Large-scale excavation high slope aging deformation mechanism analysis method under ultra-deep unloading condition

Abstract

The invention discloses an aging deformation mechanism analysis method for large-scale excavation of a high slope under an ultra-deep unloading condition, belongs to the technical field of reservoir bank slope deformation analysis, and obtains key parameters such as the volume, the sliding direction and the water level change of an accumulation body by integrating technical means such as geological exploration, hydrological monitoring and displacement measurement. By utilizing the parameters, a slope stability index, an energy function and a water-force coupling equation are constructed, and a set of complete analysis system is formed. According to the method, the accuracy and the real-time performance of slope stability evaluation are improved, and deep understanding of a slope deformation mechanism is realized through numerical simulation and an optimization algorithm. In addition, the scheme also provides prevention and control measures based on an evaluation result, and active prevention of risk management is realized. Through application of the scheme, the safety and management efficiency of the reservoir bank slope can be remarkably improved, and powerful support is provided for sustainable development of hydropower engineering.

Description

Technical Field

[0001] The invention belongs to the technical field of engineering geology, and in particular is a method for analyzing the time-dependent deformation mechanism of a large-scale excavated high slope under ultra-deep unloading conditions. Background Art

[0002] The time-dependent deformation of high slopes in large-scale excavation under ultra-deep unloading conditions refers to the deformation and stability problems of high slopes in large-scale excavation projects under the action of unloading at a large depth. In deep rock mass, the original ground stress state is changed due to excavation or other engineering activities, resulting in stress release and redistribution in the rock mass, which may cause deformation and damage to the rock mass. The excavation volume involved is large, and the slope height and excavation depth are large, which increases the risk of slope instability. Moreover, the height and slope of the slope are large, and there may be complex geological conditions, such as rock formation inclination, crack development, etc., which will affect the stability of the slope. In addition, the deformation and stability problems of the slope change over time and may gradually appear within a period of time after excavation, requiring long-term monitoring and evaluation.

[0003] Due to the complexity and uncertainty of geological conditions, it is very difficult to accurately predict the deformation and instability process of slopes. It is necessary to comprehensively consider multiple factors such as the physical and mechanical properties of the rock mass, groundwater, and external environment to conduct stability assessments, which increases the complexity of the analysis. Large-scale excavation and high slope monitoring require a large number of sensors and monitoring equipment, and the collection, transmission, and processing of monitoring data are all challenging. How to promptly identify and respond to the deformation and instability risks of slopes during construction, and formulate effective risk management measures, is an issue that requires comprehensive consideration.

[0004] Patent No.: CN115830805A points out that the existing technology has shortcomings in realizing the simulation of the entire disaster process, the publication of early warning information and the rehearsal planning of prevention and control decisions. There are still few studies on the construction of digital twins of slopes, and there is a lack of virtual-real mapping and dynamic updating characteristics during the actual excavation process of slope engineering.

[0005] On the other hand, the existing slope monitoring and early warning systems mainly rely on manual measurement and empirical judgment, and it is difficult to achieve real-time monitoring and early warning of the dynamic process of the slope. These systems have problems of low efficiency and insufficient precision in data collection, processing and analysis, and it is difficult to meet the needs of high-steep slope disaster risk prediction and prevention and control. In the process of building a digital twin model of the slope, the effective fusion of multi-source data is a key technical bottleneck. When processing multi-source data such as geological information, monitoring data and engineering parameters, the existing technology often lacks effective data fusion and information integration methods. This leads to low efficiency in the construction and update of the digital twin model of the slope, and it is difficult to achieve dynamic management and control of the entire life cycle of the slope project. For example, it is mentioned in the Chinese patent CN117830541A that the existing slope data acquisition methods and modeling technologies are changing with each passing day, but how to organically integrate the three-dimensional geological modeling of the slope with new technologies to build a digital twin model of the slope that can be dynamically updated according to the actual engineering parameters is a problem that needs to be solved by technicians in this field. Chinese patent CN118278175A emphasizes that the existing technology has a large amount of data in the digital twin scene of steep slopes, and the results of numerical simulation analysis are highly professional, which leads to low rendering efficiency and poor user cognition of steep slope information.

[0006] Therefore, the existing slope stability analysis methods still have major deficiencies in numerical simulation and risk assessment. When simulating the response of slopes under different working conditions, these methods often ignore the interaction between the slope and the surrounding environment, as well as the complexity of the internal structure of the slope. Therefore, the accuracy and reliability of the existing numerical simulation results need to be improved. Summary of the invention

[0007] The purpose of the present invention is to solve the problem that the existing technology is difficult to achieve real-time monitoring and early warning of the dynamic process of the slope, and proposes a method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions. Through technical means such as multi-source data fusion, real-time monitoring, dynamic updating and numerical simulation, the efficiency and accuracy of slope stability analysis are improved, and intelligent management and control of the entire life cycle of slope engineering are achieved. Through the technical solution of the present invention, the deficiencies in the existing technology can be effectively solved, providing a stronger guarantee for the safety and reliability of slope engineering.

