Slope stability real-time calculation method and system

Abstract

The invention relates to the field of slope stability calculation, in particular to a slope stability real-time calculation method and system. The method comprises the following steps: constructing a three-dimensional geologic model of the slope and dividing the three-dimensional geologic model into three-dimensional strips; based on the monitoring result of the three-dimensional strip block and the three-dimensional geologic model, extracting spatial features and monitoring time sequence data of the three-dimensional strip block; constructing a three-dimensional strip block stability prediction model, and performing prediction in combination with the spatial features and the monitoring time sequence data to obtain a three-dimensional strip block stability prediction value; correcting the stability prediction value of the three-dimensional strip block based on seismic data and rainfall data to obtain a corrected stability prediction value; and constructing a predicted value fusion model, and performing fusion calculation by using the predicted value fusion model according to the three-dimensional strip block stability predicted value and the corrected stability predicted value to obtain a real-time stability predicted value of the slope. The problem that slope stability calculation is not accurate and timely enough in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of slope stability calculation, and specifically to a real-time calculation method and system for slope stability. Background Art

[0002] Slope stability is crucial for engineering safety, life and property safety, and the ecological environment. However, there are many deficiencies in existing slope stability calculation methods.

[0003] Traditional two-dimensional analysis methods only consider the forces from a planar perspective, ignoring the three-dimensional characteristics of slopes. The actual slope geological structure is complex, including multiple layers of rock and soil masses, faults, joints, etc. The two-dimensional calculation results have large deviations and are difficult to meet the engineering requirements.

[0004] Although numerical simulation methods can simulate complex geological conditions, they consume a large amount of computing resources and take a long time, and are not suitable for large-scale slopes or real-time assessment scenarios. Moreover, it is difficult to obtain the parameters on which it depends, and the uncertainty of the parameters affects the accuracy of the results.

[0005] In addition, existing methods are difficult to comprehensively and accurately consider the influence of external environmental factors. Earthquakes and rainfall will change the mechanical properties and stress states of slopes. Rainfall reduces the shear strength of rock and soil masses, and the inertial forces generated by earthquakes increase additional loads. The combined action of the two may cause a significant decrease in slope stability, while traditional methods cannot accurately reflect these changes in real time.

[0006] With the development of engineering construction, the requirements for the accuracy, real-time performance, and comprehensiveness of slope stability assessment are getting higher and higher. Therefore, there is an urgent need for a calculation method for engineering construction under complex geological conditions that can comprehensively consider the three-dimensional geological structure of slopes, real-time monitoring data, and the dynamic influence of the external environment, achieve precise and dynamic monitoring and assessment, provide a reliable basis for engineering decision-making, and prevent slope instability disasters. Summary of the Invention

[0007] Aiming at the defects in the prior art, the present invention provides a real-time calculation method and system for slope stability, which solves the problem in the prior art that the stable calculation of slopes is inaccurate and untimely due to insufficient comprehensive consideration, enables the slope stability to be in real-time monitoring, and can greatly reduce slope geological disasters.

[0008] To achieve the above object, an aspect of the present invention provides a method for real-time calculation of slope stability, the method comprising: constructing a three-dimensional geological model of the slope, and dividing the three-dimensional geological model by using the three-dimensional slice method to obtain a plurality of three-dimensional slices; monitoring the three-dimensional slices, and based on the monitoring results and the three-dimensional geological model, extracting the spatial features and monitoring time series data of the three-dimensional slices; constructing a three-dimensional slice stability prediction model, and predicting by using the three-dimensional slice stability prediction model according to the spatial features and the monitoring time series data to obtain a three-dimensional slice stability prediction value; correcting the three-dimensional slice stability prediction value based on the earthquake data and rainfall data obtained in real time to obtain a corrected stability prediction value; constructing a prediction value fusion model, and performing fusion calculation by using the prediction value fusion model according to the three-dimensional slice stability prediction value and the corrected stability prediction value to obtain a real-time prediction value of the slope stability.

[0009] By constructing a three-dimensional geological model and performing three-dimensional slicing, the present invention can more meticulously and comprehensively reflect the geological structure characteristics of the slope, and is more conducive to subsequent calculation and analysis after modularization. Monitoring the three-dimensional slices and extracting spatial features and time series data provide a rich and accurate data basis for subsequent stability prediction. The three-dimensional slice stability prediction model constructed by using these data can make scientific predictions based on historical and real-time data, improving the reliability of the prediction. At the same time, correcting the prediction value based on earthquake and rainfall data fully considers the influence of external environmental factors on slope stability, further improving the accuracy of the prediction. By constructing a prediction value fusion model, the limitations of a single prediction model and correction model are avoided, and the prediction accuracy is improved again, realizing dynamic and accurate monitoring and evaluation of slope stability as a whole, providing strong support for the safety management and decision-making of slopes, helping to detect potential risks in advance, and reducing disaster losses.

[0010] Optionally, constructing the three-dimensional geological model of the slope includes: obtaining slope data of the slope, and decoupling the slope data into surface morphology data and internal structure data; constructing a digital elevation model by using the surface morphology data, and constructing a geological structural plane model by using the internal structure data; performing spatial coordinate alignment and geometric entity coupling on the digital elevation model and the geological structural plane model to form the three-dimensional geological model of the slope including external geometric features and internal mechanical interfaces.

[0011] The present invention decouples slope data into surface morphology data and internal structure data, which can deeply analyze the external morphology and internal structure of the slope respectively, making data processing more targeted. By using the surface morphology data to construct a digital elevation model and the internal structure data to construct a geological structural plane model, the external and internal characteristics of the slope are clearly presented. Through spatial coordinate alignment and geometric entity coupling of the two models, a three-dimensional geological model containing external geometric features and internal mechanical interfaces is generated, comprehensively and accurately reflecting the actual situation of the slope and improving the accuracy of the construction of the three-dimensional geological model.

