Bank slope relaxation area identification method and system based on three-dimensional geological modeling

Abstract

The invention relates to the technical field of slope monitoring, and discloses a bank slope relaxation area identification method based on three-dimensional geologic modeling, which comprises the following steps: S1, constructing a bank slope area three-dimensional geologic model; a comprehensive and accurate three-dimensional geological background is provided for subsequent identification work, and a geological basis for loose region identification is enhanced; s2, dividing the bank slope area three-dimensional geologic model into a plurality of space unit bodies, and extracting a fusion feature vector of each unit body; s3, training a prediction model for slack region identification based on the fusion feature vector; the intelligent and automatic degree of slack area identification is improved, human intervention is reduced, and the identification efficiency is improved. And S4, based on the established relaxed region prediction model, identifying a relaxed region in the bank slope region. According to the method, the three-dimensional geologic model is constructed, the structure control consistency index and the traditional geologic features are fused, the machine learning model is adopted to carry out probabilistic recognition, and the spatial clustering method is combined, so that high-precision, intelligent and continuous recognition of the bank slope relaxation area is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of slope monitoring, and in particular to a method and system for identifying a slope relaxation zone based on three-dimensional geological modeling. Background Art

[0002] As transportation, water conservancy, electricity, and urban infrastructure development expand into mountainous areas and complex terrain, the problem of geological hazards on bank slopes is becoming increasingly prominent. This is particularly true in areas with steep bank slopes, deep valleys, and areas of intense tectonic activity. Complex geological structures and intensely disturbed stress environments can easily lead to destructive features such as localized rock relaxation, interpenetrating joints, and the development of fissures, which can trigger disasters such as landslides and collapses. Bank slope relaxation zones are precursors to intense disintegration, stress relaxation, and structural instability in response to external disturbances or stress adjustments. Identifying these zones is crucial for disaster warning and mitigation design.

[0003] In existing technologies, the identification of slack zones mainly relies on manual experience, single-point monitoring, or two-dimensional data analysis, which has certain limitations. For example:

[0004] Chinese patent CN106202908B discloses a method for identifying loose areas on high slopes. This method processes GNSS point displacement data, constructs a similarity index based on the geometric distance and directional similarity between displacement vectors, and combines it with a K-means clustering method to identify loose areas. While this method reduces reliance on subjective human judgment, its identification process relies on sparse monitoring points, resulting in poor spatial continuity and difficulty adapting to geological background information with complex coupling of multiple factors.

[0005] Chinese patent CN116773772A proposes a method for measuring and calculating surrounding rock relaxation zones. This method involves drilling stress detection holes in the target detection surface, injecting a coupling agent, and then monitoring stress disturbance changes in real time. The relaxation depth distribution is determined by analyzing the stress disturbance index (SDI). This approach enables continuous monitoring and, to a certain extent, reflects the rock failure process. However, it is highly dependent on testing hardware, is limited in scope to tunnels or surrounding rock scenarios, and cannot accurately identify large-scale slopes.

[0006] In summary, the existing methods often have the following defects or deficiencies:

[0007] Limited utilization of spatial information: Many methods are based on point deformation or local geophysical exploration, lacking 3D continuous spatial structure modeling and partition analysis;

[0008] Weak ability to identify the main controlling factors of the structure: most of them do not introduce directional information of the main controlling structural planes such as joints and faults, making it difficult to capture the deep disintegration mechanism;

[0009] Data fusion is insufficient: multi-source data such as terrain, geology, hydrology and monitoring are not fully fused, and the robustness and identification accuracy of the model are limited.

[0010] Therefore, there is an urgent need for a relaxation zone identification method under complex geological conditions, which can integrate structural direction, geological environment and monitoring data, and complete quantitative identification, partition clustering and visual expression of the bank slope relaxation zone in three-dimensional space, and improve the engineering practicability and safety decision support capability. SUMMARY

[0011] To solve the technical problems proposed in the background art, the present application provides a bank slope relaxation zone identification method and system based on three-dimensional geological modeling.

[0012] The present application adopts the following technical scheme: a bank slope relaxation zone identification method based on three-dimensional geological modeling, comprising:

[0013] S1: constructing a three-dimensional geological model of the bank slope area; providing a comprehensive and accurate three-dimensional geological background for subsequent identification work, and enhancing the geological basis for relaxation zone identification;

[0014] S2: dividing the three-dimensional geological model of the bank slope area into a plurality of spatial unit bodies, and extracting a fusion feature vector of each unit body; and

[0015] The fusion feature vector fuses the traditional features and structural control consistency indexes of each spatial unit body; the structural control consistency index is used to measure the degree of coincidence between the structure main control direction and the relaxation / sliding main direction.

[0016] Discrete modeling and digital expression of complex geological structures are realized, which facilitates model training and unified processing.

[0017] S3: training a prediction model for relaxation zone identification based on the fusion feature vector; improving the intelligence and automation level of relaxation zone identification, reducing human intervention and improving identification efficiency.

[0018] S4: identifying the relaxation zone in the bank slope area based on the established relaxation zone prediction model. The potential sliding or relaxation area can be quickly and accurately identified in actual engineering, and the disaster warning capability and prevention and control level are improved.