[0008] In order to solve the above technical problems, the present invention provides a method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions, comprising the following steps:

[0009] S1. The acquired high slope geological data, slope surface model, and real-time monitored high slope deformation are used as monitoring data, and the monitoring data are transmitted to the data processing center in real time;

[0010] S2, generating a comprehensive geological-terrain model based on the combination of the slope surface model and the high slope geological data;

[0011] S3, performing slope stability analysis by numerical simulation software, inputting the comprehensive geological-topographic model into the numerical simulation software, analyzing the numerical simulation results, and identifying unstable areas and risk points;

[0012] S4, using a machine learning algorithm to establish a machine learning model, identify and classify rock masses, train the machine learning model through the monitoring data, construct and calculate the dangerous rock mass identification index, and obtain a machine learning prediction result;

[0013] S5. Import the high slope geological data, comprehensive geological-topographic model, numerical simulation results and machine learning prediction results into the digital twin platform, construct and calculate the slope morphology index, deformation rate index and deviation index between simulation and measured data, thereby calculating the comprehensive stability index, and setting the instability threshold in the digital twin platform;

[0014] S6. Input the construction parameters of slope excavation and support into the digital twin platform, perform simulation verification in the digital twin platform, and output the verification result after comparing with the instability threshold.

[0015] Preferably, S2 comprises:

[0016] S21. Use close-up photogrammetry technology and the slope surface model to build a three-dimensional model of the high slope, and generate point clouds and perform three-dimensional reconstruction using photogrammetry software;

[0017] S22, using spatial interpolation algorithm to process high slope geological data and point cloud data obtained by photogrammetry to generate a three-dimensional multi-layer digital elevation model;

[0018] S23. Combine the three-dimensional multi-layer digital elevation model with the high slope geological data to generate a comprehensive geological-topographic model.

[0019] Preferably, S4 includes:

[0020] S41, performing data cleaning, standardization and normalization processing based on the monitoring data;

[0021] S42. Based on geology, engineering and statistics, select features related to slope stability as input variables for machine learning models;

[0022] S43. Select a machine learning algorithm, use the preprocessed data to train the machine learning model, and use a cross-validation method to evaluate the predictive performance of the machine learning model;

[0023] S44. Construct and calculate the dangerous rock mass identification index in the machine learning model, use the trained machine learning model to identify and classify the rock mass, and obtain the prediction result of the dangerous rock mass.

[0024] Preferably, the step of constructing and calculating the dangerous rock mass identification index in the machine learning model comprises the following steps:

[0025] Sa. Determine the number of known dangerous rock masses in the slope through geological surveys and historical data collection;

[0026] Sb, set thresholds to distinguish between dangerous rock masses and stable rock masses, and use machine learning models to identify and classify rock masses to obtain stability prediction results for each rock mass;

[0027] Sc. Compare the stability prediction results with the results of field investigations, monitoring data or other independent verification methods to verify the stability prediction results.

[0028] Preferably, the dangerous rock mass identification index is calculated according to the following formula:

[0029]

[0030] Among them, TP represents true positives, which is the number of dangerous rock bodies correctly predicted by the model; FN represents false negatives, which is the number of dangerous rock bodies incorrectly predicted by the model as non-dangerous rock bodies.

[0031] Preferably, the slope morphology index, deformation rate index and deviation index between simulation and measured data are calculated by the following expressions respectively:

[0032] Slope morphology index calculation:

[0033]

[0034] Among them, H is the height of the slope and L is the length of the slope;

[0035] Deformation rate index calculation:

[0036]

[0037] Among them, Δd is the displacement change during the monitoring period, and Δt is the monitoring period;

[0038] Deviation index calculation between simulation and measured data:

[0039]

[0040] Among them, d sim is the simulated displacement, d meas is the measured displacement.

[0041] Preferably, the energy function Π is calculated by the following calculation formula:

[0042] Comprehensive stability index calculation:

[0043]

[0044] Among them, I max It is the product of the possible maximum values ​​of each part of the numerator in the calculation formula of the comprehensive stability index. acc It is expressed as the dangerous rock mass identification accuracy index.

[0045] Preferably, the S6 comprises:

[0046] The construction parameters of slope excavation and support are input into the digital twin platform. The numerical simulation and machine learning models in the digital twin platform are used to simulate and verify the construction process. By comparing with the calculated comprehensive stability index and instability threshold, if the simulation results show that the comprehensive stability index is lower than the instability threshold, the construction plan needs to be adjusted. The digital twin platform will provide optimization suggestions for the construction parameters, adjust the construction parameters, and re-simulate and verify until the simulation results show that the comprehensive stability index is greater than or equal to the instability threshold. If the simulation results show that the comprehensive stability index is greater than or equal to the instability threshold, the construction plan is within a safe range.

[0047] Beneficial effects of the present invention:

[0048] 1. This solution uses drone oblique photography and sensor array monitoring. This method can achieve efficient and accurate monitoring of slopes, detect small changes in slopes in a timely manner, and reduce the risks caused by untimely monitoring or inaccurate data. Combined with numerical simulation and machine learning prediction, this method can scientifically evaluate the stability and risk level of slopes, set reasonable warning thresholds, achieve early warning of slope disasters, and improve the safety management level of slope projects. The integration and functional realization of the digital twin platform make the decision support and governance plan formulation of slope projects more scientific and reasonable. By updating and analyzing data in real time, decision makers can adjust governance plans in a timely manner, optimize slope maintenance and construction, and ensure the safety and economy of the project.

