Dam geological structure stability verification method and system based on digital twin

Abstract

The present invention discloses a method and system for verifying the stability of the geological structure of a dam based on digital twins, including: first, by acquiring remote sensing images and geological exploration data of the target dam, a fine three-dimensional geological model of the dam is constructed, and a preset digital twin platform is used to perform simulation under simulated extreme conditions. Such a design can comprehensively and intuitively display the changes in the geological structure of the dam under extreme conditions, and by performing stability analysis on the three-dimensional model after simulation, the geological structure stability of the dam can be quickly and accurately evaluated, providing a scientific basis for the design, construction and operation and maintenance of water conservancy projects, and significantly improving the evaluation efficiency and accuracy.

Description

Technical Field

[0001] The present invention relates to the fields of engineering geology and hydraulic structures, and in particular to a method and system for verifying the stability of a dam geological structure based on digital twins. Background Art

[0002] In the construction of water conservancy projects, the stability of the geological structure of the dam is a key factor in ensuring the safety of the project. However, due to the complexity and variability of geological conditions, traditional geological exploration and stability assessment methods are often difficult to fully and accurately reflect the actual conditions of the dam, especially the stability performance under extreme conditions. In addition, existing methods usually rely on a large number of field tests and data analysis, which is time-consuming, labor-intensive and costly. Summary of the invention

[0003] The object of the present invention is to provide a method and system for verifying the stability of a dam geological structure based on digital twins.

[0004] In a first aspect, an embodiment of the present invention provides a method for verifying the stability of a dam geological structure based on digital twins, comprising:

[0005] Acquire remote sensing images and multiple geological exploration data for a target dam, and construct a three-dimensional geological model of the target dam corresponding to the target dam;

[0006] The three-dimensional geological model of the target dam is transmitted to a preset digital twin platform, wherein the three-dimensional geological model of the target dam is composed of a plurality of geological structure modules;

[0007] Acquire preset simulated extreme condition data, and simulate the three-dimensional geological model of the target dam based on the preset simulated extreme condition data on the preset digital twin platform to obtain a three-dimensional geological model of the dam after simulation;

[0008] Performing stability analysis on the simulated three-dimensional geological model of the dam to obtain the geological structure type corresponding to each geological structure module;

[0009] According to the multiple geological structure types, the verification result of the geological structure stability of the dam body of the target dam is obtained.

[0010] In a second aspect, an embodiment of the present invention provides a server system, including a server, wherein the server is used to execute the method described in the first aspect.

[0011] Compared with the prior art, the beneficial effects provided by the present invention include: adopting a digital twin-based dam geological structure stability verification method and system disclosed in the present invention, by acquiring remote sensing images and geological exploration data of the target dam, constructing a fine three-dimensional geological model of the dam, and using a preset digital twin platform to simulate under simulated extreme conditions. Such a design can comprehensively and intuitively display the geological structure changes of the dam under extreme conditions, and by performing stability analysis on the three-dimensional model after simulation, quickly and accurately evaluate the geological structure stability of the dam, providing a scientific basis for the design, construction and operation and maintenance of water conservancy projects, and significantly improving the evaluation efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can also be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of the steps of a method for verifying the stability of a dam geological structure based on digital twins provided in an embodiment of the present invention;

[0014] Figure 2 A schematic block diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0016] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0017] In order to solve the technical problems in the aforementioned background technology, Figure 1 A schematic flow chart of a method for verifying the stability of a dam's geological structure based on digital twins is provided in an embodiment of the present disclosure. The method for verifying the stability of a dam's geological structure based on digital twins is introduced in detail below.

[0018] Step S201, acquiring remote sensing images and multiple geological exploration data for a target dam, and constructing a three-dimensional geological model of the target dam corresponding to the target dam;

[0019] Step S202, transmitting the three-dimensional geological model of the target dam to a preset digital twin platform, wherein the three-dimensional geological model of the target dam is composed of a plurality of geological structure modules;

[0020] Step S203, obtaining preset simulated extreme condition data, and simulating the three-dimensional geological model of the target dam based on the preset simulated extreme condition data on the preset digital twin platform to obtain a three-dimensional geological model of the dam after simulation;

[0021] Step S204, performing stability analysis on the simulated three-dimensional geological model of the dam to obtain the geological structure type corresponding to each geological structure module;

[0022] Step S205: obtaining a verification result of the geological structure stability of the target dam according to the multiple geological structure types.

[0023] In an embodiment of the present invention, illustratively, the server first obtains high-resolution images of the target dam from a satellite remote sensing data source, and these images provide detailed features of the dam surface. At the same time, the server retrieves multiple geological exploration data about the target dam from a geological exploration database, including geological layer distribution, lithology, fault information, etc. Using these data, the server runs a three-dimensional modeling software to construct a three-dimensional geological model of the target dam, which reflects the geological structure of the dam in detail. The server transmits the constructed three-dimensional geological model file of the dam to a preset digital twin platform through a high-speed network, which is specifically used for virtual simulation of complex systems. On the digital twin platform, the three-dimensional geological model of the dam is decomposed into multiple geological structure modules, such as bedrock layer, cover layer, fault zone, etc., for more detailed analysis. The server obtains preset extreme condition data, such as heavy rainfall, seismic waves, etc., from the meteorological and geological disaster database. On the digital twin platform, the server runs a simulation program, applies these extreme conditions to the three-dimensional geological model of the target dam, simulates the response of the dam under these conditions, and generates a three-dimensional geological model of the dam after simulation. The server uses the built-in stability analysis algorithm to calculate the simulated three-dimensional geological model of the dam and evaluate the stability of each geological structure module under extreme conditions. Through analysis, the server can identify different types of dam geological structures, such as stable bedrock, potential sliding surface, and easily eroded cover. The server summarizes the stability analysis results of all geological structure modules and comprehensively evaluates the overall stability of the target dam under extreme conditions. Finally, the server generates a detailed dam geological structure stability verification report, including stability rating, potential risk areas, recommended reinforcement measures and other information, providing a scientific basis for the maintenance and management of the dam.

[0024] In the embodiment of the present invention, the aforementioned step S204 can be implemented through the following example execution.

[0025] Acquire a target geological structure module and a plurality of undetermined geological structure types, wherein the target geological structure module is any one of the plurality of geological structure modules;

[0026] Performing feature extraction processing on the target geological structure module and each undetermined geological structure type respectively to obtain an overall feature vector of the target geological structure module and a category feature vector of each undetermined geological structure type;

[0027] According to the overall feature vector, predicting the matching status of the target geological structure module with each preset geological structure module;

[0028] Obtaining a network embedding model corresponding to each preset geological structure module, wherein the network embedding model indicates the correlation between the to-be-determined geological structure types in the preset geological structure module;

[0029] According to the matching situation and the network embedding model corresponding to each preset geological structure module, node vector feature processing is performed on the category feature vector of each undetermined geological structure type to obtain the undetermined category feature vector corresponding to each undetermined geological structure type;

[0030] Determine the matching confidence of each undetermined geological structure type with the target geological structure module according to the undetermined category feature vector corresponding to each undetermined geological structure type;

[0031] According to the matching confidence, the target geological structure type corresponding to the target geological structure module is determined from each of the undetermined geological structure types.