[0019] As a further improvement of the above scheme, the S1 comprises:

[0020] A basic geological database is constructed through comprehensive geological survey and data acquisition;

[0021] An initial geological structure model is constructed through professional three-dimensional modeling software;

[0022] A refined three-dimensional attribute model is constructed through interpolation modeling and parameter assignment;

[0023] Model validation and calibration are performed by monitoring data and numerical simulation;

[0024] The timeliness and accuracy of the model are maintained through a dynamic updating mechanism.

[0025] The present application makes the geological model more consistent with the changes on site through dynamic updating and model calibration, and improves the timeliness and reliability of identification.

[0026] As a further improvement of the above scheme, the traditional features include topographic geometric features, geomechanical features, structure control factors, hydrological-monitoring information; wherein

[0027] The topographic geometric features include: slope θ, slope direction α, elevation H, and relative distance to shore / slope top / slope foot;

[0028] The geomechanical features include: lithology category code, elastic modulus E, Poisson's ratio v, internal friction angle φ, cohesion c, and rock layer dip / dip direction;

[0029] The structure control factors include: the nearest distance to the fault d f , local joint density, and DFN fracture penetration degree index;

[0030] The hydrological-monitoring information includes: groundwater level depth, annual average rainfall, InSAR deformation rate v INSAR , GNSS point three-dimensional displacement trend, and whether it is a high shear strain identification condition, i.e. strain rate greater than threshold value.

[0031] The present application integrates multi-dimensional geological and environmental parameters to improve the richness of model input information and identification accuracy.

[0032] As a further improvement of the above scheme, in S2:

[0033] By dividing the constructed three-dimensional geological model into spatial unit bodies of uniform size as the basic objects for subsequent feature extraction and prediction analysis, each unit body has the following spatial attributes:

[0034] Coordinate position, belonging stratum identification, and whether it passes through structure surface such as joint / fault

[0035] From the constructed three-dimensional geological model, the information of joint surface, fault surface and rock layer surface is extracted to construct the spatial attributes of each type of structure surface, including dip direction, dip angle, joint spacing, direction density, etc., and a normalized direction vector is generated for each structure surface unit;

[0036]

[0037] The structure main control vector of each unit body is calculated through the direction vector of each unit body.

[0038]

[0039] wherein w i is the weight of the i-th structural plane (which can be determined by the joint plane spacing, length, and penetrability), n j is the number of structural planes passing through the voxel.

[0040] The present application realizes comprehensive extraction of spatial information of structural planes and quantitative expression of main control structures, and provides a basis for structural consistency modeling.

[0041] As a further improvement of the above-mentioned scheme, in the S2, the following is included:

[0042] By constructing an index consistent with the sliding direction, i.e., a structure control consistency index, the structural sensitivity evaluation is realized; the structure control consistency index C j satisfies: wherein, represents the main direction of gravity or the slope normal vector of the slope where the unit is located.

[0043] Fusion of structure control consistency index and traditional features enhances the input dimension of the prediction model;

[0044] X j =[H j ,θ j ,E j ,φ j ,c j ,v j ,C j ......]

[0045] Enhance the model's ability to identify the main controlling factors of the structure and improve the accuracy of the identification of the relaxation zone.

[0046] As a further improvement of the above-mentioned scheme, in the S3, the following is included:

[0047] By fusing the fusion feature vector of the structure control consistency index as a training sample, a gradient boosting tree classification model (GBT model) is trained; wherein:

[0048] The prediction function of the model structure wherein f t (x) represents the t-th regression tree;

[0049] A binary cross-entropy loss function is used as the objective function;

[0050]

[0051] wherein: p i =σ(F(x i )) is the relaxation probability, yi ∈{0,1}

[0052] Optimizing hyperparameters using GridSearch or BayesianOptimization;

[0053] Evaluate accuracy and generalization ability using 5-fold cross-validation;

[0054] Important indicators for model evaluation include AUC, Recall, and F1-score.

[0055] The invention can improve prediction performance, optimize recognition results, and ensure generalization ability and robustness in engineering applications.

[0056] As a further improvement of the above scheme, in S4

[0057] By deploying the model to the full slope voxel grid; output the relaxation probability for each unit,

[0058] By setting a recognition threshold, mark the units exceeding the threshold as "potential relaxation area";

[0059] By using DBSCAN or connectivity-based clustering method, adjacent high-probability units are classified into a continuous relaxation area.

[0060] Output the recognition results.

[0061] The invention combines probability judgment and spatial clustering, realizes the transformation from point prediction to surface relaxation area recognition, and is convenient for practical application and visualization.

[0062] By calculating the volume, depth, deformation direction and other indicators of each relaxation area, it is used for subsequent early warning evaluation.

[0063] By superimposing the relaxation probability in the form of a heat map in a three-dimensional geological modeling platform (such as ParaView, Leapfrog).