[0049] 2. This solution integrates the collected geological data, monitoring data, numerical simulation results and machine learning prediction results into the digital twin platform, realizing the digital management and real-time monitoring of slope engineering. This integration not only optimizes the data management process, but also provides decision makers with intuitive and dynamic slope status display and risk assessment tools.

[0050] 3. This solution uses machine learning algorithms to analyze historical monitoring data and intelligently identify high-level dangerous rock masses. The application of this technology has significantly improved the intelligence level of slope monitoring. This method not only improves the accuracy of identification, but also enhances the ability to predict potential risks, providing a scientific basis for the safety management of slope projects.

[0051] 4. This solution combines geological drilling, drone oblique photography, sensor array monitoring, numerical simulation, and machine learning prediction to achieve comprehensive monitoring and analysis of high slopes. Through this multi-source data fusion, the geological characteristics and dynamic changes of the slope can be captured more accurately, providing rich data support for slope stability analysis.

[0052] 5. This scheme reflects the dependency between different indices through the multiplication operation in the comprehensive stability index formula, that is, the reduction of any factor will affect the overall stability assessment. The deviation index between simulation and measured data is used to penalize excessive deviations, that is, when the difference between simulation and measured data is large, the comprehensive stability index will decrease. By dividing by Imax, the formula ensures that the value of the comprehensive stability index is between 0 and 1, which helps to compare stability on different slopes or at different time points. Based on the comprehensive stability index and other risk assessment results, the decision support system of the digital twin platform is used to help formulate scientific maintenance and governance plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0054] Figure 1 It is a work flow chart of the present invention;

[0055] Figure 2 Schematic diagram of the surface and water level of the accumulation body in the embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0057] Example 1: Figure 1As shown in the figure, the time-dependent deformation mechanism analysis method of large-scale excavation high slope under ultra-deep unloading conditions includes the following steps:

[0058] S1, using geological drilling equipment and acoustic wave detection equipment to obtain high slope geological data, using drone oblique photography technology to obtain slope surface models, ensuring the acquisition of high-resolution surface images; using sensor arrays to monitor high slope deformation in real time, including displacement, stress and tilt, and transmit monitoring data to the data processing center in real time;

[0059] S2, using close-up photogrammetry technology combined with the surface image obtained by the drone to establish a three-dimensional model of the high slope; using a spatial interpolation algorithm to generate a three-dimensional multi-layer digital elevation model DEM, combining the DEM with the geological data of the high slope to generate a comprehensive geological-topographic model;

[0060] S3, using numerical simulation software to perform slope stability analysis, simulating the stress distribution and displacement field of the slope under different working conditions; inputting the detailed geological and topographic model of the slope obtained in S2 into the numerical simulation software; analyzing the numerical simulation results, and identifying possible unstable areas and risk points;

[0061] S4, using a machine learning algorithm to establish a machine learning model, identify and classify rock masses, use the monitoring data in S1 to train the machine learning model, and construct and calculate a dangerous rock mass identification index; using the numerical simulation results in S3 and the monitoring data in S1, continuously optimize the machine learning model so that the dangerous rock mass identification index approaches 1, and obtain a machine learning prediction result;

[0062] S5, build a digital twin platform. Select or develop a digital twin platform that can process and integrate multi-source data and support three-dimensional visualization, numerical simulation and machine learning models; import the geological data collected in step S1, the three-dimensional model established in step S2, the numerical simulation results in step S3 and the machine learning prediction results in step S4 into the digital twin platform in accordance with unified data formats and standards;

[0063] Ensure that the data in steps S1 to S4 can be synchronized to the digital twin platform in real time to achieve real-time data update and analysis. Use the three-dimensional visualization function of the digital twin platform to display the three-dimensional model of the slope, including geological structure, surface features and monitoring equipment layout. Display the monitoring data in step S1 in the form of charts, curves, etc. to intuitively reflect the real-time deformation of the slope. Integrate the numerical simulation results in step S3 into the platform to display the stress distribution, displacement field, etc. of the slope under different working conditions. Display the machine learning prediction results in step S4, such as dangerous rock mass identification and deformation trend prediction, in a visual way on the platform.

[0064] S6, based on the results of the numerical simulation of S3 and the machine learning algorithm of S4, the digital twin platform constructs and calculates the slope morphology index, the deformation rate index and the deviation index between the simulation and the measured data; the comprehensive stability index is calculated according to the dangerous rock mass identification index, the slope morphology index, the deformation rate index and the deviation index between the simulation and the measured data, and an instability threshold is set in the digital twin platform;

[0065] S7, inputting the construction parameters of slope excavation and support into the digital twin platform, performing simulation verification in the digital twin platform, outputting the verification results after comparing with the instability threshold, and determining whether to modify the construction parameters.

[0066] On the basis of the above, the present invention also proposes a system for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions, including hardware and software components for executing the above method. The system can realize real-time collection, processing, analysis and risk assessment of high slope geological data, as well as simulation verification of construction parameters to optimize slope excavation and support schemes.