[0032] In an embodiment of the present invention, exemplarily, the server selects a target geological structure module, such as a "bedrock layer module", from the three-dimensional geological model of the dam after simulation. At the same time, the server retrieves multiple undetermined geological structure types from the geological structure type database, such as "stable bedrock", "potential sliding surface", "easy to erode cover layer", etc. The server runs a feature extraction algorithm to extract features from the target geological structure module "bedrock layer module" to obtain its overall feature vector, including lithology, thickness, inclination, etc. For each undetermined geological structure type, the server also extracts features to obtain their respective category feature vectors, such as the feature vector of "stable bedrock" may include high strength, low permeability, etc. The server uses a machine learning model to predict the degree of matching between the target geological structure module "bedrock layer module" and each preset geological structure module based on its overall feature vector. The server retrieves the corresponding network embedding model of each preset geological structure module from the model library, which describes the correlation between the undetermined geological structure types, such as the strong correlation between "stable bedrock" and "not easy to slide". The server further processes the category feature vector of each undetermined geological structure type based on the matching situation and the network embedding model to enhance its features related to the target geological structure module. The server calculates the matching confidence between each undetermined geological structure type and the target geological structure module "bedrock layer module". For example, the matching confidence of "stable bedrock" may be higher. Based on the matching confidence, the server determines that the target geological structure type of the "bedrock layer module" is "stable bedrock" from multiple undetermined geological structure types. Through the above steps, the server can accurately analyze the geological structure type corresponding to each geological structure module in the three-dimensional geological model of the dam after simulation, providing an important basis for the stability assessment of the dam body.

[0033] In order to more clearly describe the solution provided by the embodiment of the present application, a more detailed description is given below.

[0034] In an embodiment of the present invention, exemplarily, the target geological structure module refers to a specific geological structure module selected from the three-dimensional geological model of the dam after simulation, which is used for subsequent stability analysis. It can be any part of the geological model, such as bedrock layer, cover layer, fault, etc. In the three-dimensional geological model of the dam, the server may select the "bedrock layer module" as the target geological structure module in order to conduct an in-depth analysis of its stability. Multiple pending geological structure types are a set of predefined geological structure types for comparison and matching with the target geological structure module. These types are determined based on geological knowledge and experience, representing different geological characteristics and stability levels. The server retrieves multiple pending types from the geological structure type database, such as "stable bedrock", "potential sliding surface", "erodible cover layer", etc., for matching analysis with the target geological structure module. Multiple geological structure modules refer to the multiple components into which the three-dimensional geological model of the dam after simulation is decomposed, each of which has unique geological characteristics and properties. These modules together constitute the geological structure of the entire dam body. In the 3D geological model of the dam, in addition to the "bedrock layer module", there may also be multiple other geological structure modules such as the "cover layer module" and the "fault module". Any geological structure module indicates that the target geological structure module can be any one of the multiple geological structure modules. The server can select any module for stability analysis as needed. In the 3D geological model of the dam, the server can select the "cover layer module" as the target geological structure module instead of the "bedrock layer module" in order to analyze its stability separately.

[0035] In addition, feature extraction processing refers to the process of extracting meaningful features or attributes that can represent the essence of data from raw data. In geological structure analysis, feature extraction processing usually involves extracting key geological features or attributes from geological structure modules or geological structure types. For the target geological structure module "bedrock layer module", feature extraction processing may include extracting key geological features such as lithology, thickness, inclination, and degree of development of joints and fissures. The overall feature vector refers to combining all key features or attributes of the target geological structure module into a vector to comprehensively describe the geological characteristics of the module. This vector contains all the important information of the module and can be used for subsequent analysis and comparison. For the "bedrock layer module", its overall feature vector may be a vector containing multiple elements such as lithology code, thickness value, inclination value, and degree of development of joints and fissures. The category feature vector refers to combining all key features or attributes of each undetermined geological structure type into a vector to describe the geological characteristics of the type. This vector represents all the important information of the type and can be used for comparison and matching with the target geological structure module. For the undetermined geological structure type "stable bedrock", its category feature vector may be a vector containing multiple elements such as high strength, low permeability, and non-weathering. The matching situation refers to the similarity or matching degree between the target geological structure module and each preset geological structure module. It reflects the proximity of the target module and the preset module in geological characteristics. Through prediction, the matching situation of the "bedrock layer module" and the "stable bedrock" preset module may be shown as a high match, while the matching situation with the "potential sliding surface" preset module may be shown as a mismatch or a low match.

[0036] In addition, the preset geological structure module refers to a set of predefined geological structure modules with specific geological characteristics and attributes. These modules are usually divided based on geological knowledge and experience for subsequent geological analysis and comparison. A set of preset modules may be stored in the server, such as "stable bedrock module", "potential sliding surface module", "erodible cover layer module", etc., each of which represents a specific geological structure type. The network embedding model is a method of mapping nodes in a network (in the embodiment of the present invention, each undetermined geological structure type) to vector representations in a low-dimensional space. These vectors capture the structural relationship and similarity between nodes and can be used to measure the correlation or distance between nodes. For the "stable bedrock module", its network embedding model may be a set of vectors that map types such as "stable bedrock", "potential sliding surface", and "erodible cover layer" to a low-dimensional space. The relative position and distance between these vectors reflect the geological correlation and similarity between them. The undetermined geological structure type refers to a set of predefined geological structure types that are used to compare and match with the target geological structure module or the preset geological structure module. These types represent different geological characteristics and stability levels. For example, "stable bedrock", "potential sliding surface", "erodible cover layer" and so on are all examples of undetermined geological structure types. Correlation refers to the similarity or correlation of the undetermined geological structure types in terms of geological characteristics, stability, formation conditions, etc. It reflects whether there is some kind of inherent connection or law between the types. In the "stable bedrock module", the "stable bedrock" type may have a low correlation with the "potential sliding surface" type because their geological characteristics and stability levels are quite different; while the "stable bedrock" type may have a high correlation with the "another stable bedrock subtype" because their geological characteristics and stability levels are similar. Matching refers to the similarity or matching degree between the target geological structure module and each preset geological structure module. This matching is based on multiple considerations such as geological characteristics, stability, and formation conditions. Through the previous prediction steps, we may get the result that the "bedrock layer module" is highly matched with the "stable bedrock preset module", but not with the "potential sliding surface preset module".