[0012] Optionally, the construction of the three-dimensional block stability prediction model includes: obtaining the historical monitoring data of the three-dimensional block, and obtaining the historical spatial characteristics and historical monitoring time series data according to the historical monitoring data and the three-dimensional geological model; calculating the stability of the three-dimensional block according to the historical spatial characteristics and the historical monitoring time series data to obtain the historical stability; using the historical spatial characteristics, the historical monitoring time series data and the historical stability to construct sample data; constructing a framework model based on the spatio-temporal graph convolutional network, and training the framework model with the sample data to obtain a three-dimensional block stability prediction model.

[0013] In the present invention, by obtaining the historical monitoring data of the three-dimensional block and combining with the three-dimensional geological model to obtain the historical spatial characteristics and the monitoring time series data, the historical information related to the block stability is comprehensively and deeply excavated, providing rich data support for subsequent analysis. Calculating the historical stability based on these data provides a reliable reference standard for model training. Using the historical spatial characteristics, time series data and historical stability to construct sample data, covering multi-dimensional key information, improves the quality and representativeness of the samples. Constructing a framework model based on the spatio-temporal graph convolutional network can effectively capture the spatio-temporal relationships in the data. Training the framework model with the sample data enables the model to learn the internal laws between the block stability and related factors, thereby establishing an accurate three-dimensional block stability prediction model and improving the prediction effect of the three-dimensional block stability prediction model.

[0014] Optionally, the historical stability satisfies the following formula: , where, is the historical stability of the th three-dimensional block, is the self-weight of the th three-dimensional block, The historical dip angle of the structural plane of the th three-dimensional block, is the historical pore water pressure of the th three-dimensional block, is the The historical sliding surface area of the three-dimensional block is the historical internal friction angle of the th ree-dimensional block,

[0015] The historical stability formula of the three-dimensional block of the present invention comprehensively considers key factors such as the self-weight of the three-dimensional block, the dip angle of the structural plane, the pore water pressure, the sliding surface area, the internal friction angle and the cohesion, etc., improving the accuracy of the calculation of the historical stability of the three-dimensional block.

[0016] Optionally, correcting the predicted value of the stability of the three-dimensional block based on the real-time acquired seismic data and rainfall data to obtain a corrected predicted value of stability includes: calculating the influence of the earthquake on the vibration of the three-dimensional block in the horizontal direction according to the seismic data to obtain a horizontal seismic influence coefficient; calculating the vibration intensity of the earthquake on the three-dimensional block according to the horizontal seismic influence coefficient to obtain an earthquake inertial force; adjusting the cohesion of the three-dimensional block according to the rainfall data to obtain the cohesion of the three-dimensional block after rainfall weakening; using the earthquake inertial force and the cohesion of the three-dimensional block after rainfall weakening to correct the predicted value of the stability of the three-dimensional block to obtain a corrected predicted value of stability.

[0017] The present invention calculates the influence of the earthquake on the vibration of the three-dimensional block in the horizontal direction, obtains a horizontal seismic influence coefficient, and further calculates an earthquake inertial force, which can accurately quantify the interference of the earthquake action on the slope stability and fully consider the dynamic load factor of the earthquake. Adjusting the cohesion of the three-dimensional block according to the rainfall data takes into account that rainfall will weaken the mechanical properties of the soil mass, which is in line with the situation that rainfall causes the reduction of slope stability in actual engineering. Using the earthquake inertial force and the cohesion of the three-dimensional block after rainfall weakening to correct the predicted value comprehensively considers two key disaster-causing factors in the external environment, making the corrected predicted value of stability more in line with the real state of the slope under complex natural conditions, and improving the accuracy, reliability and generalization ability of slope stability assessment.

[0018] Optionally, using the earthquake inertial force and the cohesion of the three-dimensional block after rainfall weakening to correct the predicted value of the stability of the three-dimensional block to obtain a corrected predicted value of stability includes: using the earthquake inertial force to correct the sliding force of the three-dimensional block to obtain a corrected sliding force; using the earthquake inertial force and the cohesion of the three-dimensional block after rainfall weakening to correct the anti-sliding force of the three-dimensional block to obtain a corrected anti-sliding force; correcting the predicted value of the stability of the three-dimensional block according to the corrected sliding force and the corrected anti-sliding force to obtain a corrected predicted value of stability.

[0019] The present invention corrects the sliding force through seismic inertial force, fully considering the influence of the additional dynamic load on the slope blocks under seismic action, making the calculation of the sliding force more in line with the actual seismic conditions. The anti-sliding force is corrected by using the seismic inertial force and the cohesion weakened by rainfall, taking into account the double weakening effects of earthquake and rainfall on the anti-sliding ability of the slope, and can more accurately reflect the true ability of the slope to resist sliding in a complex environment. The adverse effects of the two key factors of earthquake and rainfall on the slope stability are comprehensively considered, effectively improving the accuracy of the stability prediction value.

[0020] Optionally, the corrected stability prediction value satisfies the following formula: , , , wherein, is the corrected stability prediction value of the th three-dimensional block, is the corrected sliding force of the th three-dimensional block, is the corrected anti-sliding force of the th three-dimensional block, is the self-weight of the th three-dimensional block, is the dip angle of the structural plane of the th three-dimensional block, is the seismic inertial force, is the th internal friction angle of the three-dimensional block, is the cohesion of the th three-dimensional block weakened by rainfall, is the sliding surface area of the th three-dimensional block.

[0021] Through rigorous mechanical principles, the present invention comprehensively considers key factors such as the self-weight of the three-dimensional block, the dip angle of the structural plane, the seismic inertial force, the internal friction angle, the cohesion, and the sliding surface area, constructs a scientific and complete slope stability quantification system, and at the same time fully considers the dynamic influence of external adverse factors, greatly improving the accuracy of the calculation of the corrected stability prediction value.