[0067] Among them, the hardware components include geological drilling equipment, drone oblique photography system, sensor array and one or more data communication modules, which are used to transmit data to the data processing center in real time.

[0068] Software components include data preprocessing software, 3D modeling software, numerical simulation software, machine learning algorithm library and digital twin platform management software, which are used to perform data fusion, model building, stability analysis and risk assessment.

[0069] Among them, the digital twin platform management software has data visualization function, which can display the three-dimensional geological-topographic model of the slope, monitoring data, numerical simulation results and machine learning prediction results, as well as construction parameter simulation verification results.

[0070] The system also includes a decision support module for providing decision suggestions for slope excavation and support, as well as construction parameter adjustment plans based on the comprehensive stability index and the instability threshold, so as to achieve dynamic optimization of the slope engineering.

[0071] The system also includes an early warning module, which is used to automatically trigger an early warning mechanism when the monitoring data or simulation results exceed a preset threshold, and promptly notify relevant personnel to take corresponding measures to prevent slope instability.

[0072] The system also includes a user interaction interface that allows the user to input specific parameters, query analysis results, adjust simulation parameters, and interact with the system to achieve personalized slope stability analysis and management.

[0073] Example 2: Figure 1As shown in the figure, the time-dependent deformation mechanism analysis method of large-scale excavation high slope under ultra-deep unloading conditions includes the following steps:

[0074] S1, using geological drilling equipment and acoustic wave detection equipment to obtain high slope geological data, using drone oblique photography technology to obtain slope surface models, ensuring the acquisition of high-resolution surface images; using sensor arrays to monitor high slope deformation in real time, including displacement, stress and tilt, and transmit monitoring data to the data processing center in real time;

[0075] In step S1, the first thing to do is to collect geological data. Use geological drilling equipment and acoustic wave detection devices to go deep into the rock and soil to obtain detailed geological information of the slope. This includes the thickness of the rock layer, the rock characteristics and the degree of development of the cracks. Geological drilling can provide the physical and mechanical properties of the underground rock and soil, and provide basic data for subsequent stability analysis. The acoustic wave detection device assists in determining the rock layer interface and crack distribution through the propagation characteristics of sound waves in different rock layers, thereby enhancing the accuracy of geological data.

[0076] Next, the oblique photography technology of drones was used to obtain high-resolution images of the slope surface. The drones were equipped with high-definition photography equipment to capture surface features from different angles and obtain detailed surface texture and morphological information. These images not only include information such as terrain undulations and vegetation distribution, but also involve subtle changes in the surface, providing a visual basis for building an accurate three-dimensional surface model. Due to its flexibility and efficiency, the oblique photography technology of drones plays a key role in the construction of surface models for complex terrains.

[0077] Finally, a sensor array is deployed to monitor the deformation of the high slope in real time. Sensor types include displacement sensors, stress sensors, and tilt sensors, which can accurately capture small changes in the slope. The monitoring data is transmitted to the data processing center in real time through a wireless transmission system, providing real-time data support for dynamic monitoring and early warning of slope stability. The layout of the sensor array needs to take into account the geometric characteristics and potential deformation characteristics of the slope to ensure the comprehensiveness and representativeness of the monitoring data.

[0078] Through the implementation of the above step S1, it is possible to ensure that comprehensive and accurate geological and surface data are obtained, laying a solid foundation for subsequent 3D modeling, numerical simulation and machine learning analysis. At the same time, the acquisition of real-time monitoring data provides key information for dynamic management and risk assessment of slopes. The monitoring data uploaded to the data processing center is used for subsequent steps.

[0079] S2, using close-up photogrammetry technology combined with the surface image obtained by the drone to establish a three-dimensional model of the high slope; using a spatial interpolation algorithm to generate a three-dimensional multi-layer digital elevation model DEM, combining the DEM with the geological data of the high slope to generate a comprehensive geological-topographic model;

[0080] In step S2, the three-dimensional model of the high slope is first established by using close-up photogrammetry technology combined with high-resolution surface images obtained by drones. This process involves processing the images taken by drones, generating point clouds and reconstructing three-dimensional images through photogrammetry software (such as Pix4Dmapper, Agisoft Metashape, etc.) to ensure the accuracy and details of the model. Close-up photogrammetry technology can provide high-density point cloud data and provide accurate geometric information for subsequent three-dimensional modeling.

[0081] Next, a spatial interpolation algorithm, such as Kriging interpolation or inverse distance weighted (IDW) interpolation, is used to process the geological data and the point cloud data obtained by photogrammetry to generate a three-dimensional multi-layer digital elevation model (DEM). This step ensures the accuracy and resolution of the DEM and provides a basis for the terrain analysis of the slope. The generation of the DEM is to grid the point cloud data to form the elevation information of a regular grid, thereby obtaining a continuous terrain surface.

[0082] Finally, the generated DEM is combined with geological data to form a detailed geological and topographic model of the slope. This step involves superimposing geological information (such as rock layer distribution, fracture development, etc.) with the topographic information of the DEM to generate a comprehensive geological-topographic model, such as Figure 2 Through this model, we can have a more comprehensive understanding of the geological structure and topographic characteristics of the slope, and provide detailed basic data for subsequent numerical simulation and stability analysis.