[0037] In addition, node vector feature processing refers to further processing the category feature vector using the network embedding model to obtain a new feature vector that can reflect the correlation and similarity between types. For the category feature vector of "stable bedrock", we can use the network embedding model of the "stable bedrock pre-set module" to process its node vector features to obtain a new vector. This new vector not only contains the geological characteristics of "stable bedrock", but also implies its correlation with other types in the "stable bedrock pre-set module". The pending category feature vector refers to a new feature vector obtained after node vector feature processing that can more comprehensively reflect the geological characteristics of the pending geological structure type and its correlation with other types. After node vector feature processing, the category feature vector of "stable bedrock" may become a new vector, which is the pending category feature vector of "stable bedrock", which contains more geological information and correlation with other types. The pending geological structure type refers to a set of predefined geological structure types that are used for comparison and matching with the target geological structure module. These types represent different geological characteristics and stability levels, but their matching degree with the target module has not yet been determined. For example, "potential sliding surface" and "erodible cover layer" are all examples of undetermined geological structure types. The undetermined category feature vector refers to a new feature vector obtained after node vector feature processing, which can more comprehensively reflect the geological characteristics of the undetermined geological structure type and its correlation with other types. This vector contains all important information for subsequent matching analysis. For the undetermined geological structure type "potential sliding surface", its undetermined category feature vector may be a new vector containing lithology, dip, degree of development of joints and fissures, and correlation with other types. The target geological structure module refers to a specific geological structure module that has been previously selected and feature extracted. It is the focus of the current analysis and is used to match the undetermined geological structure type. For example, the "bedrock layer module" can be the target geological structure module. Matching confidence refers to the degree of matching or credibility between each undetermined geological structure type and the target geological structure module. It reflects the similarity and consistency of the geological characteristics between the undetermined type and the target module. Through matching analysis, the matching confidence between the “potential sliding surface” type and the “bedrock layer module” may be low because their geological characteristics and stability levels are quite different; while the matching confidence between the “another stable bedrock subtype” and the “bedrock layer module” may be high because their geological characteristics and stability levels are similar.

[0038] In addition, each pending geological structure type refers to a set of predefined geological structure types, which are used to compare and match with the target geological structure module. These types represent different geological characteristics and stability levels, but before the matching analysis, we are not sure which type best matches the target module. For example, the "potential sliding surface", "erodible cover layer" and "stable bedrock subtype" in the above example are all examples of pending geological structure types. Determining the target geological structure type corresponding to the target geological structure module is to select the type that best matches the target geological structure module from all pending geological structure types based on the matching confidence. This type is considered to be the type that best represents the geological characteristics of the target module. In the above example, we will select "stable bedrock subtype" as the target geological structure type for the "bedrock layer module" based on the matching confidence because it has the highest matching confidence of 0.8.

[0039] In the embodiment of the present invention, the feature extraction processing is performed on the target geological structure module and each undetermined geological structure type respectively to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type, which can be implemented through the following example.

[0040] Performing a feature extraction operation on a parameter description of the target geological structure module in at least one data dimension to obtain a module description feature of the target geological structure module in at least one data dimension;

[0041] Perform feature extraction operations on each undetermined geological structure type respectively to obtain geological structure type features of each undetermined geological structure type;

[0042] Constructing a feature set according to the geological structure type characteristics of each undetermined geological structure type and the learnable feature representation;

[0043] The module description features and the feature set are subjected to significance detection to obtain an overall feature vector of the target geological structure module and a category feature vector of each undetermined geological structure type.

[0044] In an embodiment of the present invention, exemplarily, the server first obtains the data of the target geological structure module, which may include geological exploration data, dam material properties, historical stability records, etc. Under at least one data dimension (such as geological layer distribution, lithology characteristics, joint and fissure development degree, etc.), the server performs a feature extraction operation. For example, for the dimension of geological layer distribution, the server may extract the thickness, distribution range, interlayer relationship and other characteristics of different geological layers. Through these operations, the server obtains the module description features of the target geological structure module under at least one data dimension, and these features fully describe the geological structure of the target module. The server then obtains the data of each undetermined geological structure type, which may include type definition, geological feature description, stability level, etc. For each undetermined type, the server performs a feature extraction operation and extracts its geological structure type features. For example, for the "potential sliding surface" type, the server may extract its inclination, lithology, joint and fissure development degree and other features. Through these operations, the server obtains the geological structure type features of each undetermined geological structure type, and these features fully describe the geological features of each undetermined type. The server constructs a feature set based on the geological structure type characteristics of each undetermined geological structure type and the learnable feature representation (such as the feature representation learned by the machine learning algorithm). This feature set contains the geological feature information of all undetermined types and is learnable, which means that it can be updated and optimized based on new data. The server performs a saliency test on the module description features and feature set of the target geological structure module. The purpose of the saliency test is to determine which features are significantly different between the target module and the undetermined type, thereby highlighting important geological features. Through the saliency test, the server obtains the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type. These feature vectors are the basis for subsequent matching analysis and stability verification. By executing the above steps, the server performs comprehensive feature extraction processing on the target geological structure module and each undetermined geological structure type, and obtains the corresponding feature vectors. These feature vectors will be used in subsequent matching analysis and stability verification to evaluate the stability of the dam geological structure.

[0045] In the embodiment of the present invention, the significance detection of the module description features and the feature set to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type can be performed through the following examples.

[0046] Determine a candidate module description feature under a target data dimension from the module description feature under the at least one data dimension;

[0047] Performing significance detection on the module description features and the feature set under the target data dimension to obtain a significance feature vector of the target geological structure module and a significance feature vector of each undetermined geological structure type;

[0048] Determining an iterative feature set according to the significant feature vector of the target geological structure module and the significant feature vector of each undetermined geological structure type;

[0049] Repeat the step of determining the candidate module description features under the target data dimension from the module description features under the at least one data dimension until the module description features under all data dimensions are executed, and obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type according to the current iterative feature set.

[0050] In an embodiment of the present invention, exemplarily, the server first analyzes the module description features under at least one data dimension, which may include geological layer distribution, lithological characteristics, joint and fissure development degree, etc. Through a certain strategy (such as based on the importance, relevance or stability of the features), the server determines the candidate target data dimensions, such as "geological layer distribution" and "lithological characteristics" from these data dimensions. The server extracts the module description features under these two target data dimensions as the focus of subsequent significance detection. The server performs significance detection on the module description features and feature sets under the target data dimensions. The purpose of significance detection is to find out which features are significantly different between the target module and the pending type. For example, the server may find that the thickness of a specific layer in "geological layer distribution" is significantly different between the target module and certain pending types. Through significance detection, the server obtains the significance feature vector of the target geological structure module and the significance feature vector of each pending geological structure type. The server determines the iterative feature set based on the results of the significance detection. This iterative feature set contains features that are outstanding in the significance detection, which are of great significance for distinguishing different geological structure types. The server repeatedly executes the steps of determining the module description features under the candidate target data dimension from the module description features under at least one data dimension. In each iteration, the server considers the new data dimension, performs significance detection, and updates the iterative feature set. This process continues until the module description features under all data dimensions are executed. Based on the final iterative feature set, the server obtains the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type. These feature vectors are the basis for subsequent matching analysis and stability verification. They fully reflect the differences and similarities in geological characteristics between the target module and the undetermined type. By executing the above steps, the server performs a comprehensive significance test on the target geological structure module and each undetermined geological structure type, and obtains the corresponding feature vectors. These feature vectors will be used in subsequent matching analysis and stability verification to evaluate the stability of the dam geological structure.