[0064] The invention also proposes a shore slope relaxation area recognition system based on three-dimensional geological modeling, comprising:

[0065] The geological modeling unit is used to receive geological exploration and data acquisition results, and to build a three-dimensional geological model of the shore slope area, which includes stratigraphic structure, faults, joints and other structural surface information;

[0066] The spatial unit processing unit is used to divide the three-dimensional geological model into spatial unit bodies of uniform size, and to extract the spatial position, structural surface crossing condition and direction vector information of each unit to form a structural main control vector;

[0067] a feature extraction unit configured to extract a fusion feature vector of each spatial unit, the fusion feature vector comprising a structure control consistency index and traditional geological features, the traditional geological features comprising topographic geometric features, geomechanical features, structure control factors, and hydrological-monitoring information;

[0068] a relaxed zone prediction model training unit configured to train a gradient boosting tree classification model based on the fusion feature vector, and to perform parameter optimization and accuracy evaluation on the model;

[0069] a relaxed zone identification unit configured to deploy the trained prediction model to a three-dimensional geological voxel grid, output a relaxed probability for each unit, and identify potential relaxed zones according to a set threshold value;

[0070] a relaxed zone clustering unit configured to classify adjacent high-probability units into continuous relaxed zones using a density-based spatial clustering of applications with noise (DBSCAN) or a connectivity-based algorithm, and to calculate the volume, depth, and deformation direction of each relaxed zone;

[0071] a visualization display unit configured to visualize the relaxed zone identification results in the form of a heat map superimposed on a three-dimensional geological model platform for decision-making and early warning evaluation.

[0072] The present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned shore slope relaxed zone identification method based on three-dimensional geological modeling when executing the computer program.

[0073] The present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above-mentioned shore slope relaxed zone identification method based on three-dimensional geological modeling.

[0074] Compared with the prior art, the present application has the following advantages:

[0075] The present application realizes high-precision, intelligent, and continuous identification of shore slope relaxed zones by constructing a three-dimensional geological model, fusing a structure control consistency index and traditional geological features, using a machine learning model for probabilistic identification, and combining a spatial clustering method. This method breaks through the limitations of traditional single-point monitoring and manual experience, significantly improves the identification accuracy, spatial expression ability, and automation level. At the same time, the system can output multi-dimensional parameters such as volume, depth, and deformation direction, and realize visualization display in a three-dimensional platform, enhancing the engineering practicability and early warning decision support capability of the identification results, and having good adaptability and promotional value. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 A flowchart of the shore slope relaxed zone identification method based on three-dimensional geological modeling according to the present application;

[0077] Figure 2 A structural diagram of a bank slope relaxation zone identification system based on three-dimensional geological modeling is provided for the present application. DETAILED DESCRIPTION

[0078] It should be understood that the specific embodiments described herein are merely illustrative of the present application and do not limit the present application.

[0079] The tunnel surrounding rock stability automatic monitoring and early warning method provided by the embodiments of the present application. The execution subject of the tunnel surrounding rock stability automatic monitoring and early warning method includes but is not limited to at least one of the electronic devices capable of being configured to execute the method provided by the embodiments of the present application, such as a server, a terminal and the like. In other words, the tunnel surrounding rock stability automatic monitoring and early warning method can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms, etc. Basic cloud computing services.

[0080] Embodiment 1:

[0081] Reference Figure 1 The bank slope relaxation zone identification method based on three-dimensional geological modeling provided by the present application includes:

[0082] S1: Construct a three-dimensional geological model of the bank slope area; provide a comprehensive and accurate three-dimensional geological background for subsequent identification work, and enhance the geological basis for relaxation zone identification;

[0083] S2: Divide the three-dimensional geological model of the bank slope area into a plurality of spatial unit bodies, and extract a fusion feature vector of each unit body; and

[0084] The fusion feature vector fuses the traditional features and structure control consistency indexes of each spatial unit body. The structure control consistency index is used to measure the degree of coincidence between the structure main control direction and the relaxation / sliding main direction.

[0085] Discrete modeling and feature digital expression of complex geological structures are realized, which facilitates model training and unified processing.

[0086] S3: Train a prediction model for relaxation zone identification based on the fusion feature vector; improve the intelligence and automation degree of relaxation zone identification, reduce human intervention and improve identification efficiency.

[0087] S4: Identify the relaxation zone in the slope area based on the established relaxation zone prediction model. The potential sliding or relaxation zone can be quickly and accurately identified in actual engineering, improving the disaster warning capability and prevention level.

[0088] As an optional embodiment of the present application, S1 comprises:

[0089] By comprehensive geological survey and data collection, a basic geological database is constructed;

[0090] By professional three-dimensional modeling software, an initial geological structure model is constructed;

[0091] By interpolation modeling and parameter assignment, a refined three-dimensional attribute model is constructed;

[0092] By monitoring data and numerical simulation, model verification and calibration are carried out;

[0093] Through dynamic updating mechanism, the timeliness and accuracy of the model are maintained.

[0094] Through dynamic updating and model calibration means, the geological model is more consistent with the site changes, improving the identification timeliness and reliability.

[0095] More specifically: By comprehensive geological survey and data collection (including engineering geological surveying, drilling, geophysical prospecting, in-situ testing, laboratory testing, remote sensing interpretation, etc.), a basic geological database of the study area is constructed, realizing comprehensive understanding and digital expression of the spatial distribution, geometric shape and physical and mechanical properties of key geological elements such as stratum lithology, geological structure (fault, joint, fold), weathering zoning, hydrogeological conditions, and adverse geological phenomena (such as landslides, collapses, and dangerous rock masses);

[0096] Further, by professional three-dimensional geological modeling software (such as GOCAD, MOVE, GeoStudio, FLAC3D built-in tools or professional GIS platform, etc.), an initial three-dimensional geological structure framework model based on S101 database is constructed; the spatial surface construction and connection of main geological interfaces (such as ground topography, stratum interface, fault / joint surface, weathering interface, groundwater level surface, etc.) are realized.