[0083] The following is described by an example.

[0084] Suppose a drone takes a series of images in a high slope area. First, these images need to be processed to generate a point cloud. First, image preprocessing, including denoising and correction, is performed to improve the accuracy of subsequent processing. Then feature matching is performed to identify and match common feature points in multiple images. Finally, the spatial coordinates of each point are calculated using the matched feature points through the principle of triangulation to generate a point cloud.

[0085] Using point cloud data, generate a DEM by following the steps below: Select Inverse Distance Weighted (IDW) as the interpolation algorithm. Determine the range of point cloud data to be considered during interpolation. For each grid cell, perform interpolation calculations based on the elevation values ​​of the surrounding point cloud data.

[0086] Using the IDW method, the calculation formula is:

[0087]

[0088] Among them, Z in is the interpolated elevation value, Zi is the elevation value in the point cloud data, w i is the weight, which is proportional to the inverse of the distance.

[0089] On the basis of the above, geological data (such as rock layer distribution) are further combined with DEM to generate a comprehensive geological-terrain model. In this step, the geological map is first digitized and converted into a layer that can be processed in GIS software. In the GIS software, the geological layer is superimposed with the DEM to generate a comprehensive geological-terrain model.

[0090] Analyze the overlay results and adjust the transparency or boundaries of the geological layer as needed to ensure the consistency of geological information and terrain information. Using the generated comprehensive geological-terrain model, the following calculations and analyses can be performed:

[0091] The slope of each cell can be calculated on the DEM as part of terrain analysis. It is also possible to generate terrain profiles along a specific direction to analyze terrain changes. It is also possible to identify geological faults, cracks and other structures and analyze their impact on slope stability.

[0092] Photogrammetry software that can be used includes Pix4Dmapper, Agisoft Metashape, etc., which are used for point cloud generation and 3D modeling. GIS software includes ArcGIS, QGIS, etc., which are used for spatial interpolation, data overlay and terrain analysis. Numerical analysis software includes MATLAB, Python, etc., which are used to perform more complex numerical calculations and data analysis.

[0093] S3, using numerical simulation software to perform slope stability analysis, simulating the stress distribution and displacement field of the slope under different working conditions; inputting the detailed geological and topographic model of the slope obtained in S2 into the numerical simulation software; analyzing the numerical simulation results, and identifying possible unstable areas and risk points;

[0094] In this step, first, the comprehensive geological-terrain model generated in step S2 is imported into the numerical simulation software to ensure the continuity of the geological layers and the accuracy of the contact relationship. Physical and mechanical parameters such as elastic modulus, Poisson's ratio, internal friction angle, cohesion, etc. are assigned to different rock and soil layers according to the geological data.

[0095] In this step, it is necessary to check whether the input parameters of the model are correct and ensure that all geological and topographic data have been accurately loaded into the numerical simulation software. The focus of the review includes the accuracy of material properties, boundary conditions and initial conditions. Next, the convergence check of the model is performed to ensure that the numerical solution is stable and meets the predetermined convergence criteria. In addition, it is necessary to check the rationality of the calculation results, such as the continuity of stress distribution and the smoothness of displacement field, to ensure that there are no abnormal peaks or unrealistic values.

[0096] Then set the boundary conditions. Set the boundary conditions of the slope model, including fixed boundaries, free boundaries, and possible loading conditions, such as groundwater level, rainfall infiltration, etc. Then set different working conditions, such as rainfall, earthquake, excavation, etc., to simulate the impact of these working conditions on the stability of the slope. Then run the numerical simulation software for calculation to obtain the stress distribution and displacement field of the slope under different working conditions.

[0097] Stress analysis focuses on the magnitude and direction of the principal stresses, especially the distribution of the maximum and minimum principal stresses. In a slope, areas of high stress may indicate potential shear or tension failure and require special attention. Displacement field analysis focuses on the displacement and displacement rate of a monitoring point or the entire slope. Excessive displacement or accelerated displacement changes may indicate that the slope is about to fail. By calculating safety factors, such as the Mohr-Coulomb criterion, the stability of the slope can be quantitatively evaluated.

[0098] Finally, the calculation results are analyzed to identify areas of stress concentration and excessive displacement, and to evaluate the stability of the slope. Based on the numerical simulation results, possible unstable areas and risk points are identified to provide a basis for subsequent risk assessment and management. Specifically, the visualization tools of the numerical simulation software are used to display these key indicators in the form of color coding or contour lines on the geological-topographic model to intuitively display potential risk areas. In addition, geological information such as rock formation inclination and degree of fracture development can be combined to further analyze the causes and possible development patterns of unstable areas.

[0099] S4, using a machine learning algorithm to establish a machine learning model, identify and classify rock masses, use the monitoring data in S1 to train the machine learning model, and construct and calculate a dangerous rock mass identification index; using the numerical simulation results in S3 and the monitoring data in S1, continuously optimize the machine learning model so that the dangerous rock mass identification index approaches 1, and obtain a machine learning prediction result;

[0100] In the early stage of implementing step S4, the monitoring data collected in S1 are first thoroughly preprocessed, including data cleaning, standardization and normalization, to eliminate noise and unify the data format. Subsequently, based on the knowledge of geology, engineering and statistics, features closely related to slope stability are selected, such as displacement rate, stress level, terrain slope, etc. These features will be used as input variables for the machine learning model.

