Slope stability prediction method in heavy rainfall weather

Abstract

The invention discloses a slope stability prediction method in heavy rainfall weather, and relates to the technical field of road slopes. The method comprises the following steps: collecting soil characteristic data before and after rainfall in a road slope sub-region, and fitting a water-soil characteristic function and a saturated permeation function through a least square method; the method comprises the following steps: establishing a road slope three-dimensional model by using a laser radar, obtaining soil parameters of each point of the road slope three-dimensional model based on an inverse distance square weighting method, and obtaining simulated initial soil matrix suction and simulated initial unsaturated permeability coefficients; simulating moisture movement through an unsaturated seepage control equation to obtain simulated soil moisture saturation and simulated pore water pressure of each point, screening risk points through a preset threshold value, and generating risk clusters through density clustering; and searching a potential sliding surface in the risk cluster by adopting a particle swarm optimization algorithm, calculating the safety coefficient of the sliding surface in combination with a soil strip method, and quantifying the area of the sliding surface. And fusing the safety coefficient and the area proportion to calculate a slope risk index and divide stability grades.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of highway slope, in particular to a slope stability prediction method in heavy rainfall weather. BACKGROUND

[0002] In the field of transportation infrastructure construction and operation, the stability of highway slope is significantly affected by heavy rainfall weather, and the accurate prediction of its dynamic evolution rule has always been the core problem of industry concern. With the continuous extension of China's highway network to complex terrain areas, slope instability events induced by heavy rainfall occur frequently, which not only seriously threatens traffic safety, but also causes huge economic losses. At present, the industry generally uses traditional monitoring methods combined with numerical simulation to evaluate the stability of the slope, such as obtaining soil moisture content, pore water pressure and other data by laying single-point sensors, and performing stability analysis based on simplified mechanical models.

[0003] However, the real-time evolution rule of soil mechanical parameters with moisture content during heavy rainfall is usually ignored in the prior art, and the static parameters of soil mechanical parameters in the dry state are still used, resulting in a large deviation between the potential sliding surface search result and the actual working condition, which cannot meet the high-precision early warning demand of slope instability in heavy rainfall weather. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a slope stability prediction method in heavy rainfall weather to solve the problems in the background art.

[0005] To achieve the above purpose, the present application is realized by the following technical scheme: a slope stability prediction method in heavy rainfall weather, comprising the following steps:

[0006] Step S1: sampling the soil of the highway slope sub-area before and after rainfall to obtain soil property data of the highway slope sub-area;

[0007] Step S2: curve fitting soil moisture saturation and soil matric suction in the soil property data of the highway slope sub-area by the least square method to obtain a water-soil characteristic function, curve fitting soil moisture saturation and unsaturated permeability coefficient in the soil property data of the highway slope sub-area to obtain a saturated permeability function, and curve fitting soil moisture saturation and soil cohesion in the soil property data of the highway slope sub-area to obtain a soil cohesion function;

[0008] Step S3: Three-dimensional modeling of the highway slope is performed by a UAV-mounted laser radar to obtain a three-dimensional model of the highway slope; based on the soil property data of the highway slope sub-regions, soil property data of each point in the three-dimensional model of the highway slope is obtained according to an inverse distance square weighting method, and the soil property data of each point in the three-dimensional model of the highway slope is input into a water-soil characteristic function to obtain simulated initial soil matric suction;

[0009] Step S4: Based on the simulated initial soil matric suction and simulated initial unsaturated permeability coefficient, water movement in the three-dimensional model of the highway slope is simulated by an unsaturated seepage control equation to obtain simulated soil moisture saturation and simulated soil pore water pressure of each point in the three-dimensional model of the highway slope;

[0010] Step S5: The three-dimensional model of the highway slope is screened based on the simulated soil moisture saturation and simulated soil pore water pressure to obtain risk points of the three-dimensional model of the highway slope; the risk points of the three-dimensional model of the highway slope are clustered by a density clustering method to obtain risk clusters of the three-dimensional model of the highway slope;

[0011] Step S6: A particle swarm optimization algorithm is used to search for a potential sliding surface in the risk clusters of the three-dimensional model of the highway slope, and a soil bar method is used to calculate a sliding surface safety factor of the potential sliding surface; the area of the potential sliding surface is obtained by calculating the area of the potential sliding surface; the highway slope risk index is calculated by combining the area of the potential sliding surface and the sliding surface safety factor, the highway slope risk index is compared with a preset threshold value, and a stability level is divided to realize prediction of the slope stability.

[0012] Preferably, the soil moisture saturation and soil matric suction in the soil property data of the highway slope sub-regions are curve-fitted by a least squares method to obtain a water-soil characteristic function, including the following specific steps:

[0013] The soil matric suction and soil moisture saturation during a dry period, a rainstorm peak period, and 24 hours after rain stop in the soil property data of the highway slope sub-regions are curve-fitted by a least squares method to obtain a water-soil characteristic function:

[0014]

[0015] wherein, represents the soil moisture saturation when the soil matric suction is represents the soil matric suction. represents the suction coefficient of the kth highway slope sub-region, is a curve shape parameter of the kth highway slope sub-region, is a derivative parameter of the kth highway slope sub-region, represents the soil matric suction.

[0016] Preferably, the soil moisture saturation and unsaturated permeability coefficient in the soil property data of the highway slope sub-region are curve fitted to obtain a saturated permeability function, including the following steps:

[0017] The soil moisture saturation and unsaturated permeability coefficient in the soil property data of the highway slope sub-region during the dry period, the heavy rain peak period and 24 hours after the rain stop are curve fitted by the least square method to obtain an unsaturated permeability function:

[0018]

[0019] wherein, the unsaturated permeability coefficient when the soil moisture saturation is , is the saturated permeability coefficient of the kth highway slope sub-region, is the soil moisture saturation, is an empirical index, and the default is 0.5, is the derived parameter of the kth highway slope sub-region, and the same as the of the soil property function.

[0020] Preferably, based on the soil property data of the highway slope sub-region, the soil property data of each point in the highway slope three-dimensional model is obtained according to the inverse distance square weighting method, including the following steps:

[0021] For each point (x, y, z) in the highway slope three-dimensional model of the highway slope, the simulated suction coefficient of each point (x, y, z) in the highway slope three-dimensional model is obtained by the inverse distance square weighting method:

[0022]

[0023] wherein, indicates the simulated suction coefficient of the highway slope at the coordinates (x, y, z), is the distance from the unmeasured point (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, indicates the simulated suction coefficient of the kth highway slope sub-region;

[0024] For each point (x, y, z) in the highway slope three-dimensional model of the highway slope, the simulated curve shape parameter and the simulated derived parameter of the soil and water feature curve of each point (x, y, z) in the highway slope three-dimensional model are obtained by the inverse distance square weighting method:

[0025]

[0026]