[0051] In an embodiment of the present invention, the network embedding model includes a type entity corresponding to each undetermined geological structure type, and a connection relationship between type entities, wherein the connection relationship indicates the correlation between two connected type entities; according to the matching situation and the network embedding model corresponding to each preset geological structure module, the category feature vector of each undetermined geological structure type is processed with node vector features to obtain the undetermined category feature vector corresponding to each undetermined geological structure type, which can be implemented through the following example.

[0052] For each undetermined geological structure type, determining a corresponding target type entity of the undetermined geological structure type in a corresponding network embedding model of each preset geological structure module;

[0053] For each network embedding model corresponding to the preset geological structure module, determining the relevant feature vector of the target type entity in the preset geological structure module according to the correlation between the target type entity and the remaining type entities in the network embedding model;

[0054] According to the matching situation, the relevant feature vectors of the target type entity in each preset geological structure module are integrated to obtain the integrated relevant feature vector of the undetermined geological structure type;

[0055] According to the integrated related feature vector, the category feature vector of the undetermined geological structure type is adjusted to obtain the undetermined category feature vector corresponding to the undetermined geological structure type.

[0056] In the embodiment of the present invention, exemplarily, for each undetermined geological structure type, the server determines the target type entity corresponding to it in the network embedding model corresponding to each preset geological structure module. For example, if there is a undetermined geological structure type of "potential sliding surface", the server will find the type entity representing the "potential sliding surface" in the network embedding model. For each network embedding model corresponding to the preset geological structure module, the server determines the relevant feature vector of the target type entity in the preset geological structure module according to the correlation between the target type entity and the remaining type entities in the network embedding model. For example, the server may find that the "potential sliding surface" type entity has a strong correlation with the "erodible cover" type entity in the network embedding model, so it will extract the relevant feature vector between the two type entities. The server integrates the relevant feature vectors of the target type entity in each preset geological structure module according to the matching situation, and obtains the integrated relevant feature vector of the undetermined geological structure type. For example, if the "potential sliding surface" type entity has relevant feature vectors in multiple preset geological structure modules, the server will integrate these feature vectors to form a more comprehensive integrated relevant feature vector. Finally, the server adjusts the category feature vector of the undetermined geological structure type based on the integration of relevant feature vectors, and obtains the corresponding undetermined category feature vector of the undetermined geological structure type. For example, the server may fine-tune the category feature vector of the "potential sliding surface" based on the information in the integration of relevant feature vectors to more accurately reflect its characteristics in the geological structure of the dam body. By executing the above steps, the server uses the network embedding model to process the node vector features of the category feature vector of each undetermined geological structure type, and obtains the corresponding undetermined category feature vector. These undetermined category feature vectors will be used in subsequent matching analysis and stability verification to more accurately evaluate the stability of the geological structure of the dam body.

[0057] In an embodiment of the present invention, the matching situation includes the matching content of the target geological structure module matching each preset geological structure module; according to the matching situation, the relevant feature vectors of the target type entity in each preset geological structure module are integrated to obtain the integrated relevant feature vector of the undetermined geological structure type, which can be implemented through the following example.

[0058] For each preset geological structure module, the matching content of the target geological structure module with the preset geological structure module is integrated with the relevant feature vector of the target type entity in the preset geological structure module to obtain the original integrated relevant features of the undetermined geological structure type in the preset geological structure module;

[0059] According to the original integrated related features of the undetermined geological structure type in each preset geological structure module, an integrated related feature vector of the undetermined geological structure type is determined.

[0060] In the embodiment of the present invention, exemplarily, for each preset geological structure module, the server integrates the matching content of the target geological structure module with the preset geological structure module and the relevant feature vector of the target type entity in the preset geological structure module. For example, if the target geological structure module has a good match with a preset geological structure module in terms of "geological layer distribution" and "lithological characteristics", and the "potential sliding surface" type entity in the preset geological structure module has a strong correlation with the "easy erosion cover layer" type entity, the server will integrate the matching content of these two aspects with the relevant feature vector to form the original integrated related features of the "potential sliding surface" in the preset geological structure module. The server determines the integrated related feature vector of the undetermined geological structure type according to the original integrated related features of the undetermined geological structure type in each preset geological structure module. For example, if the "potential sliding surface" has original integrated related features in multiple preset geological structure modules, the server will further integrate and optimize these features to form an integrated related feature vector that can fully reflect the characteristics of the "potential sliding surface" in the dam body geological structure. By executing the above steps, the server integrates the relevant feature vectors of the target type entity in each preset geological structure module according to the matching situation, and obtains the integrated relevant feature vectors of the undetermined geological structure type. These integrated relevant feature vectors will be used in subsequent matching analysis and stability verification to more accurately evaluate the stability of the dam geological structure.

[0061] In the embodiment of the present invention, the determination of the matching confidence of each undetermined geological structure type with the target geological structure module according to the undetermined category feature vector corresponding to each undetermined geological structure type can be implemented through the following example.

[0062] For each undetermined geological structure type, integrating the category feature vector of the undetermined geological structure type and the undetermined category feature vector to obtain a target category feature vector of the undetermined geological structure type;

[0063] According to the target category feature vector of each undetermined geological structure type, the matching confidence of each undetermined geological structure type and the target geological structure module is determined.

[0064] In the embodiment of the present invention, exemplarily, for each undetermined geological structure type, the server integrates the original category feature vector of the type and the undetermined category feature vector processed by the network embedding model. For example, for the undetermined geological structure type of "potential sliding surface", the server integrates its original category feature vector (such as geological layer distribution, lithological characteristics, etc.) and the undetermined category feature vector adjusted by the network embedding model (such as correlation with the erodible cover layer) to obtain the target category feature vector of the "potential sliding surface". The server determines the matching confidence of the target category feature vector of each undetermined geological structure type with the target geological structure module. For example, the server calculates the similarity or distance between the target category feature vector of the "potential sliding surface" and the overall feature vector of the target geological structure module, and determines the matching confidence of the "potential sliding surface" with the target geological structure module based on this similarity or distance. If the target category feature vector of the "potential sliding surface" is very similar to the overall feature vector of the target geological structure module, then the matching confidence of the "potential sliding surface" with the target geological structure module will be very high. By executing the above steps, the server integrates the category feature vectors of each undetermined geological structure type and the undetermined category feature vectors, and determines the matching confidence of the target geological structure module based on the integrated target category feature vectors. These matching confidences will be used in subsequent matching analysis and stability verification to evaluate the matching degree and stability between different geological structure types and the target geological structure module.