[0022] Optionally, the construction of the predicted value fusion model includes: performing random perturbations within a predetermined range on the internal friction angle, cohesion, and rock and soil unit weight of the three-dimensional block, and calculating the stability of the three-dimensional block according to the results of the random perturbations; calculating the standard deviation of the stability of the three-dimensional block to obtain the parameter perturbation error standard deviation; obtaining multiple measured slope data, and using the three-dimensional block stability prediction model to perform predictions according to the measured slope data to obtain the prediction results of the three-dimensional block; calculating the standard deviation of the prediction results of the three-dimensional block to obtain the prediction error standard deviation; constructing a predicted value fusion model according to the parameter perturbation error standard deviation and the prediction error standard deviation.

[0023] In the present invention, random perturbations are performed on key parameters such as the internal friction angle, cohesion, and rock and soil unit weight of the three-dimensional block, and the stability is calculated, which can simulate the uncertainty of the parameters. By calculating the standard deviation, the parameter perturbation error standard deviation is obtained, which quantifies the error caused by parameter fluctuations. Using the measured slope data in combination with the three-dimensional block stability prediction model to perform predictions and calculate the standard deviation of the prediction results, that is, the prediction error standard deviation, clarifies the error existing in the model prediction itself. Based on these two error standard deviations, a predicted value fusion model is constructed, fully considering the two factors of parameter uncertainty and model prediction error, and improving the calculation accuracy of the predicted value fusion model.

[0024] Optionally, the predicted value fusion model satisfies the following formula: , where, is the th final predicted value, is the parameter perturbation error standard deviation, is the predicted value of the stability prediction model of the th three-dimensional block, is the prediction error standard deviation, is the th corrected stability prediction value of the three-dimensional block.

[0025] The present invention comprehensively considers the parameter perturbation error standard deviation and the prediction error standard deviation, and can effectively fuse the predicted value of the stability prediction model and the corrected stability prediction value. This process fully weighs the parameter uncertainty and the model prediction error, making the finally obtained final predicted value more accurate and reliable, greatly improving the accuracy of slope stability prediction, and providing strong data support for the risk prevention and control and decision-making of slope engineering.

[0026] Another aspect of the present invention further provides a real-time slope stability calculation system, which includes a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, and the computer program includes program instructions. The processor is configured to call the program instructions to execute any one of the real-time slope stability calculation methods described in the previous aspect of the present invention.

[0027] A real-time slope stability calculation system of the present invention has a compact structure, stable performance, high integration, and simple composition. It can stably execute a real-time slope stability calculation method provided in the previous aspect of the present invention, further improving the overall applicability and practical application ability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flowchart of a real-time slope stability calculation method according to an embodiment of the present invention; Figure 2 It is a schematic structural diagram of a real-time slope stability calculation system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only for illustrative purposes and are not used to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that the present invention does not have to be practiced with these specific details. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the present invention.

[0030] Throughout the specification, references to "one embodiment", "an embodiment", "an example", or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Thus, the phrases "in one embodiment", "in an embodiment", "an example", or "an example" appearing throughout the specification do not necessarily all refer to the same embodiment or example. Additionally, the particular features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Further, those of ordinary skill in the art should understand that the diagrams provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0031] In an alternative embodiment, to solve the problem of inaccurate and untimely slope calculations in the prior art, please refer to Figure 1 , Figure 1 The method shown includes the following steps: Step S1, construct a three-dimensional geological model of the slope, and divide the three-dimensional geological model by using the three-dimensional slice method to obtain a plurality of three-dimensional slices.

[0032] Among them, the construction of the three-dimensional geological model of the slope specifically includes the following sub-steps: Step S101, obtain the slope data of the slope, and decouple the slope data into surface morphology data and internal structure data.

[0033] In this embodiment, an unmanned aerial vehicle (UAV) equipped with a high-resolution camera is used for aerial survey. The UAV is controlled to fly over the slope along a predetermined route, and the camera takes pictures at a set frequency. Through multi-view image matching technology, a large number of taken pictures are processed to construct a topographic contour model of the whole slope, accurately restoring the undulation of the slope surface. At the same time, with the help of terrestrial lidar scanning, in a non-contact measurement method, a laser beam is emitted to the slope surface and the reflected signal is received to quickly and accurately measure the distance and angle information of each point on the slope surface. Through filtering, classification and other processing of the massive point cloud data, the details of the surface morphology are further improved to obtain high-precision three-dimensional data of the slope surface.

[0034] For the acquisition of internal structure data, geological exploration means are mainly relied on. Through geological drilling, core samples are taken by drilling at different positions of the slope. In the laboratory, professional equipment such as polarizing microscopes and X-ray diffractometers are used to analyze the core samples to determine the lithology, stratigraphic layering, and to study in detail the development degree, occurrence and other characteristics of joint fissures. In addition, the seismic wave method is used. By artificially exciting seismic waves and receiving reflection and refraction signals at different positions, the structure of underground geological bodies is inverted based on information such as wave velocity and waveform. The electromagnetic method is used. Based on the electromagnetic property differences of underground geological bodies, the distribution and properties of underground geological bodies are detected, providing rich data support for comprehensively understanding the internal structure of the slope.

[0035] According to the slope characteristic attributes reflected by the data, data with similar properties are classified. The information directly reflecting the external physical form of the slope, such as the overall shape of the slope, the inclination slope of the slope surface, the elevation of the slope top and bottom, etc. are classified as surface morphology data. These data can intuitively present the external contour and topographic features of the slope and are an important basis for constructing the elevation digital model of the slope. And the data related to the internal material composition, geological structure, etc. of the slope, such as the specific type of rock and soil mass, the occurrence of geological structural planes, and the physical and mechanical parameters of the rock and soil mass, are classified as internal structure data. These data are the key basis for constructing the geological structural plane model and help to deeply analyze the mechanical properties and stability of the slope interior.

[0036] Step S102, construct an elevation digital model by using the surface morphology data, and construct a geological structural plane model by using the internal structure data.

