeability coefficient changes with water content and is related to the soil pore distribution; For each community, select its centroid sub-region as a representative to determine the dynamic safety factor of the community. The method for selecting the centroid sub-region is as follows: First, calculate the average coordinates of the geometric centers of all sub-regions within the community as the centroid position, and then select the sub-region with the closest Euclidean distance to this centroid position as the centroid sub-region; If there are multiple sub-regions at the same time, further compare the cosine similarity between their multi-dimensional feature vectors and the average feature of the community, and select the sub-region with the highest similarity as the centroid sub-region, and this centroid sub-region satisfies the following conditions: (1) It is located inside the community and is not adjacent to the isolated risk area; (2) The deviation of its geological parameters and hydrological parameters from the community average does not exceed 15%.

6. The method for predicting slope stability under rainfall conditions according to claim 5, characterized in that: Compare the dynamic safety factor with the stability threshold. According to the comparison results, divide the stability levels of each community and isolated risk area. The specific logic is as follows: When FS(t)≥FY, mark this community or isolated risk area as a stable state, indicating that the anti-sliding force of the slope in this community or isolated risk area is significantly greater than the sliding force; When 0.5*FY≤FS(t)<FY, mark this community or isolated risk area as a warning state, strengthen monitoring and restrict personnel entry; When 0.2*FY≤FS(t)<0.5*FY, mark this community or isolated risk area as a high-risk state and activate the emergency plan; When FS(t)<0.2*FY, mark this community or isolated risk area as an unstable state, indicating that a landslide has occurred or is approaching, and immediately trigger an evacuation alarm; In the formula, FS(t) represents the dynamic safety factor at time t, and FY is the preset stability threshold.

7. The method for predicting slope stability under rainfall conditions according to claim 1, characterized in that: Construct a three-dimensional digital twin model of the target slope based on the BIM platform, and accurately match the community division result with the model grid through spatial coordinate mapping; In the visualization interface, use a four-color gradient rendering scheme to display the stability levels of each community and isolated risk area in real time, and overlay and display the real-time rainfall intensity distribution; Among them, green, yellow, orange, and red represent stable state, warning state, high-risk state, and unstable state respectively; For areas where the dynamic safety factor shows a monotonically decreasing trend over three consecutive monitoring cycles, an audible and visual warning signal is automatically triggered. At the same time, the area is marked in the model with a pulse flashing effect, and a warning report is generated containing the location coordinates, the rate of change of the dynamic safety factor, and the potential slip direction. The system simultaneously records the stability evolution trajectory of each area and supports the display of the landslide development process in a timeline playback manner.

8. A device for predicting slope stability under rainfall conditions, characterized by: The device for predicting slope stability under rainfall conditions is used to execute the method for predicting slope stability under rainfall conditions according to any one of claims 1 to 7, comprising: The slope grid modeling module is used to discretize the target slope into several sub-regions using a regular grid, assign regional attributes including geological parameters, hydrological parameters, and topographic parameters to each sub-region, and integrate the real-time rainfall intensity data provided by the meteorological radar to form a slope characteristic dataset with spatiotemporal attributes; The intelligent community division module is used to characterize the local slope information of each sub-region using a multidimensional feature vector based on the geological parameters, hydrological parameters, and topographic parameters of each sub-region. The module also defines the regional synergy index between any two sub-regions based on the multidimensional feature vector and clusters the sub-regions using a spectral clustering algorithm. This allows the target slope to be divided into multiple communities and isolated risk areas. The dynamic safety assessment module is used to determine the dynamic safety factor of each community and isolated risk area based on the slope characteristic dataset, compare the dynamic safety factor with the stability threshold, and classify the stability level of each community and isolated risk area based on the comparison results. For each community, its centroid sub-region is selected as a representative to determine the dynamic safety factor of the community; The three-dimensional early warning visualization module is used to establish a three-dimensional digital twin model of the target slope, accurately match the community division results with the model through a spatial mapping algorithm, and identify the stability level of each community and isolated risk area with different colors in the visualization interface. For communities or isolated risk areas whose dynamic safety factors show a monotonically decreasing trend over three consecutive monitoring cycles, a risk warning signal is triggered, and the evolution trajectory of the area is highlighted in the three-dimensional digital twin model.

Images