[0065] In the embodiment of the present invention, the feature extraction processing is performed on the target geological structure module and each undetermined geological structure type respectively to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type, which can be implemented through the following example.

[0066] Through the dam body geological structure classification model, feature extraction processing is performed on the target geological structure module and each undetermined geological structure type respectively to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type.

[0067] In an embodiment of the present invention, exemplarily, the server uses the trained dam geological structure classification model to extract features from the target geological structure module and each undetermined geological structure type. This classification model is trained based on a large amount of dam geological structure data, and can identify and extract the key features of the dam geological structure. The server inputs the data of the target geological structure module into the classification model, and the model processes these data to extract the overall features of the target geological structure module. For example, the data of the target geological structure module may include information on geological layer distribution, lithological characteristics, geological structure, etc. The classification model will convert this information into an overall feature vector, which can fully reflect the characteristics of the target geological structure module. The server also inputs the data of each undetermined geological structure type into the classification model, and the model processes these data to extract the category features of each undetermined geological structure type. For example, for the undetermined geological structure type of "potential sliding surface", the classification model will extract key information such as its geological layer distribution, lithological characteristics, and relationship with other geological structures, and convert this information into a category feature vector. The server applied the dam body geological structure classification model to extract features from the target geological structure module and each undetermined geological structure type, and obtained the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type. These feature vectors will be used in subsequent matching analysis and stability verification to evaluate the stability and matching degree of the dam body geological structure.

[0068] In an embodiment of the present invention, the feature extraction process is performed on the target geological structure module and each undetermined geological structure type respectively through the dam geological structure classification model to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type, and the implementation can also be performed through the following examples.

[0069] Acquire a sample data set and a network embedding model corresponding to each preset geological structure module, wherein the sample data set includes a plurality of sample geological structure modules and actual geological structure types of the sample geological structure modules;

[0070] Through the dam body geological structure classification model, feature extraction processing is performed on the sample geological structure module and each undetermined geological structure type respectively to obtain a sample overall feature vector of the sample geological structure module and a sample category feature vector of each undetermined geological structure type;

[0071] According to the overall feature vector of the sample, predicting the matching status of the sample geological structure module with each preset geological structure module;

[0072] According to the matching situation and the network embedding model corresponding to each preset geological structure module, node vector feature processing is performed on the sample category feature vector of each undetermined geological structure type to obtain the undetermined sample category feature vector corresponding to each undetermined geological structure type;

[0073] Determine the matching confidence of each undetermined geological structure type with the sample geological structure module according to the undetermined sample category feature vector corresponding to each undetermined geological structure type;

[0074] According to the matching confidence and the actual geological structure type of the sample geological structure module, the model parameters in the dam geological structure classification model are optimized to obtain a trained dam geological structure classification model.

[0075] In an embodiment of the present invention, exemplarily, the server first obtains a sample data set, which includes multiple sample geological structure modules and the actual geological structure types of these sample geological structure modules. At the same time, the server also obtains the network embedding model corresponding to each preset geological structure module, which is used to describe the correlation between geological structure types. The server uses the dam geological structure classification model (in the training stage, this model is initialized or based on previous training) to perform feature extraction processing on the sample geological structure module and each undetermined geological structure type. Through processing, the server obtains the sample overall feature vector of the sample geological structure module and the sample category feature vector of each undetermined geological structure type. The server predicts the matching of the sample geological structure module with each preset geological structure module based on the sample overall feature vector. For example, the server may predict that a certain sample geological structure module is most matched with the preset geological structure module of "rock slope". The server performs node vector feature processing on the sample category feature vector of each undetermined geological structure type based on the predicted matching situation and the network embedding model corresponding to each preset geological structure module. Through processing, the server obtains the corresponding undetermined sample category feature vectors for each undetermined geological structure type, which more accurately reflects the characteristics of the undetermined geological structure type in the dam body geological structure. The server determines the matching confidence of each undetermined geological structure type with the sample geological structure module based on the corresponding undetermined sample category feature vectors for each undetermined geological structure type. For example, the server may calculate that the matching confidence of the "potential sliding surface" matching a certain sample geological structure module is 0.8. Finally, the server optimizes the model parameters in the dam body geological structure classification model based on the matching confidence and the actual geological structure type of the sample geological structure module. Through optimization, the server obtains a trained dam body geological structure classification model, which can more accurately extract the characteristics of the dam body geological structure and be used in subsequent matching analysis and stability verification. By executing the above steps, the server trains the dam body geological structure classification model and enables it to accurately extract the feature vectors of the target geological structure module and each undetermined geological structure type for subsequent matching analysis and stability verification.

[0076] In an embodiment of the present invention, the model parameters in the dam geological structure classification model are optimized according to the matching confidence and the actual geological structure type of the sample geological structure module to obtain a trained dam geological structure classification model, which can be implemented through the following examples.

[0077] Determining the error parameter of the dam body geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module;

[0078] When the error parameter does not reach a preset error parameter threshold, optimizing the model parameters in the dam body geological structure classification model according to the error parameter;

[0079] According to the matching condition between the sample geological structure module and each preset geological structure module, each sample geological structure module is divided to obtain a set of sample geological structure modules corresponding to each preset geological structure module;

[0080] For each preset geological structure module, according to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module is adjusted to obtain the adjusted network embedding model corresponding to the preset geological structure module;

[0081] According to the optimized dam body geological structure classification model and the adjusted network embedding model corresponding to each preset geological structure module, the steps of performing feature extraction processing on the sample geological structure module and each undetermined geological structure type through the dam body geological structure classification model are repeated until the error parameter of the dam body geological structure classification model reaches the preset error parameter threshold, thereby obtaining a trained dam body geological structure classification model.

[0082] In an embodiment of the present invention, exemplarily, the server calculates the error parameter of the dam body geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module. For example, if the actual type of a sample geological structure module is "rock slope", but the type with the highest matching confidence predicted by the model is "soil slope", the server will calculate the difference between the two types and use it as part of the error parameter. When the error parameter does not reach the preset error parameter threshold, the server will optimize the model parameters in the dam body geological structure classification model according to the error parameter. The optimization process may include adjusting the weight of the model, changing the learning rate, or updating the architecture of the model to reduce the prediction error and improve the accuracy of the model. The server divides the sample geological structure module according to the matching of the sample geological structure module with each preset geological structure module, and obtains a set of sample geological structure modules corresponding to each preset geological structure module. For example, the server may divide all sample geological structure modules matching "rock slope" into one set, and divide all samples matching "soil slope" into another set. For each preset geological structure module, the server will adjust the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module according to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module. For example, if the sample geological structure module in the "rock slope" set often appears with the "potential sliding surface", the server may increase the correlation between the "rock slope" and the "potential sliding surface" in the network embedding model. The server will use the optimized dam geological structure classification model and the adjusted network embedding model corresponding to each preset geological structure module to repeatedly perform the steps of extracting features from the sample geological structure module and each undetermined geological structure type through the dam geological structure classification model. This process will be iterated continuously until the error parameter of the dam geological structure classification model reaches the preset error parameter threshold, thereby obtaining the trained dam geological structure classification model. By executing the above steps, the server has comprehensively trained and optimized the dam geological structure classification model, so that it can more accurately extract and identify the characteristics of the dam geological structure, providing a reliable basis for subsequent matching analysis and stability verification.