[0097] Further, by geological interpretation inference, spatial interpolation algorithm (such as Kriging method, inverse distance weighting, etc.) and physical and mechanical parameter assignment, a refined three-dimensional geological attribute model with geological rationality and engineering practicability is constructed; and:

[0098] Accurate rock and soil physical and mechanical parameters (such as density, elastic modulus, Poisson's ratio, cohesion, internal friction angle, permeability coefficient, etc.) are assigned to each geological unit (body / surface) in the model;

[0099] Make reasonable inferences and improve models for areas of geological structural uncertainty (such as stratum pinch-out and structurally complex areas);

[0100] Complete the model's geometric topology check and geological logic verification to ensure the model's internal consistency and eliminate errors such as intersection and overlap.

[0101] In general, the present invention provides a slope relaxation zone identification method that integrates three-dimensional modeling, structural indicators and intelligent prediction models, which significantly improves the identification accuracy, intelligence level and engineering applicability, specifically: proposing a spatial unit division and attribute modeling strategy based on three-dimensional geological modeling to achieve digital and high-resolution expression of complex geological structures; integrating traditional geological characteristics and structural control consistency indicators to enhance the model's sensitivity and expression ability to the main controlling factors of the structure; constructing a machine learning prediction model driven by fusion feature vectors to achieve intelligent and automatic identification of relaxation zones and reduce human intervention; supporting dynamic update and model calibration mechanisms to synchronize geological models with actual engineering environment changes, thereby improving the real-time and accuracy of identification; and being able to output planar continuous relaxation zone results with parameters such as volume, depth, and main deformation direction, which is convenient for engineering applications and disaster warnings.

[0102] As an optional embodiment of the present invention, the traditional features include terrain geometric features, geomechanical features, structural control factors, and hydrological-monitoring information; wherein

[0103] Terrain geometric features include: slope θ, slope aspect α, elevation H, and relative distance from the shoreline / slope top / slope foot;

[0104] Geomechanical characteristics include: lithology category code, elastic modulus E, Poisson's ratio v, internal friction angle φ, cohesion c, and rock layer dip / strike;

[0105] Structural control factors include: the closest distance to the fault d f , local joint density and DFN fracture penetration index;

[0106] Hydrological monitoring information includes: groundwater depth, annual average rainfall, InSAR deformation rate v INSAR , the three-dimensional displacement trend of the GNSS point, and whether it meets the high shear strain identification condition - that is, the strain rate is greater than the threshold.

[0107] Integrate multi-dimensional geological and environmental parameters to improve the richness of model input information and recognition accuracy.

[0108] In this embodiment, more details are given below:

[0109] The construction of traditional features in this scheme combines multi-source geological information, topographic parameters and monitoring data to enrich the input dimension of the model and improve the accuracy of disaster identification. Specifically, the following dimensions of parameter extraction and quantization processing methods are included:

[0110] Acquisition and construction of topographic geometric features

[0111] The topographic features of the target area are extracted through remote sensing DEM (Digital Elevation Model) data and field surveying data, mainly including:

[0112] Slope: Calculate the slope of each grid cell using the triangulation algorithm based on the elevation difference of DEM data;

[0113] Aspect: Calculate the aspect angle of each cell based on the surface slope direction and use the sine-cosine encoding method for quantization;

[0114] Elevation: Directly extract the elevation value of the DEM grid;

[0115] Relative distance to shoreline / slope top / slope foot: Calculate the Euclidean distance from the center point of each cell to the nearest shoreline, slope top line, and slope foot line using GIS spatial analysis tools as a relative spatial position parameter input.

[0116] It should be noted that the above data can be processed by geographic information system platforms such as ArcGIS, QGIS, etc. The resolution is recommended to be no less than 10m x 10m.

[0117] In this scheme, the construction method of geomechanical features is as follows.

[0118] The following features are obtained using field geotechnical engineering investigation and indoor geotechnical physical and mechanical test data:

[0119] Rock type coding: Based on engineering geological drilling results, various rock types such as sandstone, shale, slate, granite, etc. are assigned a category code (e.g. sandstone = 1, shale = 2);

[0120] Elastic modulus E, Poisson's ratio: obtained through triaxial shear test, compression test and other indoor test methods;

[0121] Internal friction angle, cohesion: extracted from shear test (such as direct shear or triaxial test) results for subsequent stability analysis;

[0122] Rock layer dip angle / dip direction: measured by field geological surveying and inclinometer, converted to angle type features, if model compatibility is required, vector decomposition method can be used for processing.

[0123] The parameters are stored in a structured database as geological input variables for subsequent modeling.

[0124] In this scheme, the extraction of structure control factors is as follows:

[0125] This dimension focuses on quantitatively expressing the deep structural factors that control geological disasters such as landslides and collapses:

[0126] Closest distance to fault: Based on existing regional geological maps, use the GIS spatial analysis module to calculate the closest distance from the center point of the analysis unit to the known fault zone;

[0127] Local joint density: Through field structural plane investigation and high-resolution rock mass image processing (such as structural plane recognition algorithm), the number of joints per unit area is counted;

[0128] DFN fracture connectivity index: Construct a discrete fracture network (Discrete Fracture Network, DFN) model, simulate fracture growth and connectivity, quantify the connectivity level of the regional rock mass, and introduce a connectivity index (such as P32 or connectivity rate) as an input feature.

