Landslide simulation method and system based on digital twinning
By constructing a digital twin slope model and combining rainfall and water pressure sensor data to simulate landslides, the problem of slow response speed in existing landslide monitoring technologies has been solved, enabling early warning and accurate prediction of landslides.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-10
- Publication Date
- 2026-03-10
AI Technical Summary
Existing landslide monitoring mainly relies on real-time sensor detection data, which has a slow response speed and makes it difficult to achieve significant early warning.
A digital twin slope model is constructed. By initializing the slope data and combining it with data from rain gauge and water pressure sensors, rainfall prediction and landslide simulation are performed. The landslide simulation is carried out using the rainfall prediction results, and the accuracy of the simulation results is verified by water pressure data.
This enables early prediction of landslides, improves the accuracy and response speed of prediction results, and ensures the effectiveness of digital twin slope models.
Smart Images
Figure CN121637759A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of landslide simulation technology, and particularly relates to a landslide simulation method and system based on digital twins. Background Technology
[0002] Landslide simulation is a technique that uses numerical models and computer technology to reproduce and predict the process of landslides occurring on mountains or slopes under specific geological, hydrological, and external force conditions. By inputting topographic features, soil and rock parameters, and triggering factors such as rainfall or earthquakes, simulations can dynamically demonstrate the initiation, movement, and deposition processes of landslides, assess the potential scope and impact of hazards, and provide a scientific basis for risk warning, disaster prevention planning, and engineering management. Commonly used simulation methods include physics-based continuous medium models (such as the finite element method), discrete element method, and empirical statistical models.
[0003] Existing landslide monitoring mainly relies on real-time sensor detection data, which makes it difficult to predict landslides. Due to its slow response speed, it is difficult to provide significant early warnings. Summary of the Invention
[0004] The purpose of this invention is to provide a landslide simulation method based on digital twins, which aims to solve the problems of existing landslide monitoring mainly relying on real-time sensor detection data, making it difficult to predict landslides, and due to its slow response speed, it is difficult to achieve significant early warning.
[0005] This invention is implemented as follows: a landslide simulation method based on digital twins, the method comprising: Data is collected from the slope to be monitored to obtain slope data, which includes at least geomorphological data, geological data and hydrological data. Multiple rain sensors are installed on the surface of the slope, and multiple water pressure sensors are installed at different depths and in geological layers within the slope. A digital twin slope model is constructed based on the collected slope data, and the parameters of the digital twin slope model are initialized according to the geological and hydrological data in the slope data. A rainfall monitoring grid is constructed based on a digital twin slope model to obtain measurement data from each rainfall sensor. Single-point rainfall distribution data is generated based on the rainfall monitoring grid, and rainfall prediction is performed based on the single-point rainfall distribution data to obtain the rainfall prediction results. Rainfall forecast results are imported into a digital twin slope model to simulate landslides, and simulation results are obtained. Water pressure data from water pressure sensors are collected, and the simulation results are verified based on the water pressure data. Verification results are then output.
[0006] Preferably, the step of constructing a digital twin slope model based on the collected slope data, and initializing the parameters of the digital twin slope model according to the geological and hydrological data in the slope data, specifically includes: Retrieve geomorphic data from the slope data, and construct the surface texture of the digital twin slope model based on the geomorphic data; Retrieve geological data from the slope data and determine the distribution of each geological layer in the digital twin slope model based on the geological data; Retrieve hydrological data from the slope data, set hydrological parameters for each geological layer based on the hydrological data, and complete parameter initialization.
[0007] Preferably, the steps of constructing a rainfall monitoring grid based on a digital twin slope model, acquiring measurement data from each rainfall sensor, generating single-point rainfall distribution data based on the rainfall monitoring grid, and performing rainfall prediction based on the single-point rainfall distribution data to obtain the rainfall prediction result specifically include: The horizontal projection of the monitoring area is obtained based on the digital twin slope model. Based on this, the rainfall monitoring grid is divided, and the grid position corresponding to each rainfall sensor is determined. The grid is divided into a metering grid and a prediction grid. The rainfall data of the metering grid comes from the rainfall sensor. By selecting multiple metering grids around the prediction grid as references, the rainfall data of the prediction grid is calculated, and the single-point rainfall distribution data is obtained. A rainfall prediction function for each grid is constructed based on the single-point rainfall distribution data of each grid. Rainfall prediction is then performed based on the margin prediction function to obtain the rainfall prediction result.
[0008] Preferably, the steps of importing rainfall prediction results into a digital twin slope model for landslide simulation, obtaining simulation results, collecting water pressure data from a water pressure sensor, verifying the simulation results based on the water pressure data, and outputting the verification results specifically include: The rainfall forecast results are imported into the digital twin slope model to perform landslide simulation calculations and obtain simulation results. The simulation results show the state of the digital twin slope model at each time point, including the water pressure simulation results at each location. Collect water pressure data from water pressure sensors, compare the water pressure data at each moment with the water pressure simulation results in chronological order, and determine the accuracy of the water pressure simulation results. The accuracy of the water pressure simulation results is used as the simulation confidence level of the digital twin slope model. When the simulation confidence level is lower than the preset value, the digital twin slope model is calibrated.