[0083] In the embodiment of the present invention, according to the matching situation of the sample geological structure module with each preset geological structure module, each sample geological structure module is divided to obtain a set of sample geological structure modules corresponding to each preset geological structure module, which can be implemented through the following examples.

[0084] According to the matching condition between the sample geological structure module and each preset geological structure module, determining the target geological structure module category to which the sample geological structure module is classified from each preset geological structure module;

[0085] According to the target geological structure module category to which each sample geological structure module is classified, the type of each sample geological structure module is identified to obtain a sample geological structure module set corresponding to each preset geological structure module.

[0086] In an embodiment of the present invention, exemplarily, the server determines the target geological structure module category to which the sample geological structure module is classified from each preset geological structure module according to the matching of the sample geological structure module with each preset geological structure module. For example, the server may have a sample geological structure module, whose features match the two preset geological structure modules of "rock slope" and "soil slope" to a certain extent. However, after calculation, the matching confidence of the sample with "rock slope" is higher, so the server determines that the sample geological structure module is classified into the target geological structure module category of "rock slope". According to the target geological structure module category to which each sample geological structure module is classified, the server performs type identification on each sample geological structure module. The server will classify all sample geological structure modules classified into the same target geological structure module category into the same set. Continuing with the above example, the server will classify all sample geological structure modules classified into "rock slope" into one set, and this set is the sample geological structure module set corresponding to "rock slope". By executing the above steps, the server successfully divides the sample geological structure modules according to the matching between the sample geological structure modules and the preset geological structure modules, and obtains the corresponding sample geological structure module set for each preset geological structure module. This provides strong data support for subsequent model optimization and feature extraction.

[0087] In an embodiment of the present invention, according to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module is adjusted to obtain the adjusted network embedding model corresponding to the preset geological structure module, which can be implemented through the following examples.

[0088] According to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, a sample geological structure module subset of each undetermined geological structure type is determined in the sample geological structure module set corresponding to the preset geological structure module, and each sample geological structure module subset of the undetermined geological structure type includes a sample geological structure module corresponding to the corresponding undetermined geological structure type;

[0089] For each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type to determine that when the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes target matching content of the remaining undetermined geological structure types, and the target matching content indicates the correlation between the undetermined geological structure type and the remaining undetermined geological structure types;

[0090] According to the target matching content, the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module is adjusted to obtain the adjusted network embedding model corresponding to the preset geological structure module.

[0091] In an embodiment of the present invention, exemplarily, the server determines a subset of sample geological structure modules for each undetermined geological structure type in the set of sample geological structure modules corresponding to the preset geological structure module according to the actual geological structure type of the sample geological structure modules in the set. For example, for the preset geological structure module "rock slope", the server may determine a subset of sample geological structure modules for the undetermined geological structure type "potential sliding surface" in its corresponding set of sample geological structure modules, and the subset contains all sample geological structure modules containing "potential sliding surface" in the actual geological structure types. For each undetermined geological structure type, the server performs quantitative analysis on the sample geological structure modules in the subset of sample geological structure modules of the type. The goal of the server analysis is to determine that in the preset geological structure module, when the sample geological structure module includes a certain undetermined geological structure type, the sample geological structure module also includes the target matching content of the remaining undetermined geological structure types. Continuing with the above example, the server will analyze whether the samples in the preset geological structure module "rock slope" also include other undetermined geological structure types such as "joints and fissures" when the sample geological structure module includes "potential sliding surface" and determine the correlation between them. Based on the target matching content obtained by quantitative analysis, the server adjusts the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module. For example, if the server finds that "potential sliding surface" and "joints and fissures" often appear together in "rock slope", it will increase the correlation of these two undetermined geological structure types in the network embedding model. By executing the above steps, the server successfully adjusted the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module according to the actual type of the sample geological structure module, and obtained a more accurate adjusted network embedding model. This provides strong support for the subsequent classification of dam geological structure and stability verification.

[0092] In an embodiment of the present invention, for each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type to determine that when the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes the target matching content of the remaining undetermined geological structure types, which can be implemented through the following examples.

[0093] For each undetermined geological structure type, counting a first number of sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type;

[0094] Determining, from the subset of sample geological structure modules of the undetermined geological structure type, a second number of sample geological structure modules that also include the remaining undetermined geological structure types;

[0095] When it is determined according to the first number and the second number that the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes target matching content of the remaining undetermined geological structure types.

[0096] In the embodiment of the present invention, exemplarily, for each undetermined geological structure type, the server counts the first number of sample geological structure modules in the sample geological structure module subset of the type. For example, for the undetermined geological structure type "potential sliding surface", the server counts the first number of all sample geological structure modules containing "potential sliding surface" in the sample geological structure module set corresponding to the preset geological structure module "rock slope", which is assumed to be 100. In the sample geological structure module subset of each undetermined geological structure type, the server further determines the second number of sample geological structure modules that also include the remaining undetermined geological structure types. Continuing with the above example, the server will determine the second number of sample geological structure modules that also include other undetermined geological structure types such as "joints and fissures" among the 100 sample geological structure modules containing "potential sliding surface", which is assumed to be 60. Based on the first number and the second number, the server determines that in the preset geological structure module, when the sample geological structure module includes a certain undetermined geological structure type, the sample geological structure module also includes the target matching content of the remaining undetermined geological structure types. In this example, the server calculates that in "rock slope", when the sample geological structure module includes "potential sliding surface", there is a 60% (60 / 100) probability that these samples also include "joints and fissures". This ratio is the target matching content of "potential sliding surface" and "joints and fissures" in "rock slope", which indicates the correlation between these two undetermined geological structure types. By executing the above steps, the server successfully quantitatively analyzes the sample geological structure module of each undetermined geological structure type, and determines that when the sample geological structure module includes a certain undetermined geological structure type in the preset geological structure module, it also includes the target matching content of the remaining undetermined geological structure types. This provides accurate data support for the adjustment of the correlation of each undetermined geological structure type in the subsequent network embedding model.

[0097] In the embodiment of the present invention, the acquisition of the sample data set and the network embedding model corresponding to each preset geological structure module can be implemented through the following examples.