[0129] In this scheme, the acquisition and expression of hydrological-monitoring information are as follows:

[0130] In order to further enhance the dynamic response capability of the model, the following hydrological and deformation monitoring data are introduced:

[0131] Groundwater level depth: Install pore water pressure gauges or groundwater monitoring wells to record water level depth data, and combine historical monitoring sequences to construct an annual periodic variation model;

[0132] Annual average rainfall: Based on national meteorological bureau statistical data or remote sensing rainfall inversion data (such as TRMM, GPM), use interpolation algorithm to construct regional average rainfall distribution map;

[0133] InSAR deformation rate: Use time series InSAR (such as PSInSAR, SBAS-InSAR) technology to obtain the annual average subsidence or uplift rate of the monitoring point;

[0134] GNSS three-dimensional displacement trend: Install GNSS monitoring points in key areas, collect continuous three-dimensional coordinate data, and calculate the displacement rate in X, Y, Z directions;

[0135] High shear strain recognition condition: In the strain field calculated based on GNSS and InSAR deformation rate, set a threshold (such as annual strain rate > 5 × 10 -6 ) to mark high shear strain area as a disaster potential warning factor.

[0136] In summary, the present scheme can effectively improve the data richness and prediction accuracy of the geological disaster identification model through the fusion processing of the above multi-dimensional geological and environmental parameters. The present embodiment supports the connection of a multi-source data acquisition system and seamlessly connects with existing geological modeling platforms (such as GOCAD, MOVE, FLAC3D), thereby realizing dynamic and accurate evaluation of regional geological stability.

[0137] As an optional embodiment of the present application, in S2,

[0138] By dividing the constructed three-dimensional geological model into spatial unit bodies of uniform size (for example, 10m x 10m x 10m, or flexibly set according to the resolution of the geological model), as the basic object for subsequent feature extraction and prediction analysis, each unit body has the following spatial attributes.

[0139] Coordinate position (records the three-dimensional coordinates (x, y, z) of the center point of each unit), belonging stratum identification (determines which stratum unit it belongs to according to its position in the three-dimensional model, which can be realized by means of spatial interpolation or point-surface relationship), and whether it passes through the structural surface (judges whether the current unit is crossed by the joint surface, fault surface or rock layer surface, which is realized according to the relationship between the structural surface equation and the spatial position of the voxel).

[0140] Specifically, the above spatial unit division and attribute labeling can be realized by the three-dimensional geological modeling platforms such as GOCAD, MOVE, Petrel, etc. with Python or C++ script interface for automatic processing.

[0141] In detail, the spatial attributes of each type of structural surface are constructed by extracting the information of joint surface, fault surface and rock layer surface from the constructed three-dimensional geological model, including strike and dip (the geometric direction information is extracted from the structural surface triangular network by using the face normal vector projection method), joint spacing (the average spacing of adjacent structural surfaces is counted in the distribution area of the structural surface), direction density (the number of structural surfaces in each direction is counted based on unit cubic meter to form a direction density tensor), and a normalized direction vector is generated for each structural surface unit.

[0142]

[0143] The above geometric features are used to construct a structural surface database and serve as the basis for calculating the structural master control vector.

[0144] The structural master control vector of each unit body is calculated through the direction vector of each unit body.

[0145]

[0146] where w iis the weight of the i-th structural plane (which can be determined by the joint plane spacing, length, and penetrability), n j is the number of structural planes passing through the voxel.

[0147] The weight therein can be calculated based on the geometric characteristics of the structural plane, for example:

[0148] wherein L i is the structural plane length, P i is the penetrability index (such as the penetration rate, connectivity probability), S i is the average joint spacing.

[0149] The scheme can realize comprehensive extraction of spatial information of the structural plane and quantitative expression of the master structure, and provides a basis for structural consistency modeling.

[0150] It is worth mentioning that the master control vector finally obtained represents the directional characteristics of the current unit in the structure, and the direction can reflect the potential structural control trend or sliding direction of the region, thereby providing basic data for subsequent consistency analysis, slip surface prediction, or master structure identification.

[0151] As a further improvement of the above scheme, in the S2:

[0152] In the first step, the structural sensitivity evaluation is realized by constructing an index consistent with the sliding direction, i.e., a structural control consistency index C j satisfies: wherein, represents the main direction of gravity action or the slope normal vector of the slope where the unit is located.

[0153] In the second step, in order to enhance the identification ability of the model to the master control factor of the structure and improve the accuracy of the relaxation zone identification, the structural control consistency index and the traditional features are fused to enhance the input dimension of the prediction model;

[0154] X j =[H j ,θ j ,E j ,φ j ,c j ,v j ,C j ...

[0155] As an optional embodiment of the present application, the S3 includes:

[0156] The gradient boosting tree (GBT model) classification model is trained by taking the fusion feature vector of the structural control consistency index as a training sample; wherein:

[0157] Sample construction: positive samples are unit bodies in known relaxation zones, negative samples are unit bodies in stable regions; specifically: through field investigation of engineering landslide, slope instability and other disaster cases, the actual relaxation or damage area is recorded. The corresponding unit body in these areas is labeled as a "relaxation zone" sample; other areas without deformation are labeled as "stable zone" samples. The existing landslide or geological disaster database (such as the China Landslide Disaster Database, Local Geological Disaster Point Database) can also be used to extract the landslide influence range, and the voxel units falling within these ranges are labeled as "relaxation zones".