[0027] wherein, is a simulated curve shape parameter of the highway slope at coordinate (x, y, z), is a simulated derivative parameter of the highway slope at coordinate (x, y, z), is a distance from coordinate (x, y, z) of the highway slope to the kth highway slope sub-region, k represents an index of the kth highway slope sub-region, represents a simulated curve shape parameter of the kth highway slope sub-region;

[0028] For each point (x, y, z) in the highway slope three-dimensional model of the highway slope, the simulated saturated permeability coefficient of each point (x, y, z) in the highway slope three-dimensional model of the highway slope is obtained by inverse distance square weighting method:

[0029]

[0030] wherein, represents a simulated saturated permeability coefficient of the highway slope at coordinate (x, y, z), is a distance from coordinate (x, y, z) of the highway slope to the kth highway slope sub-region, k represents an index of the kth highway slope sub-region, K represents a total number of highway slope sub-regions, represents a simulated saturated permeability coefficient of the kth highway slope sub-region;

[0031] For each point (x, y, z) in the highway slope three-dimensional model of the highway slope, the simulated porosity of each point (x, y, z) in the highway slope three-dimensional model of the highway slope is obtained by inverse distance square weighting method:

[0032]

[0033] wherein, represents a simulated porosity of the highway slope at coordinate (x, y, z), is a distance from coordinate (x, y, z) of the highway slope to the kth highway slope sub-region, k represents an index of the kth highway slope sub-region, K represents a total number of highway slope sub-regions, represents a simulated porosity of the kth highway slope sub-region;

[0034] The simulated initial soil moisture saturation of each point (x, y, z) in the highway slope three-dimensional model of the highway slope is also calculated by inverse distance square weighting method , the simulated cohesion coefficient , the simulated initial internal friction angle , the simulated internal friction angle decay coefficient .

[0035] Preferably, the step of inputting the soil characteristic data of each point in the three-dimensional model of the highway slope into the soil and water characteristic function to obtain the simulated initial soil matrix suction includes the following steps:

[0036] The initial soil moisture saturation is simulated using soil property data at each point in the three-dimensional model of the highway slope. Simulated curve shape parameters and simulated suction coefficient The water and soil characteristic functions during the dry period are input to obtain the simulated initial soil matrix suction. .

[0037] Preferably, the step of simulating water movement in the three-dimensional model of the highway slope based on the simulated initial soil matrix suction and saturated permeability function, and obtaining the simulated soil moisture saturation and simulated soil pore water pressure at each point in the three-dimensional model of the highway slope through the unsaturated seepage control equation, includes the following steps:

[0038] The unsaturated seepage control equation simulates water flow on a highway slope under initial dry conditions. Therefore, the inputs needed are the simulated initial soil matrix suction and the simulated initial unsaturated permeability coefficient at each point in the 3D model of the highway slope. The initial condition for the unsaturated seepage control equation is: at t=0, , It is the pressure head at coordinates (x, y, z) at time t=0. = , The density of water is 1000 kg / m³. Where g is the acceleration due to gravity, the soil property data at each point in the three-dimensional model of the highway slope are... Simulate initial soil moisture saturation Simulated curve shape parameters and simulated saturated permeability coefficient The water and soil characteristic function during the dry period is input to obtain the simulated initial unsaturated permeability coefficient. ;

[0039] The simulated initial soil matrix suction and simulated initial unsaturated permeability coefficient at each point in the 3D model of the highway slope are used to simulate water movement through the unsaturated seepage control equation:

[0040]

[0041] in, It is a function of specific water capacity. = , It is the simulated volumetric water content at coordinates (x, y, z) at time t. It is an unsaturated osmotic function. is the pressure head at coordinate (x, y, z) at time t, and z is the elevation head;

[0042] The unsaturated seepage control equation is solved by the finite element method to obtain the pressure head h(x, y, z, t) of each point (x, y, z) at time t. Finally, the following is obtained: is the simulated volumetric water content of the highway slope at coordinate (x, y, z) at time t, pore(x, y, z);

[0043] The simulated soil moisture saturation of each point in the highway slope three-dimensional model is:

[0044]

[0045] wherein, is the simulated soil moisture saturation of the highway slope at coordinate (x, y, z) at time t, is the simulated volumetric water content of the highway slope at coordinate (x, y, z) at time t, is the simulated porosity of the highway slope at coordinate (x, y, z);

[0046] The simulated soil pore water pressure of each point (x, y, z) in the highway slope three-dimensional model of the highway slope is:

[0047]

[0048] wherein, is the simulated soil pore water pressure of the highway slope at coordinate (x, y, z) at time t, is the water density, 1000 kg / m3, g is the acceleration of gravity, () represents the inverse function of the water-soil characteristic function.

[0049] Preferably, the screening of the highway slope three-dimensional model by the simulated soil moisture saturation and the simulated soil pore water pressure to obtain the risk points of the highway slope three-dimensional model comprises the following specific steps:

[0050] After the construction of the highway slope three-dimensional model and the acquisition of the simulated soil moisture saturation and the simulated soil pore water pressure of each spatial point, the simulated soil moisture saturation and the simulated soil pore water pressure of each point in the highway slope three-dimensional model are threshold screened, and the points with the simulated soil moisture saturation and the simulated soil pore water pressure greater than the preset threshold are taken as the risk points of the highway slope three-dimensional model. ​​

[0051] Preferably, the risk points of the highway slope three-dimensional model are clustered by the density clustering method to obtain highway slope three-dimensional model risk clusters, including the following specific steps:

[0052] The risk points of the highway slope three-dimensional model are clustered by the density clustering method to obtain highway slope three-dimensional model risk clusters, the neighborhood radius and the minimum sample number in the density clustering method are initialized, the points between the highway slope three-dimensional model risk points whose Euclidean distance satisfies less than the neighborhood radius are regarded as points in the neighborhood, that is, these points are considered to be close enough in physical space and have strong relevance; the number of points in the neighborhood is used to identify core points, boundary points and noise points: if the number of points contained in the neighborhood of a certain node is not less than the set minimum sample number, the node is determined to be a core point, and the core point is a key node for clustering; the boundary point belongs to the node whose number of points in the neighborhood is less than the minimum sample number but itself is in the neighborhood of a certain core point; and the node which is neither a core point nor a boundary point is a noise point, and the relevance of such node with other nodes is weak; adjacent core point clusters are merged, that is, the cluster formed by the core points having density connection with each other is integrated into a complete clustering cluster, thereby generating the final highway slope three-dimensional model risk cluster.

[0053] Preferably, the sliding surface safety factor of the potential sliding surface is calculated by the soil strip method, including the following specific steps:

[0054] The sliding surface safety factor of each potential sliding surface in the highway slope three-dimensional model risk cluster is calculated:

[0055]

[0056] wherein, is the sliding surface safety factor of the sliding surface at t, i is the index of the i th soil strip, and N is the total number of soil strips, is the simulated soil cohesion of the i th soil strip at t, is the bottom length of the i th soil strip, is the weight of the i th soil strip, is the inclination angle of the i th soil strip, is the simulated soil pore water pressure of the i th soil strip, is the simulated internal friction angle of the i th soil strip at t.