[0037] In this embodiment, when constructing the elevation digital model using the surface morphology data, first, the acquired surface morphology data, such as the undulation of the slope surface, slope gradient, elevations of the slope top and bottom, etc., are preprocessed to remove the noise and outliers therein. Then, an interpolation algorithm in geographic information system (GIS) software is used to estimate the elevation values of unknown points based on the discrete topographic survey point data, and these discrete data are converted into continuous elevation data. By setting appropriate interpolation parameters, a high-precision digital elevation model (DEM) is generated, which can visually and accurately present the external topographic undulation of the slope.

[0038] For constructing the geological structural plane model using the internal structure data, first, the internal structure data are deeply analyzed to clarify the types of rock and soil masses, the attitudes and distribution laws of the geological structural planes. Then, professional 3D modeling software, such as 3DMine or GOCAD, etc., is used to accurately draw each geological structural plane according to the parameter information of the geological structural planes. During the modeling process, the spatial relationships and interactions between different geological structural planes are fully considered, and through operations such as Boolean operations, each geological structural plane is combined into a complete geological structural plane model, which can clearly display the geological structural characteristics inside the slope and provide important basic data for subsequent stability analysis.

[0039] Step S103, perform spatial coordinate alignment and geometric entity coupling on the elevation digital model and the geological structural plane model to form a three-dimensional geological model of the slope including external geometric features and internal mechanical interfaces.

[0040] In this embodiment, since the data sources and acquisition methods of these two models are different, their original spatial coordinate systems are not unified. First, by checking the current coordinate systems of the two models, the "Project and Transform" tool in the "Data Management Tools" of ArcGIS software is used to unify the coordinate systems of the elevation digital model and the geological structural plane model. The coordinate system of the geological structural plane model is CGCS2000. The "Project Raster" tool can be used to convert the elevation digital model, and the "Project" tool can be used to convert the geological structural plane model to convert them to the same target coordinate system.

[0041] After completing the coordinate alignment, geometric entity coupling is performed. The geological structural plane model is imported into ArcScene, and using the "Align Shape to Terrain" tool, select an appropriate alignment function, such as "Project All", to project each structural plane in the geological structural plane model onto the terrain represented by the elevation digital model, so that the geological structural plane and the elevation digital model are combined with each other to form a complete three-dimensional geological model including the external geometric features and internal mechanical interfaces of the slope.

[0042] In this embodiment, when using the three-dimensional slice method to divide the constructed three-dimensional geological model, first, a reasonable slicing idea needs to be determined according to the geological conditions and stress characteristics of the slope. Considering factors such as the distribution of rock and soil layers within the slope, the strike and dip of geological structural planes (such as faults, joints, fissures, etc.), and the potential sliding direction, etc., plan the slicing direction and approximate spacing.

[0043] Execute the specific division operation with the help of the professional geotechnical engineering analysis software FLAC3D. After importing the three-dimensional geological model into the software, use its built-in mesh generation or slicing function module to cut the model along the set direction according to the pre-determined slicing idea. During the cutting process, the software will automatically generate a series of three-dimensional slices in the shape of cuboids.

[0044] During the division process, it should be noted to ensure that the boundaries of each three-dimensional slice coincide or are parallel to the geological structural plane as much as possible to accurately reflect the mechanical properties inside the slope. At the same time, control the size and quantity of the slices to avoid insufficient analysis accuracy due to overly large slices or a sharp increase in the calculation amount due to overly small slices. After the division is completed, number and assign attributes to the obtained multiple three-dimensional slices, record information such as the spatial position, geometric dimensions, and lithological parameters of each slice, and prepare for the subsequent independent mechanical analysis and stability calculation of each three-dimensional slice, so as to more accurately evaluate the overall stability of the slope.

[0045] Step S2: Monitor the three-dimensional slices, and based on the monitoring results and the three-dimensional geological model, extract the spatial characteristics and monitoring time series data of the three-dimensional slices.

[0046] In this embodiment, the monitoring results are the monitoring data. The monitoring results include pore water pressure data, displacement data, and moisture content data. Due to the need for time series, dedicated sensors need to be arranged near the sliding surface, at the interface of rock and soil layers, or on the groundwater seepage path on the corresponding slope of the three-dimensional slice.

[0047] Install vibrating wire pore water pressure sensors. The sensor probe needs to be in close contact with the rock and soil mass inside the slice, connected to an automated data acquisition instrument through a data cable, and the pressure value is collected in real time at a frequency of once every 10 minutes. The ambient temperature is synchronously recorded for data correction to obtain the pore water pressure data.

[0048] Arrange high-precision GNSS receivers on the surface of the slice and on both sides of the internal structural plane to monitor the displacement of the corresponding slope of the three-dimensional slice in three-dimensional space in real time, supplemented by periodic calibration with a total station, and focus on monitoring the relative displacement between slices to obtain the displacement data.

[0049] TDR time domain reflectometer sensors are used and buried in layers at intervals of 0.5m along the depth direction of the strips. The moisture content is converted by measuring the soil dielectric constant. The data is collected every 30 minutes to avoid interference from instantaneous factors such as rainfall, and the moisture content data is obtained.

[0050] After the monitoring data is collected, it is necessary to pre-process the monitoring data. First, the 3σ rule is used to identify and eliminate the sudden noise of the monitoring data, such as the instantaneous fault data of the sensor. Then, based on the coordinate system of the three-dimensional geological model, the GNSS displacement data is converted into the relative displacement in the local coordinate system of the strip, eliminating the influence of the rigid body displacement of the overall slope and realizing time-space calibration. Finally, the pore water pressure, displacement, and water content data are aligned according to a unified timestamp accurate to seconds to form monitoring time series data indexed by the monitoring time.

[0051] The spatial features are extracted using the three-dimensional geological model. The spatial features include the surface inclination angle, sliding surface area, width, length and average thickness of the three-dimensional strips.