[0009] Preferably, the step of calibrating the digital twin slope model involves re-surveying the geomorphological, geological, and hydrological data of the area to be monitored, and then correcting the digital twin slope model.
[0010] Another object of the present invention is to provide a landslide simulation system based on digital twins, the system comprising: The data acquisition module is used to collect data from the slope to be monitored and obtain slope data. The slope data includes at least geomorphological data, geological data and hydrological data. Multiple rain sensors are installed on the surface of the slope, and multiple water pressure sensors are installed at different depths and in geological layers within the slope. The slope model construction module is used to construct a digital twin slope model based on the collected slope data, and to initialize the parameters of the digital twin slope model according to the geological and hydrological data in the slope data. The rainfall prediction module is used to construct a rainfall monitoring grid based on a digital twin slope model, acquire measurement data from each rainfall sensor, generate single-point rainfall distribution data based on the rainfall monitoring grid, and perform rainfall prediction based on the single-point rainfall distribution data to obtain the rainfall prediction result. The landslide simulation module is used to import rainfall prediction results into a digital twin slope model to simulate landslides, obtain simulation results, collect water pressure data from water pressure sensors, verify the simulation results based on the water pressure data, and output the verification results.
[0011] Preferably, the slope model construction module includes: The terrain construction unit is used to retrieve terrain data from the slope data and construct the surface texture of the digital twin slope model based on the terrain data; The geological layer construction unit is used to retrieve geological data from the slope data and determine the distribution location of each geological layer in the digital twin slope model based on the geological data. The hydrological parameter setting unit is used to retrieve hydrological data from the slope data, set the hydrological parameters of each geological layer based on the hydrological data, and complete the parameter initialization.
[0012] Preferably, the rainfall prediction module includes: The rainfall acquisition unit is used to obtain the horizontal plane projection of the monitoring area based on the digital twin slope model. On this basis, the rainfall monitoring grid is divided, the grid position corresponding to each rainfall sensor is determined, and the grid is divided into a metering grid and a prediction grid. The rainfall data of the metering grid comes from the rainfall sensor. The rainfall distribution calculation unit is used to select multiple metering grids around the prediction grid as references, calculate the rainfall data of the prediction grid, and obtain the single-point rainfall distribution data. The global rainfall prediction unit is used to construct the rainfall prediction function for each grid based on the single-point rainfall distribution data of each grid, and to perform rainfall prediction based on the margin prediction function to obtain the rainfall prediction result.
[0013] Preferably, the landslide simulation module includes: The landslide simulation calculation unit is used to import rainfall prediction results into the digital twin slope model to perform landslide simulation calculations and obtain simulation results. The simulation results are the state of the digital twin slope model at each time, including the water pressure simulation results at each location. The water pressure verification unit is used to collect water pressure data from the water pressure sensor, compare the water pressure data at that moment with the water pressure simulation results in chronological order, and determine the accuracy of the water pressure simulation results. The model calibration unit is used to use the accuracy of the water pressure simulation results as the simulation confidence level of the digital twin slope model. When the simulation confidence level is lower than a preset value, the digital twin slope model is calibrated.
[0014] Preferably, the step of calibrating the digital twin slope model involves re-surveying the geomorphological, geological, and hydrological data of the area to be monitored, and then correcting the digital twin slope model.
[0015] This invention provides a landslide simulation method based on digital twins. By constructing a digital twin slope model of the slope to be monitored, real-time rainfall data is imported into the digital twin slope model. Based on the simulation results, the landslide situation of the slope is predicted, thereby predicting in advance whether a landslide will occur. Furthermore, the prediction results are verified by water pressure to ensure the effectiveness of the digital twin slope model and the accuracy of the prediction results. Attached Figure Description
[0016] Figure 1 A flowchart illustrating a landslide simulation method based on digital twins, provided as an embodiment of the present invention; Figure 2 A flowchart illustrating the steps of constructing a digital twin slope model and initializing parameters provided in an embodiment of the present invention; Figure 3 A flowchart illustrating the steps for generating rainfall prediction results provided in an embodiment of the present invention; Figure 4 A flowchart illustrating the steps for outputting simulation results and verification results provided in embodiments of the present invention; Figure 5 An architecture diagram of a landslide simulation system based on digital twins provided in an embodiment of the present invention; Figure 6 An architecture diagram of a slope model construction module provided in an embodiment of the present invention; Figure 7 An architecture diagram of a rainfall prediction module provided in an embodiment of the present invention; Figure 8 This is an architectural diagram of a landslide simulation module provided in an embodiment of the present invention. Detailed Implementation
[0017] 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. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0018] like Figure 1 The diagram shows a flowchart of a landslide simulation method based on digital twins provided by an embodiment of the present invention. The method includes: S100: Data is collected from the slope to be monitored to obtain slope data. The slope data includes at least geomorphological data, geological data, and hydrological data. Multiple rainfall sensors are installed on the surface of the slope, and multiple water pressure sensors are installed at different depths and in geological layers within the slope.