[0098] Acquire a sample data set, and determine a set of sample geological structure modules for each undetermined geological structure type in the sample data set according to the actual geological structure type of the sample geological structure module in the sample data set, wherein each set of sample geological structure modules for the undetermined geological structure type includes the sample geological structure modules of the corresponding undetermined geological structure type in the sample data set;

[0099] For each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the set of sample geological structure modules of the undetermined geological structure type to determine that when the sample geological structure module includes the undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types, and the matching content indicates the correlation between the undetermined geological structure type and the remaining undetermined geological structure types;

[0100] According to the matching content, the correlation between type entities corresponding to the undetermined geological structure type in the network embedding model of each preset geological structure module is set to obtain the network embedding model corresponding to each preset geological structure module after the setting.

[0101] In an embodiment of the present invention, exemplarily, the server first obtains a sample data set, which includes multiple sample geological structure modules, each of which has an actual geological structure type label. The server determines a set of sample geological structure modules of each undetermined geological structure type in the sample data set according to the actual geological structure type of the sample geological structure module in the sample data set. For example, the server may determine all sample geological structure modules including the undetermined geological structure type of "potential sliding surface" to form a set of sample geological structure modules of "potential sliding surface". For each undetermined geological structure type, the server performs a quantitative analysis on the sample geological structure modules in the set of sample geological structure modules of this type. The goal of the server analysis is to determine that when the sample geological structure module includes a certain undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types. For example, the server will analyze the matching content of other undetermined geological structure types such as "joints and fissures" in the sample geological structure module including the "potential sliding surface". According to the matching content obtained by the quantitative analysis, the server sets the correlation between the type entities corresponding to the undetermined geological structure type in the network embedding model of each preset geological structure module. For example, if the server finds that a high proportion of the sample geological structure modules containing "potential sliding surface" also contain "joints and fissures", it will add the correlation between the two types of entities "potential sliding surface" and "joints and fissures" in the network embedding model of the preset geological structure module. By executing the above steps, the server successfully obtains the sample data set, and sets the correlation between the type entities corresponding to the undetermined geological structure type in the network embedding model of each preset geological structure module according to the information in the sample data set, and obtains the set network embedding model. This provides a strong model support for the subsequent classification of dam geological structures and stability verification.

[0102] In an embodiment of the present invention, for each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the sample geological structure module set of the undetermined geological structure type to determine that when the sample geological structure module includes the undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types, which can be implemented through the following examples.

[0103] For each undetermined geological structure type, counting a first number of sample geological structure modules in the sample geological structure module set of the undetermined geological structure type;

[0104] Determining, from the set of sample geological structure modules of the undetermined geological structure type, a second number of sample geological structure modules that also include the remaining undetermined geological structure types;

[0105] According to the first number and the second number, it is determined that when the sample geological structure module includes the undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types.

[0106] In the embodiment of the present invention, exemplarily, for each undetermined geological structure type, the server counts the first number of sample geological structure modules in the sample geological structure module set of the type. For example, for the undetermined geological structure type "potential sliding surface", the server counts the first number of all sample geological structure modules containing "potential sliding surface" in the sample data set, which is assumed to be 200. In the sample geological structure module set of each undetermined geological structure type, the server further determines the second number of sample geological structure modules that also include the remaining undetermined geological structure types. Continuing with the above example, the server will determine the second number of sample geological structure modules that also include other undetermined geological structure types such as "joints and fissures" among the 200 sample geological structure modules containing "potential sliding surface", which is assumed to be 120. Based on the first number and the second number, the server determines that when the sample geological structure module includes a certain undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types. In this example, the server calculates that among the sample geological structure modules containing "potential sliding surfaces", there is a 60% (120 / 200) probability that these samples also contain "joints and fissures". This ratio is the matching content between "potential sliding surfaces" and "joints and fissures", which indicates the correlation between these two undetermined geological structure types. By executing the above steps, the server successfully quantitatively analyzes the sample geological structure modules of each undetermined geological structure type, and determines that when the sample geological structure module includes a certain undetermined geological structure type, it also includes the matching content of the remaining undetermined geological structure types. This provides accurate data support for the setting of the correlation between the corresponding type entities of each undetermined geological structure type in the subsequent network embedding model.

[0107] In the embodiment of the present invention, the error parameter of the dam geological structure classification model is determined according to the matching confidence and the actual geological structure type of the sample geological structure module, which can be implemented through the following example.

[0108] Calculating a label error parameter according to the matching confidence and the actual geological structure type of the sample geological structure module;

[0109] Calculating the conditional entropy of the sample geological structure module according to the matching condition between the sample geological structure module and each preset geological structure module;

[0110] According to the matching condition between the sample geological structure module and each preset geological structure module, the target geological structure module category to which the sample geological structure module is classified is determined, and according to the target geological structure module category to which each sample geological structure module is classified, the corresponding sample quantity information of each preset geological structure module is determined;

[0111] Calculate the category error parameter according to the conditional entropy of each sample geological structure module and the corresponding sample number information of each preset geological structure module;

[0112] The error parameters of the dam body geological structure classification model are determined according to the label error parameters and the category error parameters.

[0113] In an embodiment of the present invention, exemplarily, the server calculates the label error parameter based on the matching confidence and the actual geological structure type of the sample geological structure module. For example, for a sample geological structure module, its actual geological structure type is "rock slope", but the type with the highest matching confidence given by the model is "soil slope", then the server will calculate the label error parameter based on this mismatch. The server calculates the conditional entropy of the sample geological structure module based on the matching of the sample geological structure module with each preset geological structure module. For example, when a sample geological structure module is matched, the matching confidences for the three preset categories of "rock slope", "soil slope" and "mixed slope" are 0.3, 0.4 and 0.3 respectively, and the server will calculate the conditional entropy of the sample based on these matching confidences. The server determines the target geological structure module category to which the sample geological structure module is classified based on the matching of the sample geological structure module with each preset geological structure module. Then, the server determines the corresponding sample number information of each preset geological structure module according to the target geological structure module category classified by each sample geological structure module. For example, among 200 samples, 100 are classified as "rock slope", 50 are classified as "soil slope", and 50 are classified as "mixed slope", then the server will record these sample number information. The server calculates the category error parameter based on the conditional entropy of each sample geological structure module and the corresponding sample number information of each preset geological structure module. The category error parameter reflects the overall uncertainty of the model when classifying samples into different preset categories. Finally, the server determines the error parameter of the dam geological structure classification model based on the label error parameter and the category error parameter. This error parameter combines the accuracy of the model in both label prediction and category classification, and provides an important quantitative indicator for subsequent model optimization and verification. By executing the above steps, the server successfully determines the error parameter of the dam geological structure classification model, which can fully reflect the accuracy and stability of the model in the classification task.

[0114] In an embodiment of the present invention, the dam body geological structure classification model is used to perform feature extraction processing on the sample geological structure module and each undetermined geological structure type, respectively, to obtain the sample overall feature vector of the sample geological structure module and the sample category feature vector of each undetermined geological structure type, which can be implemented through the following examples.