[0052] Each 3D strip is accurately extracted using the spatial analysis and measurement tools of ArcGIS, a professional geological modeling software. The built-in geological structural surface attribute query function of the model is used to directly obtain the structural surface inclination of each 3D strip, and the surface area calculation tool is used to calculate the area of ​​each 3D strip sliding surface. With the help of the 3D geometric measurement function, the width, length and average thickness of each 3D strip are measured. It should be noted that the strip number must be strictly compared during extraction to ensure that each spatial feature parameter is accurately associated with the corresponding 3D strip, and data verification is performed to avoid errors or mismatches.

[0053] Step S3, constructing a three-dimensional bar stability prediction model, and using the three-dimensional bar stability prediction model to perform predictions based on the spatial characteristics and the monitoring time series data to obtain a three-dimensional bar stability prediction value.

[0054] The construction of the three-dimensional bar stability prediction model specifically includes the following sub-steps: Step S301, obtaining historical monitoring data of the three-dimensional strip, and obtaining historical spatial characteristics and historical monitoring time series data based on the historical monitoring data and the three-dimensional geological model.

[0055] In this embodiment, the method of obtaining historical monitoring data of three-dimensional strips and obtaining historical spatial characteristics and historical monitoring time series data based on the historical monitoring data and the three-dimensional geological model is the same as the above-mentioned method of obtaining monitoring data and obtaining spatial characteristics and monitoring time series data of three-dimensional strips, and will not be repeated here.

[0056] Step S302: Calculate the stability of the three-dimensional strip based on the historical spatial features and the historical monitoring time series data to obtain the historical stability.

[0057] The historical stability satisfies the following formula: , where is the historical stability of the th three-dimensional block, is the self-weight of the th three-dimensional block, the dip angle of the historical structural plane of the th three-dimensional strip, is the historical pore water pressure of the th three-dimensional strip, is the historical sliding surface area of the th three-dimensional strip, is the historical internal friction angle of the th three-dimensional strip, is the cohesion of the th three-dimensional strip.

[0058] Step S303: Construct sample data by using the historical spatial features, the historical monitoring time series data, and the historical stability.

[0059] In this embodiment, the calculated historical stability value is accurately associated and matched with the spatial features and monitoring data at the corresponding time. Taking the time point as the sample unit, the spatial features, monitoring time series data, and historical stability value of the three-dimensional strip at each moment are integrated to construct a standardized structured sample data (such as a two-dimensional table), ensuring that each data item in each row of the sample accurately corresponds to the same moment of the same three-dimensional strip, guaranteeing the integrity, accuracy, and consistency of the data, providing a high-quality and unbiased data set for subsequent model training or analysis, and effectively supporting the construction and verification of the stability prediction model.

[0060] Step S304: Build a framework model based on the spatio-temporal graph convolutional network and train the framework model by using the sample data to obtain a three-dimensional strip stability prediction model.

[0061] In this embodiment, three-dimensional blocks are first abstracted as nodes in a graph structure, and an adjacency matrix of an undirected graph is constructed based on the spatial adjacency relationships between the blocks (such as adjacent, common structural planes, etc.). The node features fuse historical spatial features (structural plane dip angle, slip surface area, geometric dimensions, etc.) and historical monitoring time series data (temporal variations of pore water pressure, displacement, water content). A network architecture stacked with multi-layer spatio-temporal convolution modules is adopted: in the spatial dimension, the stability associations between adjacent blocks are captured through graph convolutional layers, and the spatial dependencies between nodes are learned using Laplacian matrix decomposition or attention mechanisms; in the time dimension, the dynamic evolution laws of single-node features over time are extracted through one-dimensional convolutional layers or temporal gating mechanisms.

[0062] In the model training stage, the sample data is divided into a training set, a validation set, and a test set according to the time series. The input layer encodes the spatio-temporal features of each node into a multi-dimensional tensor, with the dimension being the number of nodes multiplied by the time step and then multiplied by the feature dimension. After being alternately processed by the spatio-temporal convolutional layers, the stability prediction values are output through a fully connected layer. The mean square error is used as the loss function, and the AdamW optimizer is selected to balance the learning rate and weight decay. An early stopping mechanism is introduced during the training process to avoid overfitting, and at the same time, hyperparameters such as the number of spatial convolution kernels and the length of the time window are tuned through grid search. Finally, through end-to-end training, the model can effectively fuse the spatial correlation and time dynamics of three-dimensional blocks to generate a high-precision stability prediction model.

[0063] When using the three-dimensional block stability prediction model for prediction, first, the spatial features of the current three-dimensional block to be predicted and the real-time monitoring time series data are preprocessed. The spatial features are standardized and scaled, and the monitoring time series data is sliced according to the length of the time window during model training to form a feature sequence containing multiple time steps.

[0064] Each three-dimensional block is abstracted as a node in the graph structure, and a real-time adjacency matrix is constructed based on the spatial adjacency relationships between the blocks. It is combined with the preprocessed node features (spatial features + monitoring time series features) to form a multi-dimensional input tensor. This tensor is input into the trained three-dimensional block stability prediction model. The three-dimensional block stability prediction model captures the stability associations between adjacent blocks through spatial graph convolutional layers, extracts the dynamic evolution laws of single-block features using time convolutional layers, and after multi-layer spatio-temporal feature fusion, the stability prediction values of each three-dimensional block are output by the fully connected layer. The prediction process strictly follows the input specifications during model training to ensure the spatio-temporal consistency of the spatial adjacency relationships and real-time monitoring data, and finally, a quantitative prediction value reflecting the stability of the three-dimensional blocks under the current working conditions is obtained.

[0065] Step S4, based on the real-time acquired seismic data and rainfall data, correct the three-dimensional block stability prediction value to obtain a corrected stability prediction value.