[0019] In this step, data is collected from the slope to be monitored to obtain slope data. Different methods are used for different slope data. For geomorphological data, scanning can be performed using an airborne LiDAR module or machine vision to scan the slope to determine the elevation of various points on the surface of the slope. For geological data, the geological data is used to determine the distribution of geological layers and geomechanical parameters of the slope. The distribution of geological layers includes, for example, residual slope deposits with a thickness of 1-5 meters, thick sandstone interbedded with thin mudstone layers with a thickness of 20-30 meters, and the rock layers dipping towards the mountain. The slope has a uniform dip angle of approximately 25°, consisting of mudstone layers with a thickness of 5-15 meters, and siltstone as the underlying bedrock. Geomechanical parameters include the shear strength, unit weight, elastic modulus, and Poisson's ratio of the geological layers. Hydrological data are obtained through field measurements or based on historical experience. Hydrological data for each geological layer include saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability function, and porosity. To continuously monitor the slope's condition, multiple rain gauges, such as piezoelectric rain gauges, are installed on the slope surface. Water pressure sensors are embedded at different depths in different geological layers to detect underground water pressure.
[0020] S200 constructs a digital twin slope model based on the collected slope data, and initializes the parameters of the digital twin slope model according to the geological and hydrological data in the slope data.
[0021] In this step, a digital twin slope model is constructed based on the collected slope data. Geomorphological data from the slope data is retrieved, and the surface structural features of the digital twin slope model are constructed based on the geomorphological data to determine the texture structure of the model surface. Then, geological layers are filled into the digital twin slope model according to geological data, and the division, range, and depth of each geological layer are determined. Different geological layers are assigned different geomechanical parameters. Finally, different geological layers are assigned corresponding hydrological parameters based on hydrological data, such as saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability function, and porosity. After the settings are completed, the initialization of the digital twin slope model is completed. Subsequently, landslide simulation is carried out using rainfall as the trigger event for landslides.
[0022] The S300 constructs a rainfall monitoring grid based on a digital twin slope model, acquires measurement data from various rainfall sensors, generates single-point rainfall distribution data based on the rainfall monitoring grid, and performs rainfall prediction based on the single-point rainfall distribution data to obtain the rainfall prediction results.
[0023] In this step, a rainfall monitoring grid is constructed based on a digital twin slope model. The digital twin slope model is projected onto a horizontal plane to obtain a horizontal projection area. This horizontal projection area is divided into multiple grid cells. The locations of each rainfall sensor are determined, and the rainfall sensors are bound to their corresponding grid cells. The rainfall detected by the rainfall sensor is used as the rainfall amount for the corresponding bound grid cell. For unbound grid cells, the rainfall amount for the unbound grid cells is estimated based on the grid cells bound to adjacent rainfall sensors, thus obtaining single-point rainfall distribution data. After determining the rainfall amount for each grid cell, a rainfall prediction function is constructed for each grid cell based on the recorded single-point rainfall distribution data. Rainfall prediction is then performed for that grid cell to obtain the rainfall prediction result. The rainfall prediction result is used to predict the rainfall amount for each grid cell in future time periods.
[0024] The S400 system imports rainfall forecast results into a digital twin slope model to simulate landslides, obtains simulation results, collects water pressure data from a water pressure sensor, verifies the simulation results based on the water pressure data, and outputs the verification results.
[0025] In this step, rainfall forecast results are imported into a digital twin slope model for landslide simulation. The impact of rainfall is simulated and extrapolated using the digital twin slope model to determine the slope's state at various future time points. Water level data at each point in the digital twin slope model at each extrapolated time point is output, and the simulated water pressure at that point is determined and recorded in the simulation results. Based on the simulation results, the risk of landslides within the predicted time period is determined. Water pressure data from a water pressure sensor is collected in real time, and the simulated water pressure data is verified against the measured water pressure data. If the two do not match, it indicates that the internal structure of the monitored slope has deteriorated during long-term monitoring. The hydrogeological conditions have changed, and the previously established digital twin slope model no longer conforms to the actual situation and needs to be corrected. Therefore, the accuracy of the digital twin slope model can be evaluated during daily rainfall to complete the model correction and avoid affecting the accuracy of landslide simulation using the model. In this invention, the rainfall is derived based on each grid cell, which can more accurately determine the rainfall at each location on the slope, making the rainfall in the digital twin slope model more consistent with reality and improving the accuracy of landslide simulation. In the prior art, rainfall is derived from weather forecasts, but the rainfall at each time does not match the actual rainfall, affecting the accuracy of landslide simulation.
[0026] like Figure 2 As shown, in a preferred embodiment of the present invention, the step of constructing a digital twin slope model based on the collected slope data and initializing the parameters of the digital twin slope model according to the geological and hydrological data in the slope data specifically includes: S201, retrieve the geomorphic data from the slope data, and construct the surface texture of the digital twin slope model based on the geomorphic data.