[0057] Preferably, the highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor, including the following specific steps:

[0058] The highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor:

[0059]

[0060] wherein, is the highway slope risk index at time t, is the maximum value function, and is the maximum value of the sliding surface safety factor of all potential sliding surfaces, is the sliding surface safety factor of the potential sliding surface of the Jth highway slope three-dimensional model risk cluster, is the potential sliding surface area of the potential sliding surface of the Jth highway slope three-dimensional model risk cluster, is the total number of potential sliding surfaces.

[0061] The present application provides a slope stability prediction method in strong rainfall weather, relates to machine learning and deep learning technology, and has the following beneficial effects:

[0062] (1) The particle swarm optimization algorithm is used to search for potential sliding surfaces in the risk cluster, and the safety factor is calculated by combining the soil strip method. The core significance lies in realizing the dynamic and accurate quantification of slope instability risk. The particle swarm optimization algorithm quickly locates the most dangerous sliding surface (spherical surface equation parameterization) in the three-dimensional risk cluster through the global optimization mechanism (with the reciprocal of the sliding surface stability as the fitness), breaking through the limitations of traditional experience assumptions. The soil strip method calculates the sliding surface safety factor by discretizing the rock and soil column (the bottom surface is attached to the sliding surface, and the top surface is matched to the ground surface), coupling the real-time simulation of pore water pressure and strength decay parameters (cohesion index decay, internal friction angle linear reduction), and thus converting the hydro-mechanical dynamic response into a quantitative stability index, providing accurate basis for the position of the critical sliding surface and the instability probability for early warning decision-making.

[0063] (2) Based on the contour coefficient, high-consistency trajectory clusters are screened, which significantly improves the reliability of the early warning target. The contour coefficient calculates the overall consistency of the node features in the clustering cluster (such as displacement direction concentration and speed synchronization), and only keeps the clusters that meet the index standards. These high-consistency clusters represent dangerous areas with strong deformation coordination and clear evolution trend inside the slope (such as the slope toe shear failure zone), and exclude interference signals caused by temporary construction vibration or local collapse. The screening mechanism greatly reduces the data size of subsequent analysis, so that the computing resources are focused on the core area that truly threatens the stability of the slope.

[0064] (3) The highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor. The comprehensive evaluation index can be constructed by quantifying the "risk influence range x safety coefficient critical value". This index dynamically relates the geometric characteristics of the spherical sliding surface to the soil anti-sliding capacity, and combines the four-level stability evaluation system to quantitatively determine the spatial scale and severity of the slope instability risk in real time. BRIEF DESCRIPTION OF DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0066] Fig. 1 A flow chart of steps of a slope stability prediction method in strong rainfall weather is provided in the present application.

[0067] Fig. 2 A hierarchical diagram of steps of a slope stability prediction method in strong rainfall weather is provided in the present application. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the present application.

[0069] Please refer to Figs. 1-2 The present application provides a technical solution: a slope stability prediction method in strong rainfall weather.

[0070] Step S1: sampling the soil of a highway slope sub-region before and after rainfall to obtain soil property data of the highway slope sub-region.

[0071] The soil of a highway slope sub-region is sampled before and after rainfall to obtain soil property data of the highway slope sub-region, wherein the soil property data of the highway slope sub-region includes initial volume water content, initial pore water pressure, porosity, initial cohesion, initial internal friction angle, initial soil matric suction, and initial soil moisture saturation of each highway slope sub-region in dry period, and volume water content, pore water pressure, cohesion, internal friction angle, soil matric suction, and soil moisture saturation in rainstorm peak period and 24 hours after rain stop.

[0072] It should be noted that the highway slope sub-region includes slope toe (moisture collection area), slope waist (stress concentration zone), and slope top (tension crack prone area) of the highway slope, and a multi-parameter soil detection device is arranged in the highway slope sub-region to form a monitoring network covering hydro-mechanical sensitive points.

[0073] It should be noted that the measurement method of the soil property data of the sub-regions of the highway slope is as follows: by arranging high-precision TDR soil moisture sensors in the sub-regions of the highway slope, the initial volume moisture content in the dry period is recorded in real time, and the initial pore water pressure is measured by a synchronously arranged pore water pressure gauge. The determination of the porosity is performed by using a cutting ring method, and the undisturbed soil sample is taken at the sensor arrangement point for determination. The initial cohesion and the initial internal friction angle are determined by using a direct shear apparatus and a triaxial apparatus. The initial soil matric suction is determined by a pressure plate apparatus experiment, and the initial soil moisture saturation is calculated from the ratio of the initial volume moisture content to the porosity. The volume moisture content and the pore water pressure at the rainstorm peak period and 24 hours after the rain stop are also collected in real time by using the TDR soil moisture sensor and the pore water pressure gauge at the corresponding time period, the cohesion and the internal friction angle are determined by using a direct shear apparatus and a triaxial apparatus on the undisturbed soil sample drilled in the fixed sub-regions of the highway slope at the rainstorm peak period and 24 hours after the rain stop, the soil matric suction is determined by a pressure plate apparatus experiment at the two time periods, and the soil moisture saturation is obtained by converting the volume moisture content at the corresponding time period.

[0074] Step S2: curve fitting is performed on the soil moisture saturation and the soil matric suction in the soil property data of the sub-regions of the highway slope by using a least square method to obtain a water-soil characteristic function, curve fitting is performed on the soil moisture saturation and the unsaturated permeability coefficient in the soil property data of the sub-regions of the highway slope to obtain a saturation permeability function, and curve fitting is performed on the soil moisture saturation and the soil cohesion in the soil property data of the sub-regions of the highway slope to obtain a soil cohesion function.

[0075] The soil matric suction and the soil moisture saturation at the dry period, the rainstorm peak period and 24 hours after the rain stop in the soil property data of the sub-regions of the highway slope are curve fitted by using a least square method to obtain a water-soil characteristic function:

[0076]

[0077] wherein, represents the soil moisture saturation when the soil matric suction is represents the soil matric suction, represents the suction coefficient of the kth sub-region of the highway slope, is a curve shape parameter of the kth sub-region of the highway slope, is a derived parameter of the kth sub-region of the highway slope, represents the soil matric suction.

[0078] It should be noted that the suction coefficient and the curve shape parameter can be obtained by a pressure plate apparatus experiment, and the suction coefficient The greater, the soil is quickly saturated at lower suction, suction coefficient The smaller, the soil needs high suction to be saturated, the soil-water characteristic function reveals the relationship between soil "soil matric suction-soil moisture saturation".

[0079] Through the least squares method to the highway slope sub-regional soil characteristics data in the dry period, rainstorm peak and 24 hours after the rain, the soil moisture saturation and unsaturated permeability coefficient curve fitting, get the unsaturated permeability function:

[0080]

[0081] In which, The unsaturated permeability coefficient when the soil moisture saturation is , The saturated permeability coefficient of the kth highway slope sub-region, The soil moisture saturation, The empirical index, default is 0.5, The derived parameter of the kth highway slope sub-region, the same as the of the soil-water characteristic function.