[0101] Select a machine learning algorithm suitable for the classification task, such as random forest, gradient boosted decision tree, or deep neural network. Train the model using the preprocessed data. During the training process, use cross-validation to evaluate the model's predictive performance and adjust the model parameters to optimize the model's accuracy and generalization ability.

[0102] The construction and calculation of dangerous rock mass identification index in the machine learning model includes the following steps:

[0103] Through geological surveys and historical data collection, the number of known dangerous rock masses in the slope is determined. This includes detailed records of visible cracks, displacements, topographic changes, etc. in the rock mass. Dangerous rock masses that may exist but are not clearly manifested are counted on site to calculate the total number of dangerous rock masses that may exist in the slope. Machine learning models are used to identify and classify rock masses (such as shear cracks, compressive shear zones, high load zones, toppling deformations, etc.), and the model will give a stability prediction result for each rock mass. A threshold is set to distinguish between dangerous rock masses and stable rock masses.

[0104] Compare the model's predictions with results from field surveys, monitoring data, or other independent verification methods (e.g., geological radar, satellite remote sensing, etc.).

[0105] Dangerous rock masses are identified by determining true positives (TP) and false positives (FP):

[0106] True Positives (TP): The number of cases that the model predicts to be dangerous rock masses and actually are dangerous rock masses.

[0107] False Positives (FP): The number of cases where the model predicts a dangerous rock mass but is not actually a dangerous rock mass.

[0108] The dangerous rock mass identification index is constructed as follows:

[0109]

[0110] Among them, TP is the true positive (the number of dangerous rock bodies correctly predicted by the model), and FN is the false negative (the number of dangerous rock bodies incorrectly predicted by the model as non-dangerous rock bodies). The higher the index, the better the prediction accuracy of the model.

[0111] Then, the trained machine learning model is used to identify and classify the rock mass to obtain the prediction results of dangerous rock mass. Combined with the numerical simulation results in S3, the accuracy of the model prediction is analyzed to identify possible deviations and deficiencies of the model. Through continuous iteration, the dangerous rock mass identification index is close to 1, that is, the false negative is close to 0, thereby maximizing the prediction accuracy, and thus obtaining accurate prediction results through the machine learning model, that is, the number of dangerous rock masses accurately identified.

[0112] S5, build a digital twin platform. Select or develop a digital twin platform that can process and integrate multi-source data and support three-dimensional visualization, numerical simulation and machine learning models; import the geological data collected in step S1, the three-dimensional model established in step S2, the numerical simulation results in step S3 and the machine learning prediction results in step S4 into the digital twin platform in accordance with unified data formats and standards;

[0113] In order to build an efficient and reliable digital twin platform, this paper adopts the following technology selection and development environment construction strategies:

[0114] Adopt a microservices architecture and deploy services with containerization technology (such as Docker) to ensure the scalability and flexibility of the platform. Use message queues (such as Kafka or RabbitMQ) to process data streams and achieve real-time data collection and distribution. Deploy the digital twin platform using cloud service platforms (such as AWS, Azure, or Google Cloud) to take advantage of the elastic computing resources and storage capabilities of cloud services. Configure a continuous integration / continuous deployment (CI / CD) process and use Jenkins or GitLab CI / CD tools to automate code building, testing, and deployment.

[0115] Use data lake technology to store raw data, and use big data platforms (such as Apache Hadoop or Spark) for data processing and analysis. Use ETL (Extract, Transform, Load) tools to convert data into a unified format and store it in a data warehouse (such as PostgreSQL or MongoDB).

[0116] Integrate 3D visualization engines (such as Unity 3D or Unreal Engine) to achieve high-precision visualization of geological models and terrain. Use WebGL front-end technology to allow users to interact with 3D models through browsers, achieving cross-platform access.

[0117] Integrate numerical simulation software APIs (such as FLAC3D, ABAQUS) to automatically import and analyze numerical simulation results. Use machine learning frameworks (such as TensorFlow or PyTorch) to deploy trained models to achieve model prediction and risk assessment.

[0118] Implement role-based access control (RBAC) strategies to ensure data and model security. Integrate user authentication and authorization mechanisms such as OAuth2.0 and OpenID Connect, and use RESTful API interfaces to ensure loose coupling and easy integration between different services and components. As a preferred solution, you can also write documentation and development guidelines to ensure the maintainability and scalability of the platform.

[0119] After the digital twin platform is built, the data obtained in steps S1, S2, S3 and S4 are integrated according to a unified data format and standard. This includes geological data, 3D models, numerical simulation results and machine learning prediction results. Develop data conversion tools and interfaces to achieve seamless connection between different data sources. The data standardization process needs to ensure the consistency and accuracy of the data to facilitate effective management and use in the digital twin platform.

[0120] The 3D visualization function of the digital twin platform is used to display the geological structure, terrain features, and monitoring equipment layout of the high slope. The 3D geological-terrain model established in step S2 and the numerical simulation results in step S3 are integrated to provide users with an intuitive display of the slope status. In addition, the machine learning prediction results are displayed in a visual way on the platform, allowing users to understand the distribution of dangerous rock masses and risk assessment predicted by the model.

