A method for predicting sliding surface instability of a waste dump slope

Abstract

The application discloses a sliding surface instability prediction method for a waste dump slope, which comprises collecting waste dump slope environment data; constructing a first three-dimensional numerical analysis model based on a finite difference method, dynamically inverting and calibrating the first three-dimensional numerical analysis model based on soil moisture content data and ground surface displacement data to obtain a second three-dimensional numerical analysis model, dynamically analyzing the stability of the waste dump slope based on the second three-dimensional numerical analysis model to obtain the safety factor of the waste dump slope; and dynamically analyzing the stability of the waste dump slope again based on the third three-dimensional numerical analysis model to obtain a dynamic closed loop for predicting the stability state of the waste dump slope. The dynamic closed loop feedback mechanism can continuously adjust and optimize the analysis model according to actual monitoring data, so that the stability state of the waste dump slope can be more accurately predicted, and more powerful support can be provided for risk assessment and early warning.

Description

Technical Field

[0001] This invention relates to the field of slip surface instability prediction technology for spoil disposal site slopes, and more particularly to a method for predicting slip surface instability of spoil disposal site slopes. Background Technology

[0002] Traditional methods often rely on periodic manual monitoring of spoil heap slopes, which has poor timeliness and makes it difficult to capture sudden stability changes in a timely manner, thus failing to provide timely warnings. Traditional methods usually consider the thermal, hydrological, and mechanical effects of the soil separately, ignoring the coupling relationship between them, resulting in inaccurate prediction results. Traditional methods may use fixed parameters or only make a limited number of adjustments to the prediction model parameters, making it difficult to guarantee the long-term accuracy of the model, which often leads to errors in the prediction results. Traditional methods often require a large amount of manpower for on-site monitoring and data collection, which is costly and inefficient. Summary of the Invention

[0003] The purpose of this invention is to provide a method for predicting slip surface instability of spoil heap slopes, in order to solve the above-mentioned problems.

[0004] This invention is achieved through the following technical solution:

[0005] A method for predicting slip surface instability of spoil heap slopes includes:

[0006] Environmental data of the spoil heap slope is collected, including rainfall intensity data, groundwater level change data, soil moisture content data and surface displacement data. The three-dimensional coordinate sequence of the surface deformation monitoring points of the spoil heap slope is obtained based on the Beidou satellite positioning system.

[0007] A first three-dimensional numerical analysis model of soil thermal-hydraulic-mechanical coupling effect is constructed based on the finite difference method. The rainfall intensity data and groundwater level change data are used as dynamic boundary conditions and input into the first three-dimensional numerical analysis model to obtain the gradient distribution characteristics of pore water pressure field on the spoil disposal site slope.

[0008] Based on the soil moisture content data and surface displacement data, the first three-dimensional numerical analysis model is dynamically inverted and calibrated to obtain the second three-dimensional numerical analysis model. Based on the second three-dimensional numerical analysis model, the stability dynamic analysis of the spoil heap slope is performed to obtain the safety factor of the spoil heap slope.

[0009] When the safety factor is lower than the preset safety threshold, a displacement field interpolation model is constructed based on the three-dimensional coordinate sequence, and the geometric parameters and spatial location of the sliding surface of the spoil disposal site slope are obtained based on the displacement field interpolation model and the gradient distribution characteristics of the pore water pressure field.

[0010] The second three-dimensional numerical analysis model is analyzed and optimized based on the geometric parameters and spatial location to obtain a third three-dimensional numerical analysis model. Based on the third three-dimensional numerical analysis model, the stability dynamic analysis of the spoil heap slope is performed again to obtain a dynamic closed loop for predicting the stability state of the spoil heap slope.

[0011] Preferably, the first three-dimensional numerical analysis model for the soil thermal-hydraulic-mechanical coupling effect, constructed based on the finite difference method, includes:

[0012] The three-dimensional geometric domain of the spoil heap slope is divided into a finite difference grid, and the finite difference grid serves as the node of the first three-dimensional numerical analysis model;

[0013] The first governing equation of the first three-dimensional numerical analysis model is established based on the environmental data, wherein the first governing equation includes the heat conduction equation, the seepage equation, and the stress balance equation.

[0014] The thermal-hydraulic-mechanical coupling relationship is defined based on the first governing equation;

[0015] The environmental data is used as the boundary condition for the first three-dimensional numerical analysis model;

[0016] The first control equation is discretized using the explicit difference method to obtain a second control equation with stronger stability.

[0017] The first three-dimensional numerical analysis model is constructed based on the above steps.