[0082] It should be noted that, Can be obtained by variable head permeability experiment, the unsaturated permeability function represents the relationship between "soil moisture saturation-unsaturated permeability coefficient".

[0083] Through the least squares method to the highway slope sub-regional soil characteristics data in the dry period, rainstorm peak and 24 hours after the rain, the soil moisture saturation and soil cohesion curve fitting, get the soil cohesion function:

[0084]

[0085] In which, The soil cohesion when the soil moisture saturation is , The dry state cohesion, The cohesion coefficient of the kth highway slope sub-region, The soil moisture saturation.

[0086] Step S3: three-dimensional modeling of the highway slope by unmanned aerial vehicle carrying laser radar, get the highway slope three-dimensional model; based on the highway slope sub-regional soil characteristics data, according to the inverse distance square weighted method to get the soil characteristics data of each point in the highway slope three-dimensional model, input the soil characteristics data of each point in the highway slope three-dimensional model into the soil-water characteristic function, get the simulation initial soil matric suction.

[0087] A three-dimensional model of the highway slope is obtained by using a laser radar carried by a UAV to model the highway slope.

[0088] The UAV carrying the laser radar flies along a preset flight route, scans the slope surface by emitting laser pulses, synchronously records the reflected signals and flight posture (GNSS / IMU positioning), and generates raw point cloud data. The point cloud is denoised (removing vegetation and moving objects) and georeferenced (correcting coordinates by combining ground control points), to obtain an accurate set of three-dimensional ground points. Based on the point cloud data, a triangular mesh surface is constructed, and a continuous three-dimensional entity model is formed through surface fitting algorithm, and key terrain features (such as scarps and cracks) need to be locally encrypted and optimized. By comparing the coordinates of the measured inspection points with the model data, the error is evaluated (usually the planar accuracy is required to be ≤5 cm), and finally a three-dimensional model of the highway slope with spatial position attributes is generated.

[0089] For each point (x, y, z) in the three-dimensional model of the highway slope, the simulated suction coefficient of each point (x, y, z) in the three-dimensional model of the highway slope is obtained by inverse distance square weighting method:

[0090]

[0091] wherein, represents the simulated suction coefficient of the highway slope at coordinates (x, y, z), is the distance from the unmeasured point (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated suction coefficient of the kth highway slope sub-region.

[0092] For each point (x, y, z) in the three-dimensional model of the highway slope, the simulated curve shape parameter and the simulated derivative parameter of the soil and water feature curve of each point (x, y, z) in the three-dimensional model of the highway slope are obtained by inverse distance square weighting method:

[0093]

[0094]

[0095] wherein, is the simulated curve shape parameter of the highway slope at coordinates (x, y, z), is the simulated derivative parameter of the highway slope at coordinates (x, y, z), is the distance from the unmeasured point (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, represents the simulated curve shape parameter of the kth highway slope sub-region.

[0096] For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated saturated permeability coefficient of each point (x, y, z) in the three-dimensional model of the highway slope is obtained by the inverse distance square weighting method:

[0097]

[0098] wherein, represents the simulated saturated permeability coefficient of the highway slope at the coordinate (x, y, z), is the distance from the coordinate (x, y, z) of the highway slope to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated saturated permeability coefficient of the kth highway slope sub-region.

[0099] For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated porosity of each point (x, y, z) in the three-dimensional model of the highway slope is obtained by the inverse distance square weighting method:

[0100]

[0101] wherein, represents the simulated porosity of the highway slope at the coordinate (x, y, z), is the distance from the coordinate (x, y, z) of the highway slope to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated porosity of the kth highway slope sub-region.

[0102] The simulated initial soil moisture saturation of each point (x, y, z) in the three-dimensional model of the highway slope is also calculated by the inverse distance square weighting method , the simulated cohesion coefficient , the simulated initial internal friction angle , the simulated internal friction angle decay coefficient .

[0103] The simulated initial soil moisture saturation , the simulated curve shape parameter , and the simulated suction coefficient in the soil property data of each point in the three-dimensional model of the highway slope are input into the water and soil characteristic function in the drying period to obtain the simulated initial soil matric suction .

[0104] Step S4: Based on the simulated initial soil matrix suction and saturated permeability function, the water movement of the three-dimensional model of the highway slope is simulated by the unsaturated seepage control equation to obtain the simulated soil moisture saturation and simulated soil pore water pressure at each point in the three-dimensional model of the highway slope.

[0105] The unsaturated seepage control equation simulates water flow on a highway slope under initial dry conditions. Therefore, the inputs needed are the simulated initial soil matrix suction and the simulated initial unsaturated permeability coefficient at each point in the 3D model of the highway slope. The initial condition for the unsaturated seepage control equation is: at t=0, , It is the pressure head at coordinates (x, y, z) at time t=0. = , The density of water is 1000 kg / m³. Where g is the acceleration due to gravity, the soil property data at each point in the three-dimensional model of the highway slope are... Simulate initial soil moisture saturation Simulated curve shape parameters and simulated saturated permeability coefficient The input is given to the unsaturated permeability function during the dry period to obtain the simulated initial unsaturated permeability coefficient. .

[0106] The simulated initial soil matrix suction and simulated initial unsaturated permeability coefficient at each point in the 3D model of the highway slope are used to simulate water movement through the unsaturated seepage control equation:

[0107]

[0108] in, It is a function of specific water capacity. = , It is the simulated volumetric water content at coordinates (x, y, z) at time t. It is an unsaturated osmotic function. It is the pressure head at coordinates (x, y, z) at time t, and z is the elevation head.

[0109] It should be noted that the boundary conditions of the unsaturated seepage control equation, for example, the surface: rainfall infiltration flux boundary ( , where q(t) is the rainfall at time t, n is the outward normal vector of the surface boundary, and the rainfall intensity can be obtained from local meteorological forecast data) or pressure head boundary (when water accumulates); slope surface: atmospheric boundary (h=0); bottom: constant head boundary (groundwater level) or zero flux boundary.

[0110] The unsaturated seepage control equation was solved using the finite element method to obtain the pressure head h(x,y,z,t) at time t for each point (x,y,z). Finally, the following results were obtained: : Let be the simulated volumetric water content of the highway slope at time t in coordinates (x, y, z). = *pore(x,y,z).

[0111] It should be noted that the simulated volumetric water content is calculated based on different periods. This varies depending on the rainfall intensity. 10mm / h, Based on the soil and water characteristic curve during the peak of the rainstorm, the simulated volumetric water content is calculated. The simulated volumetric water content during the transition period between the dry period and the peak of the rainstorm can be calculated using an average weighted average method (0.6 for the dry period and 0.4 for the peak of the rainstorm) to avoid gaps in the calculated simulated volumetric water content. If the rainfall intensity is 0 mm / h for 24 hours, 0 < If the volumetric water content is m / s, then the simulated volumetric water content can be calculated based on the soil and water characteristic curve 24 hours after the rain stops. The simulated volumetric water content during the transition period between the peak of the rainstorm and 24 hours after the rain stops can be calculated by average weighting (the weight of the peak of the rainstorm is 0.6, and the weight of 24 hours after the rain stops is 0.4).