[0066] Among them, correcting the predicted value of the three-dimensional block stability based on the real-time obtained seismic data and rainfall data to obtain the corrected predicted stability value specifically includes the following sub-steps: Step S401: Calculate the influence of the earthquake on the horizontal vibration of the three-dimensional block according to the seismic data to obtain the horizontal seismic influence coefficient.

[0067] In this embodiment, the seismic data mainly refers to the peak acceleration of ground motion, which can be obtained through the local seismic monitoring network. The peak acceleration of ground motion is based on the real-time monitoring and recording of seismic waves by seismic monitoring instruments and is usually updated and released at regular time intervals.

[0068] The horizontal seismic influence coefficient satisfies the following formula: , where, is the horizontal seismic influence coefficient, is the peak acceleration of ground motion, is the acceleration due to gravity.

[0069] Step S402: Calculate the vibration intensity of the earthquake on the three-dimensional block according to the horizontal seismic influence coefficient to obtain the seismic inertia force.

[0070] The seismic inertia force satisfies the following formula: , where, is the seismic inertia force, is the horizontal seismic influence coefficient, is the self-weight of the

[0071] The self-weight of the th three-dimensional block satisfies the following formula: where, is the self-weight of the th three-dimensional block, is the unit volume weight of the th three-dimensional block, is the average thickness of the three-dimensional block, is the width of the three-dimensional block,

[0072] In this embodiment, the seismic inertial force refers to the force exerted on a slope due to its own inertia when the ground moves under the action of seismic waves during an earthquake. When seismic waves propagate to the ground and cause ground vibrations, a stationary slope, due to its inertia, will attempt to maintain its original state of motion, while the movement of the ground forces the three-dimensional blocks in the slope to change their state, and the resulting force is the seismic inertial force. Its magnitude is related to the mass of the three-dimensional blocks and the seismic acceleration. The greater the mass and the seismic acceleration, the greater the seismic inertial force.

[0073] Step S403: Adjust the cohesion of the three-dimensional blocks according to the rainfall data to obtain the cohesion of the three-dimensional blocks after rainfall weakening.

[0074] The cohesion after rainfall weakening satisfies the following formula: , where, is the cohesion of the th three-dimensional block after rainfall weakening, is the cohesion of the th three-dimensional block, is the 24-hour cumulative rainfall.

[0075] In this embodiment, is the 24-hour cumulative rainfall, which can be obtained through the rainfall monitoring equipment of the local meteorological station. These devices will record the rainfall in real time and perform statistics at hourly or shorter time intervals to finally obtain the 24-hour cumulative rainfall value.

[0076] The cohesion after rainfall weakening means that there is a certain cohesion between the particles of the rock and soil mass. Cohesion can be understood as a force that resists mutual separation within the rock and soil mass. When rainfall occurs, water seeps into the rock and soil mass, which will weaken this cohesion.

[0077] Step S404: Use the seismic inertial force and the cohesion after rainfall weakening to correct the stability prediction value of the three-dimensional blocks to obtain a corrected stability prediction value.

[0078] Among them, using the seismic inertial force and the cohesion after rainfall weakening to correct the stability prediction value of the three-dimensional blocks to obtain a corrected stability prediction value specifically includes the following sub-steps: Step S40401: Use the seismic inertial force to correct the sliding force of the three-dimensional blocks to obtain a corrected sliding force.

[0079] In this embodiment, under normal circumstances, the downward sliding force of the three-dimensional block is mainly determined by the component force of its own gravity along the structural plane, which is based on the tendency of the block to slide along the structural plane under the action of gravity. However, when an earthquake occurs, the seismic inertial force will change the stress state of the block. The horizontal component of the seismic inertial force will be superimposed or offset partially with the component force of gravity along the structural plane, having an additional impact on the downward sliding tendency of the block.

[0080] The modified downward sliding force satisfies the following formula: , where, is the modified downward sliding force, The self-weight of the th three-dimensional block, is the dip angle of the structural plane of the

[0081] th

[0082] three-dimensional block, and

[0083] is the seismic inertial force. , The modified anti-sliding force satisfies the following formula: The self-weight of the th three-dimensional block, is the dip angle of the structural plane of the th three-dimensional block, The is the seismic inertial force, The is the internal friction angle of the

[0084] th The sliding surface area of a three-dimensional block satisfies the following formula: , where is the sliding surface area of the -th three-dimensional block, is the width of the three-dimensional block, is the length of the three-dimensional block, is the dip angle of the structural plane of the -th three-dimensional block.

[0085] Step S40403: Correct the predicted value of the stability of the three-dimensional block according to the corrected downslide force and the corrected anti-slide force to obtain a corrected predicted value of stability.

[0086] The corrected predicted value of stability satisfies the following formula: , , , where is the corrected predicted value of stability of the -th three-dimensional block, is the corrected downslide force of the -th three-dimensional block, is the corrected anti-slide force of the -th three-dimensional block, is the self-weight of the -th three-dimensional block, is the dip angle of the structural plane of the -th three-dimensional block, is the seismic inertial force, is the internal friction angle of the -th three-dimensional block, is the cohesion after rainfall weakening of the -th three-dimensional block, is the sliding surface area of the -th three-dimensional block.

[0087] Step S5: Construct a predicted value fusion model, and perform fusion calculation on the predicted value of the stability of the three-dimensional block and the corrected predicted value of stability by using the predicted value fusion model to obtain a real-time predicted value of the stability of the slope.

[0088] Among them, constructing the predicted value fusion model specifically includes the following sub-steps: Step S501: Perform random perturbations within a predetermined range on the internal friction angle, cohesion and rock and soil unit weight of the three-dimensional block, and calculate the stability of the three-dimensional block according to the results of the random perturbations.