[0018] Preferably, the rainfall intensity data and groundwater level change data are input as dynamic boundary conditions into the first three-dimensional numerical analysis model to obtain the pore water pressure field gradient distribution characteristics of the spoil heap slope, including:

[0019] The rainfall intensity data is used as the rainfall infiltration boundary and input into the first three-dimensional numerical analysis model.

[0020] The groundwater level change data is used as the groundwater level boundary and input into the first three-dimensional numerical analysis model.

[0021] The pore water pressure gradient of each node in the first three-dimensional numerical analysis model is calculated based on the numerical difference method to obtain the pore water pressure field gradient distribution characteristics of the spoil disposal site slope.

[0022] Preferably, the first three-dimensional numerical analysis model is dynamically inverted and calibrated based on the soil moisture content data and surface displacement data to obtain a second three-dimensional numerical analysis model, including:

[0023] An inversion analysis module is constructed based on the soil moisture content data and surface displacement data.

[0024] Based on the particle swarm optimization algorithm and the inversion analysis module, the permeability coefficient parameters and soil shear strength parameters of the first three-dimensional numerical analysis model are dynamically inverted and calibrated to obtain the soil constitutive model parameter set with minimized error.

[0025] The second three-dimensional numerical analysis model is obtained based on the parameter set of the soil constitutive model.

[0026] Preferably, a dynamic stability analysis of the spoil heap slope is performed based on a second three-dimensional numerical analysis model to obtain the safety factor of the spoil heap slope, including:

[0027] Based on the strength reduction method, the soil shear strength parameters are reduced proportionally, and the reduced soil shear strength parameters are input into the second three-dimensional numerical analysis model.

[0028] The second three-dimensional numerical analysis model is converged. If it does not converge, the reduction is terminated. At this time, the surface displacement rate predicted by the second three-dimensional numerical analysis model is the safety factor.

[0029] Preferably, the convergence determination of the second three-dimensional numerical analysis model includes:

[0030] Calculate the surface displacement increment predicted by the second three-dimensional numerical analysis model in the adjacent reduction steps;

[0031] If the surface displacement increment is greater than the preset surface displacement increment, it is determined to be a displacement abrupt change, and thus the second three-dimensional numerical analysis model is determined to be non-convergent.

[0032] Preferably, constructing a displacement field interpolation model based on the three-dimensional coordinate sequence includes:

[0033] Based on the Kriging method, the coordinates of the surface deformation monitoring points of the spoil heap slope are interpolated according to the time window to obtain the displacement field sequence of the spoil heap slope.

[0034] The displacement field interpolation model is constructed based on the displacement field sequence.

[0035] Preferably, the geometric parameters and spatial location of the slip surface of the spoil heap slope are obtained based on the displacement field interpolation model and the gradient distribution characteristics of the pore water pressure field, including:

[0036] The displacement gradient of the spoil heap slope is obtained based on the displacement field interpolation model.

[0037] The pore water pressure gradient of the spoil heap slope is obtained based on the pore water pressure field gradient distribution characteristics.

[0038] The displacement gradient and pore water pressure gradient are fused based on preset weights to obtain a comprehensive gradient;

[0039] The gradient descent algorithm and the comprehensive gradient are used to track the displacement abrupt change region of the spoil heap slope in order to obtain the geometric parameters and spatial location of the slip surface of the spoil heap slope.

[0040] Preferably, the second three-dimensional numerical analysis model is analyzed and optimized based on the geometric parameters and spatial location to obtain a third three-dimensional numerical analysis model, including:

[0041] The geometric parameters of the slip surface of the spoil heap slope are input into a long short-term memory network for time-series prediction. The number of hidden layer neurons and the learning rate of the long short-term memory network are optimized online based on a genetic algorithm to obtain the early warning time window of the spoil heap slope.

[0042] Based on the warning time window, the predicted surface displacement rate of the second three-dimensional numerical analysis model is obtained. When the actual surface displacement is detected to exceed the predicted surface displacement rate, the parameters of the second three-dimensional numerical analysis model are updated in a rolling manner based on the Kalman filter algorithm to obtain the third three-dimensional numerical analysis model.

[0043] Preferably, the parameters of the second three-dimensional numerical analysis model are updated on a rolling basis using the Kalman filter algorithm to obtain the third three-dimensional numerical analysis model, including:

[0044] Using the parameters of the second three-dimensional numerical analysis model as state variables and the real-time monitored surface displacement data as observation variables, a prediction error covariance matrix is ​​constructed based on the state variables, and an observation covariance matrix is ​​constructed based on the observation variables.

[0045] The Kalman gain is calculated based on the prediction error covariance matrix and the observation covariance matrix. The state variables are then updated based on the observed variables and the Kalman gain to obtain the updated model parameters.