[0158] In more detail, each sample contains multiple feature dimensions: topographic parameters (slope, aspect, etc.), geological parameters (lithology, joint density, etc.), monitoring parameters (InSAR rate, GNSS displacement, etc.), structural control consistency index C j , etc.

[0159] Prediction function of model structure Where f t (x) represents the t-th regression tree;

[0160] The binary cross-entropy loss function is used as the objective function;

[0161]

[0162] Where: p i = σ (F (x i )) is the relaxation probability, y i ∈ {0, 1}

[0163] Use GridSearch or BayesianOptimization to optimize hyperparameters;

[0164] Use 5-fold cross-validation to evaluate accuracy and generalization ability;

[0165] Important indicators for model evaluation include AUC, Recall, F1-score.

[0166] This scheme can effectively improve the prediction performance, optimize the recognition result, and ensure the generalization ability and robustness of engineering application.

[0167] This scheme effectively improves the recognition accuracy of the model for structural sensitive areas (such as relaxation zones and structural sliding zones) by introducing a structural control consistency index, enhances the generalization ability and engineering applicability of the model, and can be deployed on a geological information system platform. Through the fusion of three-dimensional models and structural features, high-resolution landslide sensitive area prediction and hierarchical early warning are realized

[0168] As an optional embodiment of the present application, in the S4

[0169] By deploying the model to the full slope voxel grid; output the relaxation probability for each cell,

[0170] By setting an identification threshold, mark the cells exceeding the threshold as "potential relaxation area";

[0171] By using DBSCAN or connectivity-based clustering method, adjacent high-probability cells are grouped into a continuous relaxation area.

[0172] Output the identification results.

[0173] This scheme can combine probability judgment and spatial clustering, realize the transformation from point prediction to surface relaxation area identification, and facilitate practical application and visualization.

[0174] By calculating the volume, depth, deformation direction and other indicators of each relaxation area, for subsequent early warning evaluation.

[0175] By superimposing the relaxation probability in the form of a heat map in a three-dimensional geological modeling platform (such as ParaView, Leapfrog).

[0176] The more detailed process of S4 is as follows:

[0177] First, deploy the trained relaxation identification model to the three-dimensional voxel grid of the entire slope area, and realize the spatial zoning identification and visualization of the relaxation area based on the prediction probability results and clustering method.

[0178] Deploy the aforementioned trained gradient boosting tree classification model to the three-dimensional cell (Voxel) grid of the entire slope area. Each voxel cell has a complete input feature vector (including topography, lithology, hydrology, structure control consistency index, etc.), and after inputting the model, the probability value of the cell being a "relaxation area" is output

[0179] And set an identification probability threshold τ, such as τ = 0.6 or dynamically set through ROC curve analysis. For each cell, when its predicted probability y j > τ, mark it as a "potential relaxation cell" and record it in the set S relax This process completes the preliminary screening from model prediction output to potential risk cells.

[0180] In order to eliminate the spatial fragmentation of single-point identification and improve the continuity and engineering readability of regional identification, further spatial clustering processing is performed on the potential relaxation cell set S relax The clustering method includes but is not limited to:

[0181] 1. Density-based clustering algorithm (such as DBSCAN); Specifically: set the distance threshold ε and the minimum sample number MinPts; Cluster the high-probability units with a spatial distance less than ε and meeting the minimum density condition into a continuous relaxed area.

[0182] 2. Clustering algorithm based on topological connectivity; Specifically: construct a voxel topological adjacency graph, traverse all potential relaxed units, find all connected blocks according to the 26-neighborhood (3D) or 8-neighborhood (2D), and mark the connected units as the same region.

[0183] Each clustering region is regarded as a continuous relaxed area unit cluster, numbered and output, forming a structured recognition result.

[0184] For each recognized continuous relaxed area, further calculate its spatial and mechanical characteristic parameters for subsequent early warning evaluation and risk classification, including the following parameters:

[0185] Volume: used to count the sum of the volumes of all included units;

[0186] Maximum burial depth / average depth: used to extract the burial depth attribute of each unit and count the area range;

[0187] Main deformation direction: calculate the average direction by combining the structure master vector of each unit and the slope normal vector;

[0188] Position center: used to calculate the centroid coordinates of the relaxed area for spatial positioning and visual expression;

[0189] In this scheme, in order to improve the practicality and engineering guidance value of the recognition result, the result is directly presented in a three-dimensional geological modeling platform. The specific operation includes:

[0190] Export the voxel recognition result in VTK, OBJ or PLY format;

[0191] In the three-dimensional visualization platform such as ParaView or LeapfrogGeo, superimposed display the relaxed probability heat map;

[0192] High-risk areas are represented in red, medium-risk areas are represented in orange or yellow, forming a three-dimensional heat map driven by probability values;

[0193] Optional functions include cross-section browsing, profile analysis, and superimposed display with faults / joint surfaces.

[0194] The present application realizes the region identification and visualization conversion from point probability output to spatial continuous surface relaxed area through the above steps, greatly improves the engineering interpretability and field applicability of the model, and has good popularization and application value.