[0112] The simulated soil moisture saturation at each point in the 3D model of the highway slope is as follows:

[0113]

[0114] in, This represents the simulated soil moisture saturation of the highway slope at coordinates (x, y, z) at time t. Let be the simulated volumetric water content of the highway slope at time t in coordinates (x, y, z). The simulated porosity of the highway slope at coordinates (x, y, z) is given.

[0115] The simulated soil pore water pressure at each point (x, y, z) in the three-dimensional model of the highway slope is as follows:

[0116]

[0117] in, To simulate the soil pore water pressure of the highway slope at coordinates (x, y, z) at time t, The density of water is 1000 kg / m³. g is the acceleration due to gravity. () represents the inverse function of the soil and water characteristic function.

[0118] The simulated soil cohesion of the highway slope at each point (x, y, z) in the three-dimensional model of the highway slope is:

[0119]

[0120] wherein, is the simulated soil cohesion of the highway slope at the coordinate (x, y, z) at time t, is the cohesion in the dry state, is the simulated cohesion coefficient of the highway slope at the coordinate (x, y, z), is the simulated soil moisture saturation of the highway slope at the coordinate (x, y, z) at time t.

[0121] The simulated soil effective stress of the highway slope at each point (x, y, z) in the three-dimensional model of the highway slope is:

[0122]

[0123] wherein, is the simulated soil effective stress of the highway slope at the coordinate (x, y, z) at time t, is the unit weight of the soil, is the depth, represents the simulated soil pore water pressure of the highway slope at the coordinate (x, y, z) at time t.

[0124] The simulated internal friction angle of the highway slope at each point (x, y, z) in the three-dimensional model of the highway slope is:

[0125]

[0126] wherein, represents the simulated internal friction angle of the highway slope at the coordinate (x, y, z) at time t, represents the simulated initial internal friction angle, represents the simulated internal friction angle decay coefficient at the coordinate (x, y, z), is the simulated soil moisture saturation of the highway slope at the coordinate (x, y, z) at time t.

[0127] Step S5: screening the three-dimensional model of the highway slope by the simulated soil moisture saturation and the simulated soil pore water pressure to obtain a risk point of the three-dimensional model of the highway slope; clustering the risk point of the three-dimensional model of the highway slope by a density clustering method to obtain a risk cluster of the three-dimensional model of the highway slope.

[0128] After the construction of the three-dimensional model of the highway slope and the acquisition of the simulated soil moisture saturation and the simulated soil pore water pressure The points with simulated soil moisture saturation and simulated soil pore water pressure greater than the preset threshold value are taken as the risk points of the three-dimensional model of the highway slope.

[0129] It should be noted that the points with simulated soil moisture saturation and simulated soil pore water pressure greater than the preset threshold value are taken as the risk points of the three-dimensional model of the highway slope. For example, the critical threshold value of simulated soil moisture saturation is set according to different soil layer types. The simulated soil moisture saturation threshold value of clay is 0.85 (Montmorillonite and other clay minerals reach a layer spacing of 2 nanometers when the saturation is 85%, and the interlayer combined water film thickens to cause the cementation force to collapse by more than 60%), the simulated soil moisture saturation threshold value of silt is 0.9 (The capillary effect of silt in the 1-10 micron dominant pore collapses, and the intergranular effective stress is lost by more than 40%, and enters a quasi-liquefied sensitive state), the simulated soil moisture saturation threshold value of sand is 0.95 (The sand threshold value is 0.95, aiming at the liquefaction risk, the pore gas is closed when the saturation is 95%, causing the undrained shear strength to drop sharply, approaching the critical condition of seismic liquefaction), the pore water pressure threshold value is determined by the soil bulk density and the depth, and the pore water pressure threshold value is 0.7 *|z|, is the soil bulk density, and z is the depth.

[0130] The risk points of the three-dimensional model of the highway slope are clustered by a density clustering method to obtain a risk cluster of the three-dimensional model of the highway slope. The neighborhood radius and the minimum sample number in the density clustering method are initialized. The points between the risk points of the three-dimensional model of the highway slope whose Euclidean distance satisfies less than the neighborhood radius are regarded as points in the neighborhood, that is, it is considered that these points are close enough in the physical space and have strong relevance. Then, based on the number of points in the neighborhood, core points, boundary points and noise points are identified: if the number of points contained in the neighborhood of a node is not less than the set minimum sample number, the node is determined as a core point, and the core point is a key node of clustering; the boundary point belongs to the node whose number of points in the neighborhood is less than the minimum sample number but itself is in the neighborhood of a core point; and the node which is neither a core point nor a boundary point is a noise point, and the relevance of such node with other nodes is weak. Finally, adjacent core point clusters are merged, that is, the clusters formed by the core points having a density connection with each other are integrated into a complete clustering cluster, thereby generating the final risk cluster of the three-dimensional model of the highway slope.

[0131] Step S6: search for a potential sliding surface in the three-dimensional model risk cluster of the highway slope by using a particle swarm optimization algorithm, and calculate the sliding surface safety factor of the potential sliding surface by using a soil strip method; obtain the area of the potential sliding surface by calculating the area of the potential sliding surface; calculate the highway slope risk index by combining the area of the potential sliding surface and the sliding surface safety factor, compare the highway slope risk index with a preset threshold value, and divide the stability level to realize the prediction of the slope stability.

[0132] By using the particle swarm optimization algorithm, each particle corresponds to a potential sliding surface LZ=[ , , ,R],and the sliding surface is defined in the three-dimensional model risk cluster: the center c=( , , ), the center c needs to be in the corresponding three-dimensional model risk cluster, the spherical equation is: , R [0.5H,2H] radius range, H represents the vertical height of the highway slope, H= - , the sliding surface: S={(x,y,z) spherical surface |(x,y,z) slope rock-soil body}, the sliding surface needs to be contained in the corresponding three-dimensional model risk cluster, the number of particle swarms is M, and the maximum number of iterations is Take the reciprocal of the sliding surface stability as the fitness value in the particle swarm optimization algorithm, and divide the three-dimensional rock-soil column between the sliding surface and the top (ground surface) of the highway slope by using a soil strip.