[0046] The model parameters are input into the second three-dimensional numerical analysis model to obtain the third three-dimensional numerical analysis model.

[0047] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0048] 1. This invention uses a dynamic closed-loop feedback mechanism to continuously adjust and optimize the analysis model based on actual monitoring data, thereby more accurately predicting the stability state of spoil heap slopes and providing stronger support for risk assessment and early warning.

[0049] 2. This invention collects data in real time through sensors and combines it with the three-dimensional coordinate sequence obtained by the Beidou satellite positioning system to achieve real-time monitoring of the slope of the spoil disposal site, eliminating the influence of human intervention and improving monitoring efficiency.

[0050] 3. This invention can issue an early warning in a timely manner when the slope stability decreases to a dangerous level by setting a safety threshold and monitoring the safety factor in real time. Attached Figure Description

[0051] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0052] Figure 1 This is a schematic diagram of the overall method steps in one embodiment of the present invention; Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are for illustrative purposes only and are not intended to limit the invention. It should be noted that this invention is already in the actual research and development stage.

[0054] Example 1, such as Figure 1 As shown, the present invention proposes a method for predicting slip surface instability of spoil heap slopes, comprising:

[0055] S1 collects environmental data, including rainfall intensity data, groundwater level change data, soil moisture content data, and surface displacement data. Based on the Beidou satellite positioning system, it obtains the three-dimensional coordinate sequence of the surface deformation monitoring points of the spoil disposal site slope.

[0056] S2. Based on the finite difference method, a first three-dimensional numerical analysis model of the soil thermal-hydraulic-mechanical coupling effect is constructed. Rainfall intensity data and groundwater level change data are used as dynamic boundary conditions to input the first three-dimensional numerical analysis model to obtain the gradient distribution characteristics of the pore water pressure field of the spoil disposal site slope.

[0057] S3. Based on soil moisture content data and surface displacement data, the first three-dimensional numerical analysis model is dynamically inverted and calibrated to obtain the second three-dimensional numerical analysis model. Based on the second three-dimensional numerical analysis model, the stability dynamic analysis of the spoil disposal site slope is carried out to obtain the safety factor of the spoil disposal site slope.

[0058] S4. When the safety factor is lower than the preset safety threshold, a displacement field interpolation model is constructed based on the three-dimensional coordinate sequence. Based on the displacement field interpolation model and the gradient distribution characteristics of the pore water pressure field, the geometric parameters and spatial location of the sliding surface of the spoil disposal site slope are obtained.

[0059] S5. Based on geometric parameters and spatial location, the first three-dimensional numerical analysis model is analyzed and optimized to obtain the third three-dimensional numerical analysis model. Based on the third three-dimensional numerical analysis model, the stability dynamic analysis of the spoil disposal site slope is performed again to obtain the dynamic closed loop for predicting the stability state of the spoil disposal site slope.

[0060] In this invention, sensors are key devices for collecting various environmental data. These sensors can be deployed at different locations on the slope to monitor parameters such as rainfall intensity, groundwater level changes, soil moisture content, and surface displacement in real time. The environmental data includes rainfall intensity data, groundwater level change data, soil moisture content data, and surface displacement data. This data is crucial for assessing the stability of spoil heap slopes because it reflects the external influences and internal changes experienced by the slope. The finite difference method is a numerical analysis method used to solve partial differential equations. In the stability analysis of spoil heap slopes, the finite difference method is used to construct a first three-dimensional numerical analysis model considering the thermal-hydraulic-mechanical coupling effect of the soil to simulate the pore water pressure of the slope. Force and stress field distribution; the soil thermal-hydraulic-mechanical coupling effect refers to the interaction between temperature, moisture, and mechanical properties in soil. In spoil heap slopes, this coupling effect can significantly impact slope stability. The first three-dimensional numerical analysis model is constructed based on the finite difference method to simulate the pore water pressure and stress field distribution of spoil heap slopes. By inputting dynamic boundary conditions (such as rainfall intensity and groundwater level change data), the stability state of the slope under different conditions can be simulated. Dynamic boundary conditions refer to boundary conditions that change over time. In the stability analysis of spoil heap slopes, rainfall intensity and groundwater level change data are input as dynamic boundary conditions into the first three-dimensional numerical analysis model to reflect the slope's dynamic boundary conditions. External influences; the gradient distribution characteristics of the pore water pressure field refer to the changing trend of pore water pressure within the slope; dynamic inversion calibration refers to adjusting the parameters of the numerical analysis model using real-time monitoring data to improve the model's accuracy. By combining soil moisture content and surface displacement data to perform dynamic inversion calibration on the first three-dimensional numerical analysis model, a soil constitutive model parameter set matching the monitoring data can be generated. The safety factor is an important indicator for assessing slope stability; when the safety factor is lower than a preset threshold, it indicates that the slope stability has decreased to an unacceptable level; the displacement field interpolation model is constructed based on the three-dimensional coordinate sequence of surface displacement monitoring points and is used to extend discrete monitoring point data into a continuous displacement field. The displacement field interpolation model can reveal the overall distribution of slope surface displacement. The gradient descent algorithm, an optimization algorithm used to find the minimum value of a function, is used here to track abrupt changes in the displacement field to determine the location of the sliding surface. The third-dimensional numerical analysis model is the result of optimizing the original model based on the geometric parameters and spatial location of the sliding surface. By optimizing the model parameters, the prediction accuracy and reliability of the model can be improved. Dynamic closed-loop prediction refers to the dynamic prediction of slope stability by continuously updating model parameters and monitoring data. Here, through dynamic closed-loop prediction, the stability of the slope can be monitored in real time, and timely mitigation measures can be taken to ensure the safety and stability of the slope.