[0195] The application can also integrate a multi-stage learning mechanism, and the specific operation is as follows:

[0196] Step 1: Identify potential relaxation zones by using the GBT model in the above scheme, achieve the effect of preliminary screening of high-risk spatial units, and record the probability value of each unit belonging to the “relaxation zone” output by the output as P GBT；

[0197] Step 2, analyze the time sequence deformation data through the LSTM model, achieve the effect of identifying the relaxation evolution trend; for the potential relaxation area identified in step 1, extract the historical displacement sequence (such as Δt=[δ1, δ2, δ3, δ4......δ T ]) of the monitoring points such as InSAR and GNSS corresponding to the potential relaxation area, input these time series into the LSTM (Long Short-Term Memory Neural Network) model for time sequence modeling, identify the relaxation evolution process with trend growth or mutation characteristics, and output the evolution risk score R LSTM

[0198] Step 3, by fusing the spatial probability and the time evolution score, achieve the effect of accurately predicting the high-risk relaxation zone; adopt a weighted fusion strategy to comprehensively process the outputs of the two models, for example:

[0199] P final =α·P GBT +(1-α)·R LSTM

[0200] Wherein, α is the adjustable fusion coefficient. Finally, P final is taken as the comprehensive risk score to determine the “high-risk relaxation zone”.

[0201] Through the above improvement, a complete risk modeling path from “static spatial attributes” to “dynamic deformation trend” can be realized, which greatly improves the identification accuracy of the model for “gradual change type” and “hidden type” relaxation zones, supports real-time updating and early warning response, and has good scalability and engineering application value.

[0202] Embodiment 2:

[0203] Referring to Figure 2 , the application also proposes a bank slope relaxation zone identification system based on three-dimensional geological modeling, comprising:

[0204] A geological modeling unit is used to receive geological survey and data acquisition results and construct a three-dimensional geological model of the bank slope area, and the model includes stratigraphic structure, faults, joint information and other structural surface information; the geological modeling unit is used to output structured three-dimensional geological model data.

[0205] A spatial unit processing unit is configured to divide the three-dimensional geological model into spatial unit bodies of uniform size, and extract spatial position, structure surface crossing condition and directional vector information of each unit to form a structure master directional vector; the spatial unit processing unit is configured to form spatial unit body data containing structure directional information and geometric position attributes.

[0206] A feature extraction unit is configured to extract a fusion feature vector of each spatial unit, the fusion feature vector including a structure control consistency index and traditional geological features including topographic geometric features, geomechanical features, structure control factors and hydrological-monitoring information; the feature extraction unit is configured to output a multi-dimensional fusion feature vector dataset corresponding to each spatial unit.

[0207] A relaxed zone prediction model training unit is configured to train a gradient boosting tree classification model based on the fusion feature vector, and perform parameter optimization and precision evaluation on the model;

[0208] A relaxed zone identification unit is configured to deploy the trained prediction model to a three-dimensional geological voxel grid, output a relaxed probability for each unit body, and identify potential relaxed zones according to a set threshold; the relaxed zone identification unit is configured to output an optimized prediction model, which can output a relaxed probability for any unit body.

[0209] A relaxed zone clustering unit is configured to use a density-based clustering (DBSCAN) or connectivity-based algorithm to classify adjacent high-probability units as continuous relaxed zones, and calculate the volume, depth and deformation direction of each relaxed zone; the relaxed zone clustering unit is configured to output a unit body set labeled with a relaxed probability.

[0210] A visualization display unit is configured to visualize and display the relaxed zone identification result in the form of a heat map superimposed on a three-dimensional geological model platform, for assisting decision-making and early warning evaluation. The visualization display unit supports visual display layers and analysis results for human-computer interaction.

[0211] In the present scheme, the data flow and function cooperation between the units are performed in the following order and dependency relationship.

[0212] Geological modeling unit→output three-dimensional model data;

[0213] Spatial unit processing unit→receive model division unit and extract structure information;

[0214] Feature extraction unit→aggregate geological and structural features to form model input;

[0215] Prediction model training unit→use samples to train and identify the model;

[0216] Relaxed zone identification unit→deploy the model to the global grid and output the probability result;

[0217] Relaxation zone clustering unit: aggregate probability results, identify continuous risk areas;

[0218] Visualization display unit: render and output spatial visualization results.

[0219] The application provides a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the bank slope relaxation zone identification method based on three-dimensional geological modeling when executing the computer program.

[0220] The application provides a computer readable storage medium, which stores a computer program, and the computer program implements the bank slope relaxation zone identification method based on three-dimensional geological modeling when executed by a processor.

[0221] The above formulas are all dimensionless numerical calculations, the formulas are obtained by software simulation of a large amount of data to obtain a formula closest to the actual situation, and the preset parameters and threshold values in the formulas are set by a person skilled in the art according to actual conditions.

[0222] The above embodiments can be realized by software, hardware, firmware or any combination thereof, in whole or in part. When realized by software, the above embodiments can be realized in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0223] The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (for example, infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center and the like containing one or more available medium collections. The available medium can be a magnetic medium (for example, floppy disk, hard disk, magnetic tape), an optical medium (for example, DVD), or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0224] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and module can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0225] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the division of the above-described device embodiments is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple devices or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different parts can be indirect couplings or communication connections through some interfaces, devices or modules, and can be electric, mechanical or in other forms.

[0226] The modules illustrated as separated components can or can not be physically separated, and the components illustrated as modules can or can not be physical modules, and can be located in one place or distributed on multiple network modules. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0227] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0228] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. The artificial intelligence is to use a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results.