[0133] It should be noted that the soil strip division is performed on the three-dimensional rock-soil column between the actual sliding surface and the top (ground surface) of the highway slope. The essence of soil strip division is to discretize the potential sliding body into a series of three-dimensional prisms perpendicular to the sliding direction. The spatial range of each soil strip is jointly defined by the bottom sliding surface and the top ground surface: the bottom boundary strictly adheres to the actual sliding surface defined by the parameterization (i.e., the intersection surface of the spherical surface and the slope rock-soil body), and the top boundary corresponds to the natural ground surface curve of the highway slope. The specific generation process is as follows: sliding surface discretization: the actual sliding surface is cut into N segments (usually 30 segments) along the sliding direction, and each segment of the sliding surface serves as the bottom surface of the soil strip. For example, in a circular arc sliding surface, the sliding surface curve is equally divided according to the central angle, and the connecting lines of adjacent division points form the bottom surface line segment of the soil strip. Ground surface projection: vertically project each sliding surface division point onto the slope ground surface curve (obtained through a digital elevation model) to form the corresponding ground points. The connecting lines of adjacent ground points form the top boundary of the soil strip. Lateral closure: by establishing four vertical sides connecting the corresponding vertices of the bottom surface and the top surface, a closed three-dimensional prism is formed: two end faces: the quadrilaterals formed by connecting the end points of the bottom surface and the corresponding ground points. Two sides: the inclined surfaces connecting adjacent bottom-top point pairs along the sliding direction. Spatial parameter analysis: for each generated soil strip prism, automatically calculate the key geometric parameters: bottom length (arc length of sliding surface segmentation), average depth (vertical distance from bottom surface to top surface), volume (prism spatial volume), inclination angle (angle between bottom surface line segment and horizontal plane), ensuring that each soil strip completely contains all rock-soil bodies between the sliding surface and the ground surface, reflecting the geometric characteristics of the sliding surface and preserving the spatial form of the ground surface topography. For example, in a slope toe circular arc sliding surface, the soil strips close to the center are higher (reflecting the thick soil layer at the slope top), and the soil strips far from the center are shallower (corresponding to the thin soil layer at the slope toe), accurately matching the actual material distribution of the slope.

[0134] Calculate the sliding surface safety factor of each potential sliding surface in the risk cluster of the three-dimensional model of the highway slope:

[0135]

[0136] wherein, is the sliding surface safety factor of the sliding surface at time t, i is the index of the ith soil strip, and N is the total number of soil strips, is the simulated soil cohesion of the ith soil strip at time t, is the bottom length of the ith soil strip, is the weight of the ith soil strip, is the inclination angle of the ith soil strip, is the simulated pore water pressure of the ith soil strip, is the simulated internal friction angle of the ith soil strip at time t.

[0137] It should be noted that the simulated soil cohesion of the ith soil strip the simulated internal friction angle of the i-th slice , the simulated soil cohesion and the simulated internal friction angle corresponding to the center point of the potential sliding surface of the bottom of the i-th slice; the weight of the i-th slice , = wherein, is the bulk density of the i-th slice, is the volume of the i-th slice.

[0138] For each particle in the particle swarm optimization algorithm, the velocity and position are updated as follows:

[0139]

[0140] wherein, represents the velocity of the m-th particle in the dd+1-th iteration of the wd-th feature, is the inertia weight in the particle swarm optimization algorithm, which controls the degree of reservation of the historical velocity of the particle, represents the velocity of the m-th particle in the dd-th iteration of the wd-th feature, and represent the individual learning factor and the social learning factor in the particle swarm optimization algorithm, both of which are usually taken as 2.0, and are random numbers, which avoid local optimum of the particle swarm optimization algorithm, represents the wd-th feature value of the historical optimal position of the m-th particle, represents the wd-th feature value of the global optimal position found by all particles in the particle swarm optimization algorithm.

[0141]

[0142] wherein, represents the position of the m-th particle in the dd+1-th iteration of the wd-th feature, represents the position of the m-th particle in the dd-th iteration of the wd-th feature, represents the velocity of the m-th particle in the dd+1-th iteration of the wd-th feature.

[0143] By continuously updating the particle position, the particle that makes the sliding surface safety factor minimum is searched. When the number of iterations reaches the maximum number of iterations , the iteration is stopped or the global optimal fitness value (i.e., the reciprocal of the minimum sliding surface safety factor) of all particles fluctuates by less than 1× ​​When the particle swarm optimization algorithm is determined to have reached a stable convergence state, the iteration is stopped. After the iteration is stopped, the sliding surface corresponding to the particle with the smallest safety factor of the sliding surface in the highway slope three-dimensional model risk cluster is taken as the potential sliding surface of the highway slope three-dimensional model risk cluster. Each highway slope three-dimensional model risk cluster corresponds to a potential sliding surface, and the safety factor of the sliding surface of each potential sliding surface is .

[0144] The area of the potential sliding surface is obtained by calculating the area of the potential sliding surface.

[0145]

[0146] wherein, is the potential sliding surface area of the potential sliding surface of the J Lth highway slope three-dimensional model risk cluster, is the corresponding spherical radius of the potential sliding surface of the J Lth highway slope three-dimensional model risk cluster, is the spherical cap height of the potential sliding surface of the J Lth highway slope three-dimensional model risk cluster.

[0147] It should be noted that the spherical cap height represents the maximum depth of the sliding surface in the slope, that is, the vertical distance from the slope toe to the potential sliding surface.

[0148] The highway slope risk index is calculated by combining the potential sliding surface area and the safety factor of the sliding surface.

[0149]

[0150] wherein, is the highway slope risk index at time t, is a maximum value function that takes the maximum value of all potential sliding surface safety factors of the sliding surface, is the safety factor of the sliding surface of the potential sliding surface of the J Lth highway slope three-dimensional model risk cluster, is the potential sliding surface area of the potential sliding surface of the J Lth highway slope three-dimensional model risk cluster, is the total number of potential sliding surfaces.

[0151] The highway slope risk index is calculated in real time and dynamically compared with a preset threshold to achieve accurate prediction of the stability of the slope. The specific implementation process is as follows: a four-level stability evaluation system is established, and when 0.3, it is determined to be in a stable state (blue warning), at which time the slope is in a safe range and only requires routine monitoring; when 0.3 0.6, it enters an alert state (yellow warning), indicating that the local soil body starts to soften, and the system automatically increases the monitoring frequency to 1 time per hour and starts crack inspection; when 0.6 0.8 is upgraded to high-risk state (orange alert), at which time the sliding surface accelerates development, immediately close the slope foot lane and deploy emergency rescue materials; when >0.8 triggers emergency state (red alert), which means that the slope instability is inevitable, the system automatically issues a full-section closure instruction and starts the personnel evacuation plan.

[0152] In the field of transportation infrastructure, the problem of highway slope instability caused by heavy rainfall has long restricted engineering safety and operational efficiency. Traditional single-point monitoring and simplified numerical models cannot accurately capture the spatial variation and dynamic evolution of hydro-mechanical parameters of the slope. Therefore, there is an urgent need for a fine prediction method that integrates multi-source data and intelligent algorithms.