[0061] This embodiment includes: constructing a first three-dimensional numerical analysis model of the soil thermal-hydraulic-mechanical coupling effect based on the finite difference method, including:

[0062] A1. Divide the three-dimensional geometric domain of the spoil heap slope into a finite difference grid, and use the finite difference grid as the node of the first three-dimensional numerical analysis model;

[0063] A2. The first governing equation of the first three-dimensional numerical analysis model based on environmental data, wherein the governing equations include the heat conduction equation, the seepage equation and the stress balance equation;

[0064] A3. Define the thermo-hydraulic-mechanical coupling relationship of the first three-dimensional numerical analysis model based on the governing equations;

[0065] A4. Use environmental data as the boundary conditions of the first three-dimensional numerical analysis model;

[0066] A5. Based on the explicit difference method, the control equations of the first three-dimensional numerical analysis model are discretized to obtain a second control equation with stronger stability.

[0067] A6. Based on the above steps, construct the first three-dimensional numerical analysis model of the soil thermal-hydraulic-mechanical coupling effect.

[0068] It should be noted that the finite difference mesh divides the three-dimensional geometric domain of the spoil heap slope into many small, regular mesh elements. These mesh elements serve as nodes in the first three-dimensional numerical analysis model, used to discretize the continuous geometric domain for numerical calculations. The governing equations are fundamental equations describing physical phenomena, including the heat conduction equation, seepage equation, and stress balance equation. The heat conduction equation describes the temperature distribution and conduction process within the slope, the seepage equation describes the flow and infiltration of water within the slope, and the stress balance equation describes the stress distribution and equilibrium state within the slope. The thermo-water-mechanical coupling relationship refers to the interaction between temperature, moisture, and mechanical properties in the soil; changes in temperature affect the physical properties of the soil. Physical properties, such as permeability coefficient and shear strength, are affected by changes in moisture content, which in turn affect soil volume changes and stress state. Changes in mechanical properties, in turn, affect soil temperature and moisture distribution. Boundary conditions refer to the physical conditions or constraints on the boundaries of the numerical analysis model. Boundary conditions are used to simulate the boundary behavior of actual slopes, such as rainfall intensity and groundwater level changes. The explicit difference method is a numerical discretization method used to solve partial differential equations. Based on the explicit difference method, the governing equations can be discretized, transforming continuous partial differential equations into discrete algebraic equations, thus facilitating numerical solutions. The explicit difference method has the advantages of high computational efficiency and strong stability, and can more accurately simulate the thermal-hydraulic-mechanical coupling effect and stability state of slopes.

[0069] In an optional embodiment, rainfall intensity data and groundwater level change data are input as dynamic boundary conditions into a first three-dimensional numerical analysis model to obtain the gradient distribution characteristics of the pore water pressure field on the spoil heap slope, including:

[0070] B1. Input the rainfall intensity data as the rainfall infiltration boundary into the first three-dimensional numerical analysis model;

[0071] B2. Input the groundwater level change data as the groundwater level boundary into the first three-dimensional numerical analysis model;

[0072] B3. The pore water pressure gradient of each node of the first three-dimensional numerical analysis model is calculated based on the numerical difference method to obtain the pore water pressure field gradient distribution characteristics of the spoil disposal site slope.

[0073] It should be noted that the rainfall infiltration boundary refers to the boundary conditions in the first three-dimensional numerical analysis model that simulate how rainfall infiltrates into the slope surface. By inputting rainfall intensity data as the rainfall infiltration boundary into the model, the infiltration and distribution of water within the slope during actual rainfall can be simulated. The groundwater level boundary refers to the boundary conditions in the first three-dimensional numerical analysis model that represent the groundwater level. By inputting groundwater level change data as the groundwater level boundary into the model, the flow and distribution of groundwater in the actual slope can be simulated.