[0089] In this embodiment, randomly disturbing the internal friction angle, cohesion, and unit weight of the soil-rock mass of the three-dimensional strip within a predetermined range means randomly generating different parameter values within the predetermined range. According to past engineering experience, geological exploration data, and relevant specifications, the reasonable fluctuation ranges of these three parameters are determined. The internal friction angle can fluctuate within ±5° of the measured value, the cohesion can vary within ±20% of the measured value, and the unit weight of the soil-rock mass can float within ±5 kN / m³ of the measured value. Then, using the random number generation method in Monte Carlo simulation, a series of different parameter combinations are randomly generated within the predetermined range, and then the stability of the three-dimensional block is calculated using these parameter combinations. Through no less than 1000 stability calculations, a large number of stability calculation results of the three-dimensional strip are obtained.

[0090] Step S502: Calculate the standard deviation of the stability of the three-dimensional strip to obtain the standard deviation of the parameter perturbation error.

[0091] In this embodiment, first, all the stability calculation values are summarized and sorted out, and the average value of these stability values is calculated. This step is to obtain a reference benchmark value to measure the deviation degree of each stability value from the average level. Then, the difference between each stability value and the average value is calculated in turn, and these differences are squared. The purpose of the squaring operation is to eliminate the positive and negative effects of the differences, so that all deviation situations are reflected in positive numbers, so as to perform subsequent statistical calculations more reasonably. Immediately afterwards, the squared differences are summed up and divided by the total number of stability values to obtain the variance. However, the numerical magnitude of the variance is quite different from the stability value itself, which is not conducive to intuitively reflecting the degree of data dispersion. Therefore, the square root operation is performed on the obtained variance, and the finally obtained value is the standard deviation of the parameter perturbation error.

[0092] This standard deviation reflects the degree of dispersion of the stability calculation results caused by random parameter perturbations. If the standard deviation is small, it means that these stability calculation results after random perturbations are relatively concentrated, and the uncertainty of the parameters has a relatively small impact on the stability. On the contrary, if the standard deviation is large, it means that the stability calculation results are more dispersed, and the uncertainty of the parameters has a more significant impact on the stability of the three-dimensional strip.

[0093] Step S503: Obtain multiple measured slope data, and use the three-dimensional strip stability prediction model to perform predictions according to the measured slope data to obtain the prediction results of the three-dimensional strip.

[0094] In this embodiment, the data acquisition is the same as above, and will not be elaborated again. The essence of the current embodiment is to verify the three-dimensional strip stability prediction model. Through no less than 1000 predictions, a large number of prediction results are obtained.

[0095] Step S504, calculate the standard deviation of the prediction results of the three-dimensional strip to obtain the standard deviation of the prediction error.

[0096] In this embodiment, the principle is the same as that of step S502 and will not be elaborated here.

[0097] Step S505, construct a prediction value fusion model based on the standard deviation of the parameter perturbation error and the standard deviation of the prediction error.

[0098] The prediction value fusion model satisfies the following formula: , where, is the th final prediction value, is the standard deviation of the parameter perturbation error, is the th prediction value of the stability prediction model of the three-dimensional strip, is the standard deviation of the prediction error, is the th corrected stability prediction value of the three-dimensional strip.

[0099] In this implementation, the construction of the prediction value fusion model is based on the principle of error weighted average. The standard deviation of the parameter perturbation error reflects the degree of dispersion of the stability calculation results caused by randomly perturbing parameters such as the internal friction angle, cohesion, and rock and soil unit weight of the three-dimensional strip. It reflects the influence of parameter uncertainty on the results. The standard deviation of the prediction error measures the error fluctuation of the stability prediction model itself during the prediction process.

[0100] By weighting and summing the prediction value of the stability prediction model and the corrected stability prediction value according to the squares of their respective error standard deviations, and then dividing by the sum of the squares of the two error standard deviations, the final prediction value is obtained. The purpose of doing this is that when the error standard deviation corresponding to a certain prediction value is small, it indicates that the reliability of this prediction value is relatively high, and a greater weight will be given to it in the final fusion result, so that the final prediction value is more inclined to the prediction result with high reliability, comprehensively considering two different sources of errors, and improving the accuracy and reliability of the prediction.

[0101] Finally, substitute the obtained stability prediction value and corrected stability prediction value of the three-dimensional strip into the stability prediction model of the three-dimensional strip to obtain the final prediction value.

[0102] It should be clear that when there is no influence of earthquake and heavy rainfall factors, the prediction value of the stability prediction model of the three-dimensional strip will not be corrected, then the The predicted value of the stability prediction model of a three-dimensional block is equal to the corrected stability predicted value of the th three-dimensional block, satisfying the formula: . Substituting this formula into the predicted value fusion model, the formula is obtained: . It shows that when there is no influence of earthquake and heavy rainfall factors, the final fusion value is the predicted value of the stability prediction model of the three-dimensional block.

[0103] Associate and bind the number of each three-dimensional block with the corresponding final predicted value. In the GIS software, set the color mapping rule according to the stability threshold. For example, set the area with a stability value greater than 1.2 to green, indicating a stable state; set the area with a stability value between 1.0 and 1.2 to yellow, representing a potentially unstable area; set the area with a stability value less than 1.0 to red, that is, a dangerous area. Through this way of color gradient, the spatial distribution of the stability value can be clearly shown on the three-dimensional slope model. At the same time, use the contour generation tool to generate a stability contour map based on the final predicted value. The density of the contour lines can further reflect the change gradient of the stability value, enabling users to more intuitively identify the areas with lower stability and the areas with larger stability changes, thus providing strong support for slope stability analysis and the formulation of protection measures.

[0104] As Figure 2 shown, on the other hand, the present invention also provides a slope stability real-time calculation system, including: a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, and the computer program includes program instructions. The processor is configured to call the program instructions to execute the relevant steps of the relevant embodiments in a slope stability real-time calculation method of the present invention.

[0105] For the slope stability real-time calculation system provided by the present invention, each functional component can be integrated in a processing component, or each component can exist physically alone, or two or more components can be integrated in one component. The above-mentioned integrated components can be implemented in the form of hardware or in the form of software functions, further enhancing the overall applicability and practical application ability of the present invention.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and the description of the present invention.