[0027] In this step, the topographic data in the slope data is retrieved. The topographic data can be obtained by measuring the terrain below using a drone equipped with a laser rangefinder, or by recognizing it using a binocular vision solution to generate a depth map of the slope, thereby generating the corresponding topographic data, and thus constructing a digital twin slope model to form the surface texture of the digital twin slope model.
[0028] S202, retrieve the geological data from the slope data, and determine the distribution location of each geological layer in the digital twin slope model based on the geological data.
[0029] S203: Retrieve hydrological data from the slope data, set hydrological parameters for each geological layer based on the hydrological data, and complete parameter initialization.
[0030] In this step, geological data from the slope data is retrieved, and the digital twin slope model is filled layer by layer based on the geological data to determine the distribution location of each geological layer and the corresponding geomechanical parameters. After assigning the above parameters, hydromechanical parameters, including saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability coefficient function, and porosity, are assigned to the digital twin slope model based on hydrological data, thus confirming the initial state of the digital twin slope model.
[0031] like Figure 3 As shown, in a preferred embodiment of the present invention, the steps of constructing a rainfall monitoring grid based on a digital twin slope model, acquiring measurement data from each rainfall sensor, generating single-point rainfall distribution data based on the rainfall monitoring grid, and performing rainfall prediction based on the single-point rainfall distribution data to obtain the rainfall prediction result specifically include: S301 obtains the horizontal plane projection of the monitoring area based on the digital twin slope model, divides the rainfall monitoring grid on this basis, determines the grid position corresponding to each rainfall sensor, and divides the grid into a metering grid and a prediction grid. The rainfall data of the metering grid comes from the rainfall sensor.
[0032] In this step, the horizontal projection of the monitoring area is obtained based on the digital twin slope model. That is, the digital twin slope model is projected onto the horizontal plane to obtain the horizontal projection of the monitoring area. The entire horizontal projection of the monitoring area is divided into multiple fixed-size blocks according to a preset size. Each block is a rainfall monitoring grid, which is a grid unit. At the same time, the position of the rainfall sensor is also projected onto the horizontal plane and finally falls within a grid. The grid containing the rainfall sensor is the metering grid. The rainfall in the metering grid can be directly measured by the rainfall sensor. The remaining grids are prediction grids. No rainfall sensor is set in the prediction grid. The rainfall in the prediction grid is calculated based on the rainfall in the adjacent metering grid.
[0033] S302: Select multiple metering grids around the prediction grid as references, calculate the rainfall data of the prediction grid, and obtain the single-point rainfall distribution data.
[0034] In this step, each time a prediction grid is selected, the nearest multiple metering grids are used as references to calculate the rainfall for that prediction grid. Specifically, the rainfall of the i-th surrounding metering grid per unit time is obtained. Based on the distance between the measurement grid and the prediction grid The weight of the econometric grid is determined by the proximity of it to the prediction grid; the closer the econometric grid is to the prediction grid, the more significant its influence. The weight of the i-th econometric grid is... Represented as: in, This is a power parameter, the value of which is set according to requirements. In this embodiment, it is set to 2. The rainfall per unit time for this predicted grid is expressed as: The following example illustrates this: the distances between prediction grid A and its five adjacent metering grids B, C, D, E, and F are 3.24, 4.75, 3.8, 4.2, and 3.55 meters, respectively; and the hourly rainfall rates for the five metering grids B, C, D, E, and F are 18.5, 19.2, 18.0, 19.0, and 18.8 mm / h, respectively. The predicted rainfall for grid A is expressed as: Based on this, the unit time rainfall of each prediction grid at that moment is determined, and the single-point rainfall distribution data is obtained.
[0035] S303: Construct a rainfall prediction function for each grid based on the single-point rainfall distribution data of each grid, and perform rainfall prediction based on the margin prediction function to obtain the rainfall prediction result.
[0036] In this step, a rainfall prediction function for each grid is constructed based on the single-point rainfall distribution data of each grid. The single-point rainfall distribution data of each grid at each time is statistically analyzed and represented in coordinate form, with the horizontal axis representing time and the vertical axis representing the rainfall per unit time. This constructs a series of rainfall coordinates belonging to the grid. The rainfall prediction function corresponding to the grid is output through function fitting. The rainfall prediction function outputs the rainfall per unit time at each time in the future, thus obtaining the rainfall prediction result.
[0037] like Figure 4 As shown, in a preferred embodiment of the present invention, the steps of importing rainfall prediction results into a digital twin slope model for landslide simulation, obtaining simulation results, collecting water pressure data from a water pressure sensor, verifying the simulation results based on the water pressure data, and outputting the verification results specifically include: S401. Import the rainfall forecast results into the digital twin slope model to perform landslide simulation calculations and obtain simulation results. The simulation results show the state of the digital twin slope model at each time point, including the water pressure simulation results at each location.
[0038] In this step, the rainfall forecast results are imported into the digital twin slope model. Specifically, landslide simulation can be performed using MatDEM. During the simulation, the water pressure simulation results at each point in the digital twin slope model are recorded in real time. The landslide situation at each moment in the digital twin slope model is also recorded, as well as the time node when the landslide occurs.