[0074] In an optional embodiment, a second three-dimensional numerical analysis model is obtained by dynamically inverting and calibrating the first three-dimensional numerical analysis model based on soil moisture content data and surface displacement data, including:

[0075] C1. Construct an inversion analysis module based on soil moisture content data and surface displacement data;

[0076] C2. Based on the particle swarm optimization algorithm and the inversion analysis module, the permeability coefficient parameters and soil shear strength parameters of the first three-dimensional numerical analysis model are dynamically inverted and calibrated to obtain the soil constitutive model parameter set with minimized error;

[0077] C3. A second three-dimensional numerical analysis model is obtained based on the parameter set of the soil constitutive model.

[0078] It should be noted that the inversion analysis module is an analytical framework that combines soil moisture content data and surface displacement data. It is used to calibrate the parameters of the first three-dimensional numerical analysis model. Through inversion analysis, actual monitoring data can be used to optimize model parameters, thereby improving the accuracy and reliability of the model. The particle swarm optimization algorithm is used to search for the optimal combination of model parameters to minimize the error between the model prediction results and the actual monitoring data. The soil constitutive model parameter set is a set of parameters describing the mechanical behavior of soil, including permeability coefficient, shear strength parameters, etc. Through inversion calibration, the soil constitutive model parameter set with the smallest error with the monitoring data can be obtained. The second three-dimensional numerical analysis model refers to the model after dynamic inversion calibration. Its parameters have been optimized and adjusted according to the actual monitoring data. The calibrated model can more accurately reflect the actual stability state of the spoil heap slope, providing a more reliable basis for slope treatment and design.

[0079] In an optional embodiment, a dynamic stability analysis of the spoil heap slope is performed based on a second three-dimensional numerical analysis model to obtain the safety factor of the spoil heap slope, including:

[0080] D1. Based on the strength reduction method, the soil shear strength parameters are reduced proportionally, and the reduced soil shear strength parameters are input into the second three-dimensional numerical analysis model.

[0081] D2. Perform a convergence check on the second three-dimensional numerical analysis model. If it does not converge, terminate the reduction process. At this point, the surface displacement rate predicted by the second three-dimensional numerical analysis model is the safety factor.

[0082] It should be noted that the strength reduction method is a numerical method used to assess slope stability. Its basic principle is to progressively reduce the shear strength parameters of the soil (such as the internal friction angle and cohesion) until the slope reaches a limit equilibrium state (i.e., an unstable state), thereby calculating the slope's safety factor. In numerical calculations, convergence judgment refers to the process of determining whether the calculation results tend to be stable. By performing convergence judgment on the second- and third-dimensional numerical analysis models, it can be determined whether the model has found a stable solution, i.e., whether the slope has reached a limit equilibrium state. If the model does not converge, it indicates that the slope may not be able to remain stable under current conditions. Surface displacement rate refers to the displacement velocity of monitoring points on the slope surface over time. It is one of the important indicators for assessing slope stability. An increase in surface displacement rate may indicate that the slope is experiencing instability or sliding. The safety factor reflects the stability state of the slope under current conditions. When the safety factor is lower than a preset threshold, it indicates that the slope's stability has decreased to an unacceptable level, requiring remedial measures and early warning.

[0083] In an optional embodiment, the convergence determination of the second three-dimensional numerical analysis model includes:

[0084] E1. Calculate the surface displacement increment predicted by the second three-dimensional numerical analysis model in the adjacent reduction steps;

[0085] E2. If the surface displacement increment is greater than the preset surface displacement increment, it is determined to be a displacement mutation, and thus the second three-dimensional numerical analysis model is determined to be non-convergent.

[0086] It should be noted that the surface displacement increment refers to the difference in surface displacement predicted by the second and third-dimensional numerical analysis models in adjacent reduction steps. This increment reflects the change in surface displacement of the slope during the parameter reduction process and can be used to assess slope stability. The preset surface displacement increment is a threshold used to determine whether the surface displacement increment is significant. In practical applications, a reasonable preset surface displacement increment value will be set according to the specific conditions and stability requirements of the slope. When the surface displacement increment exceeds this preset value, it indicates that the surface displacement of the slope has undergone a sudden change, which may mean that the stability state of the slope has changed significantly. A displacement mutation refers to the phenomenon of a sudden increase in the surface displacement increment. In slope stability analysis, a displacement mutation usually means that the stability state of the slope has changed significantly and instability may be imminent. When the second and third-dimensional numerical analysis models do not converge, it usually means that the slope has reached the limit equilibrium state (instability state). At this time, the model cannot continue to perform effective numerical calculations, the reduction process needs to be terminated, and the slope stability needs to be assessed based on the current results.