[0121] Integrate real-time monitoring functions into the digital twin platform to dynamically update the real-time data collected by the sensor array in step S1 to the platform.

[0122] As a preferred solution, we can also combine numerical simulation and machine learning prediction results to develop an intelligent decision support system to provide a scientific basis for slope stability analysis and risk management. The system can automatically evaluate the slope status based on monitoring data and prediction results, indicate possible risks, and recommend corresponding preventive or response measures.

[0123] S6, based on the results of the numerical simulation of S3 and the machine learning algorithm of S4, the digital twin platform constructs and calculates the slope morphology index, the deformation rate index and the deviation index between the simulation and the measured data; the comprehensive stability index is calculated according to the dangerous rock mass identification index, the slope morphology index, the deformation rate index and the deviation index between the simulation and the measured data, and an instability threshold is set in the digital twin platform;

[0124] In step S6, the numerical simulation results obtained in step S3 and the dangerous rock mass identification index obtained by the machine learning algorithm in step S4 are first integrated into the digital twin platform. The numerical simulation results provide key data such as stress and displacement of the slope under different working conditions, while the machine learning model provides an accuracy index for dangerous rock mass identification based on monitoring data.

[0125] Using the numerical simulation results and monitoring data, the following indexes are constructed and calculated according to the calculation formula provided previously:

[0126] Slope morphology index calculation:

[0127]

[0128] Among them, H is the height of the slope and L is the length of the slope, which are obtained through actual measurement.

[0129] Deformation rate index calculation:

[0130]

[0131] Among them, Δd is the displacement change during the monitoring period, and Δt is the monitoring period.

[0132] Deviation index calculation between simulation and measured data:

[0133]

[0134] Among them, d sim is the simulated displacement, d meas is the measured displacement.

[0135] Comprehensive stability index calculation:

[0136]

[0137] Among them, I max is the maximum value of the product of all possible impact factors.

[0138] The following is an analysis of the components in the formula:

[0139] Slope shape index I from : This index reflects the geometric characteristics of the slope, such as slope gradient and slope height, which directly affect the stability of the slope.

[0140] Deformation rate index I rate : This index indicates the deformation rate of the slope. Rapid deformation may indicate impending instability.

[0141] Deviation index I between simulation and measured data dev :This index measures the difference between numerical simulation results and field measured data. dev The smaller it is, the closer the simulation result is to the actual situation, which means the reliability of the simulation is higher.

[0142] Dangerous rock mass identification accuracy index I acc : This index is evaluated through a machine learning model and reflects the accuracy of the model in identifying dangerous rock masses. It is a quantitative indicator of the performance of the intelligent identification system.

[0143] Normalization factor I max : This factor is used to normalize the comprehensive stability index to ensure that its value is between 0 and 1. This involves the maximum value of the product of all possible influencing factors, that is, the product of the possible maximum values ​​of the numerator parts in the calculation formula of the comprehensive stability index, so that the stability index under different circumstances is comparable.

[0144] Comprehensive stability index Istab :The index is a quantitative assessment that integrates morphology, deformation rate, simulation accuracy and identification of dangerous rock masses. It provides a standardized stability assessment result through normalization.

[0145] The multiplication operation in the formula reflects the dependence between different indices, that is, the reduction of any factor will affect the overall stability assessment. dev It is used to penalize excessive deviations, that is, when the simulation and measured data differ greatly, the comprehensive stability index will decrease. max , the formula ensures that the value of the comprehensive stability index is between 0 and 1, which helps to compare the stability of different slopes or at different time points.

[0146] According to the comprehensive stability index and other risk assessment results, the decision support system of the digital twin platform is used to help formulate scientific maintenance and governance plans. For example, the threshold of the comprehensive stability index is set to 0.85 in the digital twin platform, and a prompt is given when the actual calculated value is less than the threshold.

[0147] S7, inputting the construction parameters of slope excavation and support into the digital twin platform, performing simulation verification in the digital twin platform, outputting the verification result after comparing it with the instability threshold, and determining whether to modify the construction parameters.

[0148] In step S7, the construction parameters of slope excavation and support are first input into the digital twin platform. These parameters may include excavation depth, excavation rate, support structure type, support time point, etc. The digital twin platform should have a user interface that allows engineers to easily input and adjust these parameters. The input parameters will be used to drive numerical simulation software and machine learning models to simulate the impact of the construction process on slope stability.

[0149] The construction process is simulated and verified using numerical simulation and machine learning models in the digital twin platform. The simulation results will show the impact of changes in construction parameters on slope stability, including changes in stress distribution, displacement field, and dangerous rock mass identification index. The safety of the construction plan is evaluated by comparing it with the comprehensive stability index and instability threshold calculated in step S6. If the simulation results show that the comprehensive stability index is lower than the instability threshold, it indicates that the construction plan may cause slope instability and needs to be adjusted.