[0039] S402: Collect water pressure data from the water pressure sensor, compare the water pressure data at that moment with the water pressure simulation results in chronological order, and determine the accuracy of the water pressure simulation results.
[0040] S403 uses the accuracy of the water pressure simulation results as the simulation confidence level of the digital twin slope model. When the simulation confidence level is lower than the preset value, the digital twin slope model is calibrated.
[0041] In this step, water pressure data from water pressure sensors is collected. Based on the water pressure simulation results, the simulated water pressure at corresponding times is calculated step by step. The difference between the two is calculated. When the difference is lower than the preset pressure value, the water pressure simulation result is considered accurate. The number of accurate water pressure simulation results and the total number of water pressure simulation results are counted. The ratio of the two is the accuracy rate of the water pressure simulation results. The more accurate the water pressure simulation results, the closer the current digital twin slope model is to the actual slope structure. When the accuracy rate decreases, it indicates that the actual geological and hydrological parameters of the slope have changed, and the prediction results generated based on the digital twin slope model are also inaccurate. When the simulation confidence is lower than the preset value, such as below 50%, it indicates that the geological structure or hydrological parameters have changed, and the digital twin slope model needs to be calibrated, that is, the above data is re-collected to correct the digital twin slope model. Based on the landslide simulation results, it is determined whether a landslide will occur within the prediction time and whether there is a landslide trend, so as to generate early warning information.
[0042] like Figure 5 As shown, an embodiment of the present invention provides a landslide simulation system based on digital twins, the system comprising: The data acquisition module 100 is used to acquire data from the slope to be monitored and obtain slope data. The slope data includes at least geomorphological data, geological data and hydrological data. Multiple rain sensors are installed on the surface of the slope, and multiple water pressure sensors are installed at different depths and in geological layers within the slope.
[0043] In this system, the data acquisition module 100 collects data from the slope to be monitored, obtaining slope data. Different methods are used for different slope data. For geomorphological data, scanning can be performed using an airborne LiDAR module or machine vision to scan the slope to determine the elevation of various points on the surface of the slope. For geological data, the geological data is used to determine the distribution of geological layers and geomechanical parameters of the slope. The distribution of geological layers includes, for example, residual slope deposits with a thickness of 1-5 meters, thick sandstone interbedded with thin mudstone layers with a thickness of 20-30 meters, and so on. The slope dips in the same direction as the hillside, at an angle of approximately 25°. The layers consist of mudstone, 5-15 meters thick, and siltstone as the underlying bedrock. Geomechanical parameters include the shear strength, unit weight, elastic modulus, and Poisson's ratio of the geological layers. Hydrological data are obtained through field measurements or based on historical experience. Hydrological data for each geological layer include saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability function, and porosity. To continuously monitor the slope's condition, multiple rain gauges are installed on the slope surface. These rain gauges can be piezoelectric rain gauges, while water pressure sensors are embedded at different depths in different geological layers to detect underground water pressure.
[0044] The slope model construction module 200 is used to construct a digital twin slope model based on the collected slope data, and to initialize the parameters of the digital twin slope model according to the geological and hydrological data in the slope data.
[0045] In this system, the slope model construction module 200 constructs a digital twin slope model based on the collected slope data. It retrieves geomorphological data from the slope data, constructs the surface structural features of the digital twin slope model based on the geomorphological data, determines the texture structure of the model surface, and then fills the digital twin slope model with geological layers according to geological data. It determines the division, range, and depth of each geological layer and assigns different geomechanical parameters to different geological layers. Finally, it assigns corresponding hydrological parameters to different geological layers according to hydrological data, such as saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability coefficient function, and porosity. After the settings are completed, the initialization of the digital twin slope model is completed. Subsequently, landslide simulation is performed using rainfall as the trigger event for landslides.
[0046] The rainfall prediction module 300 is used to construct a rainfall monitoring grid based on a digital twin slope model, acquire measurement data from various rainfall sensors, generate single-point rainfall distribution data based on the rainfall monitoring grid, and perform rainfall prediction based on the single-point rainfall distribution data to obtain the rainfall prediction result.
[0047] In this system, the rainfall prediction module 300 constructs a rainfall monitoring grid based on a digital twin slope model. The digital twin slope model is projected onto a horizontal plane to obtain a horizontal projection area. This horizontal projection area is divided into multiple grid cells. The locations of each rainfall sensor are determined, and the sensors are bound to their corresponding grid cells. The rainfall detected by the sensor is used as the rainfall amount for the bound grid cell. For unbound grid cells, the rainfall is estimated based on the grid cells bound to adjacent sensors, obtaining single-point rainfall distribution data. After determining the rainfall amount for each grid cell, a rainfall prediction function is constructed for each grid cell based on the recorded single-point rainfall distribution data. Rainfall prediction is then performed on this function to obtain the rainfall prediction result. This result is used to predict the rainfall amount for each grid cell in future time periods.
[0048] The landslide simulation module 400 is used to import rainfall prediction results into a digital twin slope model to simulate landslides, obtain simulation results, collect water pressure data from a water pressure sensor, verify the simulation results based on the water pressure data, and output the verification results.