[0087] In an optional embodiment, a displacement field interpolation model is constructed based on a three-dimensional coordinate sequence, including:

[0088] F1. Based on the Kriging method, the coordinates of the surface deformation monitoring points of the spoil heap slope are interpolated according to the time window to obtain the displacement field sequence of the spoil heap slope.

[0089] F2. Construct a displacement field interpolation model based on the displacement field sequence.

[0090] It should be noted that Kriging is a spatial interpolation method that uses data from known points to estimate the values ​​at unknown points. In displacement field interpolation of spoil heap slopes, Kriging is used to interpolate deformation information at any location on the slope surface based on the coordinate data of deformation monitoring points, thereby constructing a continuous displacement field. A displacement field sequence refers to a series of displacement field data arranged in chronological order. The displacement field interpolation model is a mathematical model built based on the displacement field sequence, used to predict and simulate the displacement field distribution of the slope. The displacement field interpolation model can use the deformation data of known monitoring points to interpolate displacement information at any location on the slope surface, thus providing a more comprehensive understanding of the slope's deformation characteristics.

[0091] In an optional embodiment, the geometric parameters and spatial location of the slip surface of the spoil heap slope are obtained based on the displacement field interpolation model and the gradient distribution characteristics of the pore water pressure field, including:

[0092] G1. Obtain the displacement gradient of the spoil disposal site slope based on the displacement field interpolation model;

[0093] G2. Obtain the pore water pressure gradient of the spoil heap slope based on the gradient distribution characteristics of the pore water pressure field;

[0094] G3. The displacement gradient and pore water pressure gradient are fused based on preset weights to obtain a comprehensive gradient;

[0095] G4. Based on the gradient descent algorithm and the comprehensive gradient, the displacement change region of the spoil disposal site slope is tracked to obtain the geometric parameters and spatial location of the slip surface of the spoil disposal site slope.

[0096] It should be noted that the displacement gradient refers to the rate of change of the displacement field in space, reflecting the severity and direction of the deformation of the slope surface. The comprehensive gradient is the gradient obtained by fusing the displacement gradient and the pore water pressure gradient based on preset weights. In order to comprehensively consider the influence of deformation and moisture on slope stability, the displacement gradient and the pore water pressure gradient can be fused according to certain weights to obtain the comprehensive gradient. The gradient descent algorithm is an optimization algorithm used to find the minimum value of a function. The gradient descent algorithm is used to track the region of abrupt displacement change. By calculating the direction of the comprehensive gradient, the gradient descent algorithm can iteratively search along the direction of gradient descent until the minimum point of the region of abrupt displacement change is found, thereby determining the geometric parameters and spatial location of the sliding surface. The geometric parameters of the sliding surface refer to the parameters describing the shape and size of the sliding surface. The spatial location of the sliding surface refers to the specific location of the sliding surface in the slope, including the starting point, ending point, and sliding direction of the sliding surface.

[0097] In an optional embodiment, the second three-dimensional numerical analysis model is analyzed and optimized based on geometric parameters and spatial location to obtain a third three-dimensional numerical analysis model, including:

[0098] H1. Input the geometric parameters of the slip surface of the spoil heap slope into the Long Short-Term Memory network for time-series prediction. Based on the genetic algorithm, optimize the number of hidden layer neurons and the learning rate of the Long Short-Term Memory network online to obtain the early warning time window of the spoil heap slope.

[0099] H2. Based on the early warning time window, the predicted surface displacement rate of the second three-dimensional numerical analysis model is obtained. When the actual surface displacement exceeds the predicted surface displacement rate, the parameters of the second three-dimensional numerical analysis model are updated in a rolling manner based on the Kalman filter algorithm to obtain the third three-dimensional numerical analysis model.

[0100] It should be noted that Long Short-Term Memory (LSTM) networks are a special type of recurrent neural network capable of learning long-term dependencies in time-series data. Here, it is used to make time-series predictions based on the geometric parameters of the slip surface to assess the stability trend of the slope over future periods. Genetic algorithms are used to optimize the number of hidden layer neurons and the learning rate of the LTM network online to improve its predictive performance and accuracy. The number of hidden layer neurons in the LTM network determines the network's complexity and learning ability, significantly impacting its predictive performance. Optimizing the number of hidden layer neurons using genetic algorithms can help find the optimal network structure for predicting the stability of spoil heap slopes. Predicted surface displacement rate refers to the rate of surface displacement predicted by the LTM network over future periods based on historical data and the current state. This can be compared with the actual monitored surface displacement rate to verify the accuracy of the prediction. Rolling update refers to the process of continuously updating model parameters based on the latest monitoring data. Rolling update ensures that the model parameters remain consistent with the actual situation, thus reflecting changes in slope stability in a timely manner. By using the Kalman filter algorithm to continuously update the parameters of the second-dimensional numerical analysis model, an optimized model can be obtained, providing a more accurate basis for slope stability assessment.