[0153] The particle swarm optimization algorithm is used to search for potential sliding surfaces in the risk cluster, and the safety factor is calculated by the soil bar method. The core significance lies in the dynamic and accurate quantification of slope instability risk. The particle swarm optimization algorithm quickly locates the most dangerous sliding surface (spherical surface equation parameterization) in the three-dimensional risk cluster through the global optimization mechanism (with the reciprocal of the stability of the sliding surface as the fitness), breaking through the limitations of traditional empirical assumptions. The soil bar method calculates the safety factor of the sliding surface by discretizing the rock and soil column (the bottom surface is attached to the sliding surface, and the top surface matches the ground surface curve), coupling the real-time simulation of pore water pressure and strength decay parameters (cohesion index decay, internal friction angle linear reduction), and calculating the safety factor of the sliding surface. This converts the hydro-mechanical dynamic response into a quantitative stability index, providing accurate information on the critical sliding surface position and instability probability for early warning decision-making.

[0154] Based on the simulation of initial soil matric suction and saturated permeability function, the core significance of the non-saturated seepage control equation in simulating water movement lies in dynamically tracking the three-dimensional spatio-temporal evolution of pore water pressure and saturation during the rainwater infiltration process, and reflecting the nonlinear decay law of soil strength (cohesion, internal friction angle) with water content in real time. This breaks through the limitations of traditional static parameters and provides dynamic mechanical basis for accurate early warning of slope instability.

[0155] The risk index of highway slope is calculated by integrating the area of potential sliding surface and the safety factor of sliding surface. The comprehensive evaluation index can be constructed by quantifying the "risk influence range x safety factor critical value". This index dynamically relates the geometric characteristics of the spherical sliding surface (such as radius, spherical crown height) to the soil anti-sliding ability, and combines the four-level stability evaluation system (blue alert to red alert) to quantitatively reflect the spatial scale and severity of slope instability risk in real time.

[0156] It has to be noted that, in the present document, the terms "first", "second", etc. merely serve to identify different entities or actions and do not necessarily require or imply any actual relationship or order between these entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. In other words, without further restriction, reference to elements will not, without more limitations, exclude additional, unrecited elements of a process, method, article, or apparatus.

[0157] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous modifications and changes can be made to the embodiments without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.

Claims

1. A method of predicting the stability of a slope in heavy rainfall weather, characterized by: The method comprises the following steps: Step S1: sampling the soil of the sub-regions of the highway slope before and after rainfall to obtain soil characteristic data of the sub-regions of the highway slope; Step S2: performing curve fitting on the soil moisture saturation and soil matric suction in the soil characteristic data of the sub-regions of the highway slope by the least square method to obtain a water-soil characteristic function, performing curve fitting on the soil moisture saturation and the unsaturated permeability coefficient in the soil characteristic data of the sub-regions of the highway slope to obtain a saturated permeability function, and performing curve fitting on the soil moisture saturation and the soil cohesion in the soil characteristic data of the sub-regions of the highway slope to obtain a soil cohesion function; Step S3: performing three-dimensional modeling on the highway slope by a UAV carrying a laser radar to obtain a three-dimensional model of the highway slope; obtaining the soil characteristic data of each point in the three-dimensional model of the highway slope according to the inverse distance square weighting method based on the soil characteristic data of the sub-regions of the highway slope, and inputting the soil characteristic data of each point in the three-dimensional model of the highway slope into the water-soil characteristic function to obtain simulated initial soil matric suction; Step S4: simulating water movement in the three-dimensional model of the highway slope by an unsaturated seepage control equation based on the simulated initial soil matric suction and simulated initial unsaturated permeability coefficient to obtain the simulated soil moisture saturation and simulated soil pore water pressure of each point in the three-dimensional model of the highway slope; Step S5: screening the three-dimensional model of the highway slope by the simulated soil moisture saturation and simulated soil pore water pressure to obtain risk points of the three-dimensional model of the highway slope; and clustering the risk points of the three-dimensional model of the highway slope by the density clustering method to obtain risk clusters of the three-dimensional model of the highway slope; Step S6: searching for a potential sliding surface in the risk clusters of the three-dimensional model of the highway slope by a particle swarm optimization algorithm, calculating a sliding surface safety factor of the potential sliding surface by a soil bar method, calculating an area of the potential sliding surface, obtaining a highway slope risk index by combining the area of the potential sliding surface and the sliding surface safety factor, comparing the highway slope risk index with a preset threshold value, and dividing a stability level to realize prediction of the stability of the slope.

2. The method for predicting the slope stability in heavy rainfall weather according to claim 1, characterized in that: The curve fitting on the soil moisture saturation and soil matric suction in the soil characteristic data of the sub-regions of the highway slope by the least square method to obtain the water-soil characteristic function comprises the following specific steps: The curve fitting on the soil matric suction and soil moisture saturation in the dry period, rainstorm peak period and 24 hours after rain of the sub-regions of the highway slope by the least square method to obtain the water-soil characteristic function comprises the following steps: ； wherein, represents the soil moisture saturation at a soil matric suction of represents the suction coefficient of the kth highway slope sub-region, is the curve shape parameter of the kth highway slope sub-region, is the derived parameter of the kth highway slope sub-region, represents the soil matric suction.​ 3. The method for predicting the slope stability in heavy rainfall weather according to claim 2, characterized in that: The curve fitting on the soil moisture saturation and unsaturated permeability coefficient in the soil characteristic data of the sub-regions of the highway slope to obtain the saturated permeability function comprises the following steps: The curve fitting on the soil moisture saturation and unsaturated permeability coefficient in the dry period, rainstorm peak period and 24 hours after rain of the sub-regions of the highway slope by the least square method to obtain the unsaturated permeability function comprises the following steps: ； wherein, is the unsaturated hydraulic conductivity for a soil moisture saturation of is the saturated hydraulic conductivity of the kth highway slope sub-area, is the soil moisture saturation, is an empirical exponent, by default 0.5, is the derived parameter of the kth highway slope sub-area, same as the of the soil water characteristic function.​ 4. The method for predicting the slope stability in heavy rainfall weather according to claim 3, characterized in that: The obtaining of the soil characteristic data of each point in the three-dimensional model of the highway slope according to the inverse distance square weighting method based on the soil characteristic data of the sub-regions of the highway slope comprises the following steps: For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated suction coefficient of each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope is obtained by the inverse distance square weighting method: ； wherein, represents the simulated suction factor at the coordinate (x, y, z) on the highway slope, is the distance from the unmeasured point (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated suction factor of the kth highway slope sub-region; For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated curve shape parameter and the simulated derivative parameter of the water and soil characteristic curve of each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope are obtained by the inverse distance square weighting method: ； ； wherein, is a simulated curve shape parameter of the highway slope at coordinates (x, y, z), is a simulated derivative parameter of the highway slope at coordinates (x, y, z), is a distance from the highway slope at coordinates (x, y, z) to the kth highway slope sub-region, k represents an index of the kth highway slope sub-region, represents a simulated curve shape parameter of the kth highway slope sub-region; For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated saturated permeability coefficient of each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope is obtained by the inverse distance square weighting method: ； wherein, represents the simulated saturated permeability coefficient of the highway slope at coordinate (x, y, z), is the distance from the highway slope at coordinate (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated saturated permeability coefficient of the kth highway slope sub-region; For each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope, the simulated porosity of each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope is obtained by the inverse distance square weighting method: ； wherein, represents the simulated porosity of the highway slope at coordinate (x, y, z), is the distance from the highway slope at coordinate (x, y, z) to the kth highway slope sub-region, k represents the index of the kth highway slope sub-region, and K represents the total number of highway slope sub-regions, represents the simulated porosity of the kth highway slope sub-region; The simulated initial soil moisture saturation of each point (x, y, z) in the three-dimensional model of the highway slope is also calculated by the inverse distance square weighting method , simulating the cohesion coefficient , simulating the initial internal friction angle , simulating the internal friction angle attenuation coefficient .