[0049] In this system, the landslide simulation module 400 imports rainfall forecast results into a digital twin slope model for landslide simulation. It uses the digital twin slope model to simulate and extrapolate the impact of rainfall, determining the slope's state at various future time points. It outputs water level data at each point in the digital twin slope model at each extrapolation time node, determining the simulated water pressure at that point and recording it in the simulation results. Based on the simulation results, it determines whether there is a risk of landslide within the predicted time period. Furthermore, it verifies the simulated water pressure data by real-time acquisition of water pressure data from water pressure sensors and comparing it with the measured water pressure data. If the two do not match, it indicates that the monitored slope has experienced landslides during a long monitoring period. In this process, the internal hydrogeological conditions have changed, and the previously established digital twin slope model no longer conforms to the actual situation and needs to be corrected. Therefore, during daily rainfall, the accuracy of the digital twin slope model can be evaluated to complete the model correction and avoid affecting the accuracy of landslide simulation using the model. In this invention, the rainfall is derived based on each grid cell, which can more accurately determine the rainfall at each location on the slope, making the rainfall in the digital twin slope model more consistent with reality and improving the accuracy of landslide simulation. In the prior art, rainfall is derived from weather forecasts, but the rainfall at each time does not match the actual rainfall, affecting the accuracy of landslide simulation.
[0050] like Figure 6 As shown, in a preferred embodiment of the present invention, the slope model construction module 200 includes: The terrain construction unit 201 is used to retrieve terrain data from the slope data and construct the surface texture of the digital twin slope model based on the terrain data.
[0051] In this module, the terrain construction unit 201 retrieves terrain data from the slope data. The terrain data can be obtained by measuring the terrain below using a drone equipped with a laser rangefinder, or by recognizing it using a binocular vision scheme to generate a depth map of the slope, thereby generating the corresponding terrain data and constructing a digital twin slope model to form the surface texture of the digital twin slope model.
[0052] Geological layer construction unit 202 is used to retrieve geological data from the slope data and determine the distribution location of each geological layer in the digital twin slope model based on the geological data.
[0053] The hydrological parameter setting unit 203 is used to retrieve hydrological data from the slope data, set the hydrological parameters of each geological layer based on the hydrological data, and complete the parameter initialization.
[0054] In this module, geological data from the slope data is retrieved, and the digital twin slope model is filled layer by layer based on the geological data to determine the distribution location of each geological layer and the corresponding geomechanical parameters. After assigning the above parameters, hydromechanical parameters are assigned to the digital twin slope model based on hydrological data, including saturated permeability coefficient, soil-water characteristic curve, unsaturated permeability coefficient function, and porosity, thus confirming the initial state of the digital twin slope model.
[0055] like Figure 7 As shown, in a preferred embodiment of the present invention, the rainfall prediction module 300 includes: Rainfall acquisition unit 301 is used to obtain the horizontal plane projection of the monitoring area based on the digital twin slope model, divide the rainfall monitoring grid on this basis, determine the grid position corresponding to each rainfall sensor, and divide the grid into a metering grid and a prediction grid. The rainfall data of the metering grid comes from the rainfall sensor.
[0056] In this module, the rainfall acquisition unit 301 obtains the horizontal projection of the monitoring area based on the digital twin slope model. That is, the digital twin slope model is projected onto the horizontal plane to obtain the horizontal projection of the monitoring area. The entire horizontal projection of the monitoring area is divided into multiple fixed-size blocks according to the preset size. Each block is a rainfall monitoring grid, which is a grid unit. At the same time, the position of the rainfall sensor is also projected onto the horizontal plane and finally falls within a grid. The grid containing the rainfall sensor is the metering grid. The rainfall in the metering grid can be directly measured by the rainfall sensor. The remaining grids are prediction grids. No rainfall sensor is set in the prediction grid. The rainfall in the prediction grid is calculated based on the rainfall in the adjacent metering grid.
[0057] The rainfall distribution calculation unit 302 is used to select multiple metering grids around the prediction grid as references, calculate the rainfall data of the prediction grid, and obtain single-point rainfall distribution data.
[0058] In this module, the rainfall distribution calculation unit 302 selects a prediction grid at a time and uses the multiple metering grids closest to that prediction grid as references to calculate the rainfall for that prediction grid. Specifically, it obtains the rainfall of the i-th surrounding metering grid within a unit of time. Based on the distance between the measurement grid and the prediction grid The weight of the econometric grid is determined by the proximity of it to the prediction grid; the closer the econometric grid is to the prediction grid, the more significant its influence. The weight of the i-th econometric grid is... , is represented as: in, This is a power parameter, the value of which is set according to requirements. In this embodiment, it is set to 2. The rainfall per unit time for this predicted grid is expressed as: The following example illustrates this: the distances between prediction grid A and its five adjacent metering grids B, C, D, E, and F are 3.24, 4.75, 3.8, 4.2, and 3.55 meters, respectively; and the hourly rainfall rates for the five metering grids B, C, D, E, and F are 18.5, 19.2, 18.0, 19.0, and 18.8 mm / h, respectively. The predicted rainfall for grid A is expressed as: Based on this, the unit time rainfall of each prediction grid at that moment is determined, and the single-point rainfall distribution data is obtained.