[0150] Based on the results of the simulation verification, the digital twin platform will provide optimization suggestions for construction parameters. If the current construction parameters lead to reduced slope stability, the platform will recommend adjusting the construction plan, such as changing the excavation depth, adding support structures, or optimizing the construction sequence. Engineers can adjust the construction parameters based on the platform's recommendations and re-simulate verification until the optimal construction plan is found. This process can be iterated until the simulation results show that the construction plan is within the safe range, that is, the comprehensive stability index is higher than the instability threshold.

[0151] The above specific embodiments are preferred embodiments of the present invention, and are not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the present specific embodiments. All equivalent changes made in accordance with the shape, structure, and method of the present invention are within the protection scope of the present invention.

Claims

1. The time-dependent deformation mechanism analysis method of large-scale excavation of high slopes under ultra-deep unloading conditions is characterized by: The steps include: S1. The acquired high slope geological data, slope surface model, and real-time monitored high slope deformation are used as monitoring data, and the monitoring data are transmitted to the data processing center in real time; S2, generating a comprehensive geological-terrain model based on the combination of the slope surface model and the high slope geological data; S3, performing slope stability analysis by numerical simulation software, inputting the comprehensive geological-topographic model into the numerical simulation software, analyzing the numerical simulation results, and identifying unstable areas and risk points; S4, using a machine learning algorithm to establish a machine learning model, identify and classify rock masses, train the machine learning model through the monitoring data, construct and calculate the dangerous rock mass identification index, and obtain a machine learning prediction result; S5. Import the high slope geological data, comprehensive geological-topographic model, numerical simulation results and machine learning prediction results into the digital twin platform, construct and calculate the slope morphology index, deformation rate index and deviation index between simulation and measured data, thereby calculating the comprehensive stability index, and setting the instability threshold in the digital twin platform; S6. Input the construction parameters of slope excavation and support into the digital twin platform, perform simulation verification in the digital twin platform, and output the verification result after comparing with the instability threshold.

2. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 1 is characterized in that: The S2 includes: S21. Use close-up photogrammetry technology and the slope surface model to build a three-dimensional model of the high slope, and generate point clouds and perform three-dimensional reconstruction using photogrammetry software; S22, using spatial interpolation algorithm to process high slope geological data and point cloud data obtained by photogrammetry to generate a three-dimensional multi-layer digital elevation model; S23. Combine the three-dimensional multi-layer digital elevation model with the high slope geological data to generate a comprehensive geological-topographic model.

3. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 1 is characterized in that: The S4 includes: S41, performing data cleaning, standardization and normalization processing based on the monitoring data; S42. Based on geology, engineering and statistics, select features related to slope stability as input variables for machine learning models; S43. Select a machine learning algorithm, use the preprocessed data to train the machine learning model, and use a cross-validation method to evaluate the predictive performance of the machine learning model; S44. Construct and calculate the dangerous rock mass identification index in the machine learning model, use the trained machine learning model to identify and classify the rock mass, and obtain the prediction result of the dangerous rock mass.

4. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 3 is characterized in that: The method of constructing and calculating the dangerous rock mass identification index in the machine learning model includes the following steps: Sa. Determine the number of known dangerous rock masses in the slope through geological surveys and historical data collection; Sb, set thresholds to distinguish between dangerous rock masses and stable rock masses, and use machine learning models to identify and classify rock masses to obtain stability prediction results for each rock mass; Sc. Compare the stability prediction results with the results of field investigations, monitoring data or other independent verification methods to verify the stability prediction results.

5. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 4 is characterized in that: The dangerous rock mass identification index is calculated according to the following formula: Among them, TP represents true positives, which is the number of dangerous rock bodies correctly predicted by the model; FN represents false negatives, which is the number of dangerous rock bodies incorrectly predicted by the model as non-dangerous rock bodies.

6. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 1 is characterized in that: The slope morphology index, deformation rate index and deviation index between simulation and measured data are calculated by the following expressions: Slope morphology index calculation: Among them, H is the height of the slope and L is the length of the slope; Deformation rate index calculation: Among them, Δd is the displacement change during the monitoring period, and Δt is the monitoring period; Deviation index calculation between simulation and measured data: Among them, d sim is the simulated displacement, d meas is the measured displacement.

7. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 1 is characterized in that: The energy function Π is calculated by the following calculation formula: Comprehensive stability index calculation: Among them, I max It is the product of the possible maximum values ​​of each part of the numerator in the calculation formula of the comprehensive stability index. acc It is expressed as the dangerous rock mass identification accuracy index.

8. The method for analyzing the time-dependent deformation mechanism of large-scale excavation of high slopes under ultra-deep unloading conditions according to claim 1 is characterized in that: The S6 includes: The construction parameters of slope excavation and support are input into the digital twin platform. The numerical simulation and machine learning models in the digital twin platform are used to simulate and verify the construction process. By comparing with the calculated comprehensive stability index and instability threshold, if the simulation results show that the comprehensive stability index is lower than the instability threshold, the construction plan needs to be adjusted. The digital twin platform will provide optimization suggestions for the construction parameters, adjust the construction parameters, and re-simulate and verify until the simulation results show that the comprehensive stability index is greater than or equal to the instability threshold. If the simulation results show that the comprehensive stability index is greater than or equal to the instability threshold, the construction plan is within a safe range.

Images