[0229] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for identifying bank slope relaxation zones based on three-dimensional geological modeling, characterized in that: include S1: Construct a three-dimensional geological model of the bank slope area; S2: Divide the three-dimensional geological model of the bank slope area into several spatial units, and extract the fusion feature vector of each unit; and The fusion feature vector integrates the traditional features of each spatial unit and the structural control consistency index; the structural control consistency index is used to measure the degree of consistency between the main control direction of the unit structure and the main relaxation / slip direction; S3: Prediction model for relaxation zone identification based on fused feature vector training; S4: Identify the relaxation zone in the bank slope area based on the established relaxation zone prediction model.

2. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: Said S1 comprises: Build a basic geological database through comprehensive geological survey and data collection; Construct an initial geological structure model using professional 3D modeling software; Construct a refined three-dimensional attribute model through interpolation modeling and parameter assignment; Model validation and calibration through monitoring data and numerical simulations; Maintain the timeliness and accuracy of the model through a dynamic update mechanism.

3. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: The traditional features in S2 include terrain geometry, geomechanical features, structural control factors, and hydrological-monitoring information; in Topographic geometric features include but are not limited to: slope θ, slope aspect α, elevation H, and relative distance from the shoreline / slope crest / slope foot; Geomechanical characteristics include but are not limited to: lithology category code, elastic modulus E, Poisson's ratio v, internal friction angle φ, cohesion c, and rock layer dip / strike; Structural control factors include but are not limited to: the closest distance to the fault d f , local joint density and DFN fracture penetration index; Hydrological monitoring information includes but is not limited to: groundwater depth, annual average rainfall, InSAR deformation rate v INSAR , the three-dimensional displacement trend of the GNSS point, and whether it meets the high shear strain identification condition - that is, the strain rate is greater than the threshold.

4. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: In S2: The constructed 3D geological model is divided into spatial units of uniform size, which serve as the basic objects for subsequent feature extraction and prediction analysis. Each unit includes but is not limited to the following spatial attributes: coordinate position, stratigraphic identity, and whether it passes through joints / faults. Extracting joint, fault, and rock surface information from the constructed 3D geological model to construct spatial attributes of each type of structural surface, including but not limited to strike, dip, joint spacing, and direction density, and generating a normalized direction vector for each structural surface unit; Calculate the structural master vector of each unit body through the direction vector of each unit body; Among them, w i is the weight of the i-th structural surface, n j is the number of structural faces passing through the voxel.

5. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: In S2: By constructing an index consistent with the sliding direction, namely the structural control consistency index, the structural sensitivity assessment is achieved; the structural control consistency index C j satisfy: in, Indicates the main direction of gravity or slope normal vector of the slope where the unit is located; Integrate structural control consistency indicators with traditional features to enhance the input dimension of the prediction model; X j =[H j ,i j ,E j ,f j ,c j ,v j ,C j ......]。 6. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: In the S3: The fusion feature vector of the consistency index is controlled by the fusion structure as a training sample to train the gradient boosting tree classification model; wherein: Prediction function of the model structure where f t (x) represents the t-th regression tree; Use binary cross entropy loss function as the objective function; Where: p i =σ(F(x i )) is the relaxation probability, y i ∈{0, 1} Optimize hyperparameters using GridSearch or BayesianOptimization; Use 5-fold cross validation to evaluate accuracy and generalization ability; Important indicators for model evaluation include AUC, Recall, and F1-score.

7. The method for identifying bank slope relaxation zones based on three-dimensional geological modeling according to claim 1, characterized in that: The S4 includes By deploying the model to the entire bank slope voxel grid and outputting the relaxation probability for each cell, By setting the identification threshold, the cells exceeding the threshold are marked as "potential relaxation areas"; By using DBSCAN or connectivity-based clustering methods, adjacent high-probability units are grouped into a continuous relaxation region and the identification results are output.

8. A bank slope relaxation zone identification system based on three-dimensional geological modeling, characterized in that: include: A geological modeling unit is used to receive geological survey and data collection results and construct a three-dimensional geological model of the slope area, the model including the following structural surface information: stratum structure, faults, and joints; A spatial unit processing unit is used to divide the three-dimensional geological model into spatial units of uniform size, and extract the spatial position, structural surface crossing and direction vector information of each unit to form a structural master control vector; A feature extraction unit is used to extract a fused feature vector of each spatial unit, wherein the fused feature vector includes a structural control consistency index and traditional geological features, wherein the traditional geological features include terrain geometric features, geomechanical features, structural control factors, and hydrological-monitoring information; The slack zone prediction model training unit is used to train the gradient boosting tree classification model based on the fused feature vector, and to optimize the model parameters and evaluate its accuracy; The relaxation zone identification unit is used to deploy the trained prediction model to the 3D geological voxel grid, output the relaxation probability for each unit volume, and identify potential relaxation zones based on the set threshold; Relaxation zone clustering unit, which is used to classify adjacent high-probability units into continuous relaxation zones using density clustering (DBSCAN) or connectivity-based algorithms, and calculate the volume, depth and deformation direction of each relaxation zone; The visualization display unit is used to overlay the relaxation zone identification results in the form of a heat map on the three-dimensional geological model platform for visualization, which is used to assist decision-making and early warning assessment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method for identifying a slope relaxation zone based on three-dimensional geological modeling according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying a bank slope relaxation zone based on three-dimensional geological modeling according to any one of claims 1 to 7 is implemented.

Images