[0059] The global rainfall prediction unit 303 is used to construct the rainfall prediction function of each grid based on the single-point rainfall distribution data of each grid, and to perform rainfall prediction based on the margin prediction function to obtain the rainfall prediction result.
[0060] In this module, the global rainfall prediction unit 303 constructs a rainfall prediction function for each grid based on the single-point rainfall distribution data of each grid. It statistically analyzes the single-point rainfall distribution data of each grid at each time and represents it in coordinate form, with the horizontal axis representing time and the vertical axis representing the rainfall per unit time. This constructs a series of rainfall coordinates belonging to the grid. The rainfall prediction function corresponding to the grid is output through function fitting. The rainfall prediction function outputs the rainfall per unit time at each time in the future, thus obtaining the rainfall prediction result.
[0061] like Figure 8 As shown, in a preferred embodiment of the present invention, the landslide simulation module 400 includes: The landslide simulation calculation unit 401 is used to import the rainfall prediction results into the digital twin slope model to perform landslide simulation calculations and obtain simulation results. The simulation results are the state of the digital twin slope model at each time, including the water pressure simulation results at each location.
[0062] In this module, the landslide simulation calculation unit 401 imports the rainfall prediction results into the digital twin slope model. Specifically, landslide simulation can be performed using MatDEM. During the simulation, the water pressure simulation results at each point in the digital twin slope model are recorded in real time. The water pressure simulation results are also recorded, as well as the landslide situation at each moment in the digital twin slope model and the time node of the landslide occurrence.
[0063] The water pressure verification unit 402 is used to collect water pressure data from the water pressure sensor, compare the water pressure data corresponding to that moment with the water pressure simulation results in chronological order, and determine the accuracy of the water pressure simulation results.
[0064] The model calibration unit 403 is used to use the accuracy of the water pressure simulation results as the simulation confidence of the digital twin slope model. When the simulation confidence is lower than a preset value, the digital twin slope model is calibrated.
[0065] In this module, water pressure data from water pressure sensors is collected. Based on the water pressure simulation results, the simulated water pressure at corresponding times is calculated step by step. The difference between the two is calculated. When the difference is lower than the preset pressure value, the water pressure simulation result is considered accurate. The number of accurate water pressure simulation results and the total number of water pressure simulation results are counted. The ratio of the two is the accuracy rate of the water pressure simulation results. The more accurate the water pressure simulation results, the closer the current digital twin slope model is to the actual slope structure. When the accuracy rate decreases, it indicates that the actual geological and hydrological parameters of the slope have changed, and the prediction results generated based on the digital twin slope model are also inaccurate. When the simulation confidence is lower than the preset value, such as below 50%, it indicates that the geological structure or hydrological parameters have changed, and the digital twin slope model needs to be calibrated, that is, the above data is re-collected to correct the digital twin slope model. Based on the landslide simulation results, it is determined whether a landslide will occur within the prediction time and whether there is a landslide trend, so as to generate early warning information.
[0066] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A landslide simulation method based on digital twinning, characterized by, The method comprises: data acquisition is performed on a slope body to be monitored to obtain slope body data, the slope body data at least comprising topographic data, geological data and hydrological data, a plurality of rainfall sensors are arranged on the surface of the slope body, and a plurality of water pressure sensors are arranged in different depths and geological layers in the slope body; a digital twin slope body model is constructed based on the collected slope body data, and the digital twin slope body model is parameterized according to the geological data and the hydrological data in the slope body data; a rainfall monitoring grid is constructed based on the digital twin slope body model, measurement data of each rainfall sensor is obtained, single-point rainfall distribution data is generated according to the rainfall monitoring grid, rainfall prediction is performed based on the single-point rainfall distribution data, and a rainfall prediction result is obtained; the rainfall prediction result is imported into the digital twin slope body model to perform landslide simulation, a simulation result is obtained, water pressure data of the water pressure sensors is collected, the simulation result is verified based on the water pressure data, and a verification result is output. 2.The landslide simulation method based on digital twinning according to claim 1, wherein, The step of constructing the digital twin slope body model based on the collected slope body data and parameterizing the digital twin slope body model according to the geological data and the hydrological data in the slope body data specifically comprises: topographic data in the slope body data is called, and surface texture of the digital twin slope body model is constructed based on the topographic data; geological data in the slope body data is called, and distribution positions of each geological layer in the digital twin slope body model are determined according to the geological data; hydrological data in the slope body data is called, and hydrological parameters of each geological layer are set based on the hydrological data, and parameterization is completed. 