[0101] In an optional embodiment, the parameters of the second three-dimensional numerical analysis model are updated in a rolling manner based on the Kalman filter algorithm to obtain the third three-dimensional numerical analysis model, including:

[0102] I1. Using the parameters of the second three-dimensional numerical analysis model as state variables and the real-time monitored surface displacement data as observation variables, a prediction error covariance matrix is ​​constructed based on the state variables, and an observation covariance matrix is ​​constructed based on the observation variables.

[0103] I2. Calculate the Kalman gain based on the prediction error covariance matrix and the observation covariance matrix, and update the state variables based on the observation variables and the Kalman gain to obtain the updated model parameters;

[0104] I3. Input the model parameters into the second three-dimensional numerical analysis model to obtain the third three-dimensional numerical analysis model.

[0105] It should be noted that the Kalman gain is a key parameter in the Kalman filter algorithm. It determines the weight of model predictions and observation data when updating state variables. By calculating the Kalman gain, the Kalman filter can intelligently fuse model predictions and observation data to minimize estimation errors and thus obtain more accurate estimates of state variables.

[0106] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting the sliding surface instability of a waste dump slope, comprising the steps of: collecting waste dump slope environmental data, including rainfall intensity data, groundwater level change data, soil moisture content data, and surface displacement data, and obtaining a three-dimensional coordinate sequence of a surface deformation monitoring point of the waste dump slope based on a Beidou satellite positioning system; constructing a first three-dimensional numerical analysis model of soil thermal-water-force coupling effects based on a finite difference method, inputting the rainfall intensity data and the groundwater level change data as dynamic boundary conditions into the first three-dimensional numerical analysis model, and obtaining the pore water pressure field gradient distribution characteristics of the waste dump slope; dynamically inverting and calibrating the first three-dimensional numerical analysis model based on the soil moisture content data and the surface displacement data, obtaining a second three-dimensional numerical analysis model, and dynamically analyzing the stability of the waste dump slope based on the second three-dimensional numerical analysis model to obtain the safety factor of the waste dump slope; when the safety factor is lower than a preset safety threshold, constructing a displacement field interpolation model based on the three-dimensional coordinate sequence, obtaining the geometric parameters and spatial position of the sliding surface of the waste dump slope based on the displacement field interpolation model and the pore water pressure field gradient distribution characteristics; analyzing and optimizing the second three-dimensional numerical analysis model based on the geometric parameters and spatial position to obtain a third three-dimensional numerical analysis model, and again dynamically analyzing the stability of the waste dump slope based on the third three-dimensional numerical analysis model to obtain a dynamic closed loop for predicting the stability state of the waste dump slope; constructing a first three-dimensional numerical analysis model of soil thermal-water-force coupling effects based on a finite difference method, comprising the steps of: dividing the three-dimensional geometric domain of the waste dump slope into a finite difference grid, which serves as a node of the three-dimensional numerical analysis model; establishing a first control equation of the first three-dimensional numerical analysis model based on the environmental data, wherein the first control equation includes a heat conduction equation, a seepage equation, and a stress balance equation; defining a thermal-water-force coupling relationship based on the first control equation; inputting the environmental data as boundary conditions of the first three-dimensional numerical analysis model; discretizing the first control equation based on an explicit difference method to obtain a second control equation with stronger stability; and constructing the first three-dimensional numerical analysis model based on the above steps; inputting the rainfall intensity data and the groundwater level change data as dynamic boundary conditions into the first three-dimensional numerical analysis model to obtain the pore water pressure field gradient distribution characteristics of the waste dump slope, comprising the steps of: inputting the rainfall intensity data as a rainfall infiltration boundary into the first three-dimensional numerical analysis model; inputting the groundwater level change data as a groundwater level boundary into the first three-dimensional numerical analysis model; and calculating the pore water pressure gradient of each node of the first three-dimensional numerical analysis model based on a numerical difference method to obtain the pore water pressure field gradient distribution characteristics of the waste dump slope. ​ ​ ​ ​ ​ 2. The method for predicting the sliding surface instability of the waste dump slope according to claim 1, characterized in that, ​ ​ ​ ​ ​ ​ ​ 3. The method for predicting the sliding surface instability of the waste dump slope according to claim 2, characterized in that: ​ ​ ​ ​ 4. The method for predicting sliding surface instability of a waste dump slope according to claim 3, characterized in that: Performing dynamic inversion calibration on the first three-dimensional numerical analysis model based on the soil moisture content data and the ground surface displacement data to obtain a second three-dimensional numerical analysis model, comprising: Constructing an inversion analysis module based on the soil moisture content data and the ground surface displacement data; Performing dynamic inversion calibration on the permeability coefficient parameters and the soil shear strength parameters of the first three-dimensional numerical analysis model based on a particle swarm optimization algorithm and the inversion analysis module to obtain a soil constitutive model parameter set with minimized error; Obtaining the second three-dimensional numerical analysis model based on the soil constitutive model parameter set.