5. The method for predicting the slope stability in heavy rainfall weather according to claim 4, characterized in that: The soil property data of each point in the three-dimensional model of the highway slope is input into the water and soil characteristic function to obtain the simulated initial soil matric suction, including the following steps: simulating initial soil moisture saturation degree in the soil property data of each point in the highway slope three-dimensional model , simulating curve shape parameters and simulating suction coefficient input into the water and soil characteristic function of the dry period to obtain simulated initial soil matric suction .

6. The method for predicting the slope stability in heavy rainfall weather according to claim 5, characterized in that: Based on the simulated initial soil matric suction and the saturated permeability function, the water movement in the three-dimensional model of the highway slope is simulated by the unsaturated seepage control equation to obtain the simulated soil moisture saturation and the simulated soil pore water pressure of each point in the three-dimensional model of the highway slope, including the following steps: The unsaturated seepage control equation simulates water flow on a highway slope under initial dry conditions. Therefore, the inputs needed are the simulated initial soil matrix suction and the simulated initial unsaturated permeability coefficient at each point in the 3D model of the highway slope. The initial condition for the unsaturated seepage control equation is: at t=0, , It is the pressure head at coordinates (x, y, z) at time t=0. = , The density of water is 1000 kg / m³. Where g is the acceleration due to gravity, the soil property data at each point in the three-dimensional model of the highway slope are... Simulate initial soil moisture saturation Simulated curve shape parameters and simulated saturated permeability coefficient The water and soil characteristic function during the dry period is input to obtain the simulated initial unsaturated permeability coefficient. ; The simulated initial soil matric suction and the simulated initial unsaturated permeability coefficient of each point in the three-dimensional model of the highway slope are simulated by the unsaturated seepage control equation to simulate the water movement: ； wherein is the specific water capacity function, = , is the simulated volumetric water content at coordinates (x, y, z) at time t, is the unsaturated hydraulic conductivity function, is the pressure head at coordinates (x, y, z) at time t, and z is the elevation head. The non-saturated seepage control equation is solved by the finite element method to obtain the pressure water head h(x, y, z, t) of each point (x, y, z) at t time; and finally the : is the simulated volume water content of the highway side slope at t time under the coordinate (x, y, z), = pore(x, y, z); Then the simulated soil moisture saturation of each point in the three-dimensional model of the highway slope is: ； wherein, S (x, y, z, t) represents the simulated soil moisture saturation at coordinate (x, y, z) of the highway slope at time t, VWC (x, y, z, t) is the simulated volumetric water content at coordinate (x, y, z) of the highway slope at time t, Por (x, y, z) is the simulated porosity of the highway slope at coordinate (x, y, z); Then the simulated soil pore water pressure of each point (x, y, z) in the three-dimensional model of the highway slope of the highway slope is: ； wherein is the simulated soil pore water pressure of the road side slope at time t at coordinate (x, y, z), is the water density, 1000 kg / m3, g is the gravitational acceleration, () denotes the inverse function of the water soil characteristic function.

7. The method for predicting the slope stability in heavy rainfall weather according to claim 6, characterized in that: The three-dimensional model of the highway slope is screened by the simulated soil moisture saturation and the simulated soil pore water pressure to obtain the three-dimensional model risk points of the highway slope, including the following specific steps: After the construction of the three-dimensional model of the highway slope and the acquisition of the simulated soil moisture saturation of each spatial point and the simulated soil pore water pressure , the points with simulated soil moisture saturation and simulated soil pore water pressure greater than the preset threshold are regarded as the risk points of the three-dimensional model of the highway slope by threshold screening of the simulated soil moisture saturation and simulated soil pore water pressure of each point in the three-dimensional model of the highway slope.

8. The method for predicting the slope stability in heavy rainfall weather according to claim 7, characterized in that: The three-dimensional model risk points of the highway slope are clustered by the density clustering method to obtain the three-dimensional model risk clusters of the highway slope, including the following specific steps: The three-dimensional model risk points of the highway slope are clustered by the density clustering method to obtain the three-dimensional model risk clusters of the highway slope, the neighborhood radius and the minimum sample number in the density clustering method are initialized, the points between the three-dimensional model risk points of the highway slope whose Euclidean distance satisfies less than the neighborhood radius are regarded as points in the neighborhood, that is, these points are considered to be close enough in physical space and have strong relevance; based on the number of points in the neighborhood, core points, boundary points and noise points are identified: if the number of points contained in the neighborhood of a certain node is not less than the set minimum sample number, the node is determined as a core point, and the core point is the key node of clustering; the boundary point belongs to the node whose number of points in the neighborhood is less than the minimum sample number but itself is in the neighborhood of a certain core point; The nodes that are neither core points nor boundary points are noise points, and the relevance of these nodes to other nodes is weak; adjacent core point clusters are merged, that is, the clusters formed by the core points that have density connection with each other are integrated into a complete clustering cluster, thereby generating the final three-dimensional model risk clusters of the highway slope.

9. The method for predicting the slope stability in heavy rainfall weather according to claim 8, characterized in that: The method for calculating the sliding surface safety factor of the potential sliding surface through the soil strip method comprises the following specific steps: The sliding surface safety factor of each potential sliding surface in the risk cluster of the three-dimensional model of the highway slope is calculated: ； wherein, is a safety factor of the sliding surface at time t, i is an index of the ith soil strip, and N is the total number of soil strips, is the simulated soil cohesion of the ith soil strip at time t, is the length of the bottom of the ith soil strip, is the weight of the ith soil strip, is the inclination angle of the ith soil strip, is the simulated pore water pressure of the ith soil strip, is the simulated internal friction angle of the ith soil strip at time t.

10. The method for predicting the slope stability in heavy rainfall weather according to claim 9, characterized in that: The highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor, and the method comprises the following specific steps: The highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor: The highway slope risk index is calculated by combining the potential sliding surface area and the sliding surface safety factor: ； wherein, is the highway slope risk index at time t, is the maximum value function, and is the maximum value of the sliding surface safety factor of all potential sliding surfaces, is the sliding surface safety factor of the potential sliding surface of the JLandth highway slope three-dimensional model risk cluster, is the potential sliding surface area of the potential sliding surface of the JLandth highway slope three-dimensional model risk cluster, is the total number of potential sliding surfaces.

Images