Claims

1. A real-time calculation method for slope stability, characterized in that: The method comprises: Constructing a three-dimensional geological model of the slope, and dividing the three-dimensional geological model using a three-dimensional strip method to obtain a plurality of three-dimensional strips; Monitoring the three-dimensional strips and extracting spatial features and monitoring time series data of the three-dimensional strips based on the monitoring results and the three-dimensional geological model; Constructing a three-dimensional bar stability prediction model, and using the three-dimensional bar stability prediction model to perform predictions based on the spatial characteristics and the monitoring time series data to obtain a three-dimensional bar stability prediction value; The three-dimensional block stability prediction value is corrected based on the seismic data and rainfall data acquired in real time to obtain a corrected stability prediction value; A prediction value fusion model is constructed, and a fusion calculation is performed using the prediction value fusion model according to the three-dimensional strip stability prediction value and the modified stability prediction value to obtain a real-time prediction value of the slope stability.

2. A real-time calculation method for slope stability according to claim 1, characterized in that: The step of constructing the three-dimensional geological model of the slope comprises: Acquiring slope data of the slope, and decoupling the slope data into surface morphology data and internal structure data; constructing an elevation digital model using the surface morphology data, and constructing a geological structure surface model using the internal structure data; The elevation digital model and the geological structure surface model are spatially aligned with each other and coupled with geometric entities to form a three-dimensional geological model of the slope including external geometric features and internal mechanical interfaces.

3. A real-time calculation method for slope stability according to claim 1, characterized in that: The construction of the three-dimensional strip stability prediction model comprises: Acquire historical monitoring data of the three-dimensional strip, and obtain historical spatial characteristics and historical monitoring time series data based on the historical monitoring data and the three-dimensional geological model; Calculating the stability of the three-dimensional strip according to the historical spatial characteristics and the historical monitoring time series data to obtain historical stability; constructing sample data using the historical spatial features, the historical monitoring time series data, and the historical stability; A framework model is constructed based on a spatiotemporal graph convolutional network, and the framework model is trained using the sample data to obtain a three-dimensional strip stability prediction model.

4. A real-time calculation method for slope stability according to claim 3, characterized in that: The historical stability satisfies the following formula: ， in, For the The historical stability of a three-dimensional block, For the The self-weight of a three-dimensional block, No. The historical structural surface inclination of the three-dimensional strip, For the The historical pore water pressure of a three-dimensional strip, For the The historical sliding surface area of ​​a three-dimensional strip, For the The historical internal friction angle of a three-dimensional strip, For the The cohesion of a three-dimensional strip.

5. A real-time calculation method for slope stability according to claim 1, characterized in that: The three-dimensional block stability prediction value is corrected based on the real-time acquired seismic data and rainfall data to obtain the corrected stability prediction value, which includes: According to the earthquake data, calculating the influence of the earthquake on the horizontal vibration of the three-dimensional strip, and obtaining the horizontal earthquake influence coefficient; Calculate the vibration intensity of the earthquake on the three-dimensional strip according to the horizontal earthquake influence coefficient to obtain the earthquake inertia force; Adjusting the cohesion of the three-dimensional strips according to the rainfall data to obtain the cohesion of the three-dimensional strips after weakening by rainfall; The three-dimensional block stability prediction value is corrected by using the earthquake inertial force and the cohesion weakened by rainfall to obtain a corrected stability prediction value.

6. A real-time calculation method for slope stability according to claim 5, characterized in that: The method of using the earthquake inertia force and the cohesion weakened by rainfall to correct the three-dimensional block stability prediction value to obtain the corrected stability prediction value includes: Using the seismic inertial force to correct the sliding force of the three-dimensional strip to obtain a corrected sliding force; Using the earthquake inertia force and the cohesion weakened by rainfall to correct the anti-sliding force of the three-dimensional strip, a corrected anti-sliding force is obtained; The three-dimensional strip stability prediction value is corrected according to the corrected sliding force and the corrected anti-sliding force to obtain a corrected stability prediction value.

7. A real-time slope stability calculation method according to claim 6, characterized in that: The modified stability prediction value satisfies the following formula: ， ， ， in, For the The modified stability prediction value of the three-dimensional strip, For the The modified sliding force of a three-dimensional strip, For the The modified anti-slip force of a three-dimensional strip, For the The dead weight of a three-dimensional bar, For the The inclination angle of the structural surface of the three-dimensional strip, is the earthquake inertia force, For the The internal friction angle of a three-dimensional strip, For the The cohesion of a three-dimensional strip weakened by rainfall, For the The sliding surface area of ​​a three-dimensional bar.

8. A real-time calculation method for slope stability according to claim 1, characterized in that: The construction of the prediction value fusion model includes: Performing random disturbances within a predetermined range on the internal friction angle, cohesion and weight of the three-dimensional strip, and calculating the stability of the three-dimensional strip according to the results of the random disturbances; Calculating the standard deviation of the stability of the three-dimensional strip to obtain the standard deviation of parameter disturbance error; Acquire a plurality of measured slope data, and perform prediction using the three-dimensional strip stability prediction model according to the measured slope data to obtain a prediction result of the three-dimensional strip; Calculating the standard deviation of the prediction results of the three-dimensional strip to obtain the prediction error standard deviation; A prediction value fusion model is constructed according to the parameter disturbance error standard deviation and the prediction error standard deviation.

9. A real-time slope stability calculation method according to claim 8, characterized in that: The predicted value fusion model satisfies the following formula: ， in, For the The final predicted value, is the standard deviation of parameter disturbance error, For the The predicted value of the stability prediction model of a three-dimensional strip, is the standard deviation of the prediction error, For the Modified stability prediction values ​​for three-dimensional bars.

10. A real-time slope stability calculation system, characterized in that: include: A processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute a real-time slope stability calculation method as described in any one of claims 1 to 9.

Images