5. The method for predicting the sliding surface instability of the waste dump slope according to claim 4, characterized in that, Performing dynamic stability analysis on the waste dump slope based on the second three-dimensional numerical analysis model to obtain a safety factor of the waste dump slope, comprising: Performing reduction on the soil shear strength parameters in proportion based on a strength reduction method, and inputting the reduced soil shear strength parameters into the second three-dimensional numerical analysis model; Performing convergence judgment on the second three-dimensional numerical analysis model, and if it does not converge, terminating the reduction, at which time the ground surface displacement rate predicted by the second three-dimensional numerical analysis model is the safety factor.

6. The method for predicting sliding surface instability of a waste dump slope according to claim 5, wherein Performing convergence judgment on the second three-dimensional numerical analysis model, comprising: Calculating the ground surface displacement increment predicted by the second three-dimensional numerical analysis model in the adjacent reduction step; If the ground surface displacement increment is greater than a preset ground surface displacement increment, determining that the displacement is suddenly changed, and thus determining that the second three-dimensional numerical analysis model does not converge.

7. The method for predicting sliding surface instability of a waste dump slope according to claim 5, characterized in that, Constructing a displacement field interpolation model based on the three-dimensional coordinate sequence, comprising: Interpolating the coordinates of the surface deformation monitoring points of the waste dump slope in a time window based on a Kriging method to obtain a displacement field sequence of the waste dump slope; Constructing the displacement field interpolation model based on the displacement field sequence.

8. The method for predicting sliding surface instability of a waste dump slope according to claim 3, characterized in that, Obtaining the geometric parameters and spatial positions of the sliding surface of the waste dump slope based on the displacement field interpolation model and the gradient distribution characteristics of the pore water pressure field, comprising: Obtaining the displacement gradient of the waste dump slope based on the displacement field interpolation model; Obtaining the pore water pressure gradient of the waste dump slope based on the gradient distribution characteristics of the pore water pressure field; Fusing the displacement gradient and the pore water pressure gradient based on a preset weight to obtain a comprehensive gradient; Tracking the displacement sudden change area of the waste dump slope based on a gradient descent algorithm and the comprehensive gradient to obtain the geometric parameters and spatial positions of the sliding surface of the waste dump slope.

9. The method for predicting sliding surface instability of a waste dump slope according to claim 8, characterized in that, Performing analysis and optimization on the second three-dimensional numerical analysis model based on the geometric parameters and spatial positions to obtain a third three-dimensional numerical analysis model, comprising: Inputting the geometric parameters of the sliding surface of the waste dump slope into a long short-term memory network to perform time series prediction, and performing online optimization on the number of hidden layer neurons and the learning rate of the long short-term memory network based on a genetic algorithm to obtain a warning time window of the waste dump slope; Obtaining the predicted ground surface displacement rate of the second three-dimensional numerical analysis model based on the warning time window, and when the actual ground surface displacement is monitored to exceed the predicted ground surface displacement rate, performing rolling update on the parameters of the second three-dimensional numerical analysis model based on a Kalman filtering algorithm to obtain the third three-dimensional numerical analysis model.

10. The method for predicting sliding surface instability of a waste dump slope according to claim 9, wherein, The parameters of the second three-dimensional numerical analysis model are updated based on a Kalman filtering algorithm to obtain a third three-dimensional numerical analysis model, and the third three-dimensional numerical analysis model comprises: The parameters of the second three-dimensional numerical analysis model are taken as state variables, real-time monitored surface displacement data are taken as observation variables, a prediction error covariance matrix is constructed based on the state variables, and an observation covariance matrix is constructed based on the observation variables; A Kalman gain is calculated based on the prediction error covariance matrix and the observation covariance matrix, and the state variables are updated based on the observation variables and the Kalman gain to obtain updated model parameters; The model parameters are input into the second three-dimensional numerical analysis model to obtain a third three-dimensional numerical analysis model.

Images