3.The digital-twin-based landslide simulation method of claim 1, wherein, The step of constructing the rainfall monitoring grid based on the digital twin slope body model, obtaining measurement data of each rainfall sensor, generating single-point rainfall distribution data according to the rainfall monitoring grid, performing rainfall prediction based on the single-point rainfall distribution data, and obtaining a rainfall prediction result specifically comprises: a horizontal plane projection of a monitoring area is obtained based on the digital twin slope body model, the rainfall monitoring grid is divided based on the horizontal plane projection, the grid position corresponding to each rainfall sensor is determined, and the grid is divided into a measurement grid and a prediction grid, and rainfall data of the measurement grid is derived from the rainfall sensor; a plurality of measurement grids around the prediction grid are selected as references, rainfall data of the prediction grid is calculated, and single-point rainfall distribution data is obtained; a rainfall prediction function of each grid is constructed based on the single-point rainfall distribution data of the grid, rainfall prediction is performed based on the rainfall prediction function, and a rainfall prediction result is obtained. 4.The landslide simulation method based on digital twinning according to claim 1, wherein, The step of importing the rainfall prediction result into the digital twin slope body model to perform landslide simulation, obtaining a simulation result, collecting water pressure data of the water pressure sensors, verifying the simulation result based on the water pressure data, and outputting a verification result specifically comprises: the rainfall prediction result is imported into the digital twin slope body model to perform landslide simulation calculation, a simulation result is obtained, the simulation result is a state of the digital twin slope body model at each moment, including water pressure simulation results at each position; water pressure data corresponding to the moment is compared with the water pressure simulation result in chronological order, and the accuracy of the water pressure simulation result is determined. The accuracy of the water pressure simulation result is used as the simulation confidence of the digital twin slope model, and when the simulation confidence is lower than a preset value, the digital twin slope model is calibrated. 5.The landslide simulation method based on digital twinning according to claim 4, characterized in that, The step of calibrating the digital twin slope model comprises re-surveying the topographic data, geological data and hydrological data of the to-be-monitored region, and modifying the digital twin slope model. 6.A landslide simulation system based on digital twinning, characterized in that, The system comprises: a data acquisition module configured to acquire data of a to-be-monitored slope to obtain slope data, the slope data comprising at least topographic data, geological data and hydrological data, the surface of the slope being provided with a plurality of rainfall sensors, and the slope being provided with a plurality of water pressure sensors at different depths and geological layers; a slope model construction module configured to construct a digital twin slope model based on the acquired slope data, and to initialize parameters of the digital twin slope model according to the geological data and the hydrological data in the slope data; a rainfall prediction module configured to construct a rainfall monitoring grid based on the digital twin slope model, to acquire measurement data of each rainfall sensor, to generate single-point rainfall distribution data according to the rainfall monitoring grid, to predict rainfall based on the single-point rainfall distribution data, and to obtain a rainfall prediction result; a landslide simulation module configured to import the rainfall prediction result into the digital twin slope model to simulate a landslide, to acquire water pressure data of the water pressure sensors, to verify the simulation result based on the water pressure data, and to output a verification result.
7. The digital-twin-based landslide simulation system of claim 6, wherein, The slope model construction module comprises: a topography construction unit configured to call the topographic data in the slope data, and to construct surface textures of the digital twin slope model based on the topographic data; a geological layer construction unit configured to call the geological data in the slope data, and to determine distribution positions of each geological layer in the digital twin slope model according to the geological data; a hydrological parameter setting unit configured to call the hydrological data in the slope data, to set hydrological parameters of each geological layer based on the hydrological data, and to complete parameter initialization.
8. The digital-twin-based landslide simulation system of claim 6, wherein, The rainfall prediction module comprises: a rainfall acquisition unit configured to acquire a horizontal projection of a monitoring region based on the digital twin slope model, to divide a rainfall monitoring grid based on the horizontal projection, to determine a grid position corresponding to each rainfall sensor, to divide the grid into a measurement grid and a prediction grid, and to acquire rainfall data of the measurement grid from the rainfall sensor; a rainfall distribution calculation unit configured to select a plurality of measurement grids around a prediction grid as a reference center, to calculate rainfall data of the prediction grid, and to obtain single-point rainfall distribution data; a global rainfall prediction unit configured to construct a rainfall prediction function of each grid based on the single-point rainfall distribution data of the grid, to predict rainfall based on the rainfall prediction function, and to obtain a rainfall prediction result.
9. The digital-twin-based landslide simulation system of claim 6, wherein, The landslide simulation module comprises: a landslide simulation calculation unit configured to import the rainfall prediction result into the digital twin slope model, to perform landslide simulation calculation, to obtain a simulation result, and to determine a state of the digital twin slope model at each time point, including water pressure simulation results at each position; a water pressure verification unit configured to acquire water pressure data of the water pressure sensors, to compare the water pressure data corresponding to the time point with the water pressure simulation result in a time sequence, to determine an accuracy of the water pressure simulation result, and to output a verification result. A model calibration unit is configured to take the accuracy of the water pressure simulation result as the simulation confidence of the digital twin slope model, and calibrate the digital twin slope model when the simulation confidence is lower than a preset value.
10. The digital-twin-based landslide simulation system of claim 9, wherein, The calibration of the digital twin slope model is achieved by re-surveying the topographic data, geological data and hydrological data of the area to be monitored, and modifying the digital twin slope model.