[0115] By using the dam body geological structure classification model, a feature extraction operation is performed on the sample geological structure module to obtain a sample module description feature of the sample geological structure module;

[0116] Through the dam body geological structure classification model, feature extraction operations are performed on each undetermined geological structure type to obtain sample geological structure type features of each undetermined geological structure type; according to the sample geological structure type features of each undetermined geological structure type and the learnable feature representation, a feature set is constructed;

[0117] Performing significance detection on the sample module description features and the feature set to obtain the sample overall feature vector of the sample geological structure module and the sample category feature vector of each undetermined geological structure type;

[0118] The method optimizes the model parameters in the dam body geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module to obtain a trained dam body geological structure classification model, including:

[0119] According to the matching confidence and the actual geological structure type of the sample geological structure module, the model parameters in the dam geological structure classification model and the learnable feature representation are optimized to obtain a trained dam geological structure classification model.

[0120] In an embodiment of the present invention, exemplarily, the server performs a feature extraction operation on the sample geological structure module through the dam geological structure classification model. For example, the server may have a sample geological structure module containing multiple geological structure features, such as rock layer inclination, joint fissure distribution, etc. Through the classification model, the server can extract the numerical representation of these features to form the sample module description features. The server performs a feature extraction operation on each undetermined geological structure type through the dam geological structure classification model. For example, for the undetermined geological structure type "potential sliding surface", the server extracts features related to the type, such as the inclination of the sliding surface, the roughness of the sliding surface, etc., to form the sample geological structure type features. The server constructs a feature set based on the sample geological structure type features of each undetermined geological structure type and the learnable learnable feature representations. These learnable feature representations are parameters that can be adjusted by the model during the training process, which are used to better capture and represent the features of the geological structure type. The server performs a saliency test on the sample module description features and the feature set. The purpose of the saliency test is to determine which features are important for distinguishing different geological structure types. Through detection, the server can obtain the sample overall feature vector of the sample geological structure module and the sample category feature vector of each undetermined geological structure type. The server optimizes the model parameters and learnable feature representation in the dam geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module. For example, if the model has a low matching confidence when classifying a sample geological structure module as a "potential sliding surface" and the actual type of the sample is a "rock slope", the server will adjust the model parameters and learnable feature representation to improve the model's classification accuracy for similar samples in the future. Through continuous optimization, the server finally obtained a trained dam geological structure classification model, which can more accurately classify and evaluate the stability of the dam according to the geological structure characteristics of the dam. By executing the above steps, the server successfully performed feature extraction processing and optimized the dam geological structure classification model, providing a more accurate classification and evaluation tool for subsequent dam stability verification.

[0121] The embodiment of the present invention provides a computer device 100, which includes a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device 100 executes the aforementioned dam geological structure stability verification method based on digital twins. Figure 2 As shown, Figure 2 This is a structural block diagram of a computer device 100 provided in an embodiment of the present invention. The computer device 100 includes a memory 111 , a processor 112 , and a communication unit 113 .

[0122] To achieve data transmission or interaction, the memory 111, the processor 112, and the communication unit 113 are electrically connected to each other directly or indirectly. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0123] For illustrative purposes, the foregoing description is made with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the present disclosure to the precise form disclosed. Numerous modifications and variations are possible in accordance with the above teachings. These embodiments are selected and described in order to best illustrate the principles of the present disclosure and its practical application, so that those skilled in the art can best utilize the present disclosure and utilize various embodiments with different modifications to suit the intended specific application.

Claims

1. A dam geological structure stability verification method based on digital twins, characterized in that: include: Acquire remote sensing images and multiple geological exploration data for a target dam, and construct a three-dimensional geological model of the target dam corresponding to the target dam; The three-dimensional geological model of the target dam is transmitted to a preset digital twin platform, wherein the three-dimensional geological model of the target dam is composed of a plurality of geological structure modules; Acquire preset simulated extreme condition data, and simulate the three-dimensional geological model of the target dam based on the preset simulated extreme condition data on the preset digital twin platform to obtain a three-dimensional geological model of the dam after simulation; Performing stability analysis on the simulated three-dimensional geological model of the dam to obtain the geological structure type corresponding to each geological structure module; According to the multiple geological structure types, a verification result of the geological structure stability of the target dam is obtained; The stability analysis of the three-dimensional geological model of the dam after the simulation is performed to obtain the geological structure type corresponding to each geological structure module, including: Acquire a target geological structure module and a plurality of undetermined geological structure types, wherein the target geological structure module is any one of the plurality of geological structure modules; Performing feature extraction processing on the target geological structure module and each undetermined geological structure type respectively to obtain an overall feature vector of the target geological structure module and a category feature vector of each undetermined geological structure type; According to the overall feature vector, predicting the matching situation of the target geological structure module with each preset geological structure module; the matching situation includes the matching content of the target geological structure module with each preset geological structure module; Acquire a network embedding model corresponding to each preset geological structure module, wherein the network embedding model indicates the correlation between the undetermined geological structure types in the preset geological structure module; the network embedding model includes a type entity corresponding to each undetermined geological structure type, and a connection relationship between the type entities, wherein the connection relationship indicates the correlation between two connected type entities; For each undetermined geological structure type, determining a corresponding target type entity of the undetermined geological structure type in a corresponding network embedding model of each preset geological structure module; For each network embedding model corresponding to the preset geological structure module, determining the relevant feature vector of the target type entity in the preset geological structure module according to the correlation between the target type entity and the remaining type entities in the network embedding model; For each preset geological structure module, the matching content of the target geological structure module with the preset geological structure module is integrated with the relevant feature vector of the target type entity in the preset geological structure module to obtain the original integrated relevant features of the undetermined geological structure type in the preset geological structure module; Determining an integrated related feature vector of the undetermined geological structure type according to the original integrated related features of the undetermined geological structure type in each preset geological structure module; According to the integrated related feature vector, the category feature vector of the undetermined geological structure type is adjusted to obtain the undetermined category feature vector corresponding to the undetermined geological structure type; For each undetermined geological structure type, integrating the category feature vector of the undetermined geological structure type and the undetermined category feature vector to obtain a target category feature vector of the undetermined geological structure type; Determining the matching confidence of each undetermined geological structure type with the target geological structure module according to the target category feature vector of each undetermined geological structure type; According to the matching confidence, the target geological structure type corresponding to the target geological structure module is determined from each of the undetermined geological structure types.

2. The method according to claim 1, characterized in that The step of performing feature extraction processing on the target geological structure module and each undetermined geological structure type to obtain an overall feature vector of the target geological structure module and a category feature vector of each undetermined geological structure type includes: Performing a feature extraction operation on a parameter description of the target geological structure module in at least one data dimension to obtain a module description feature of the target geological structure module in at least one data dimension; Perform feature extraction operations on each undetermined geological structure type respectively to obtain geological structure type features of each undetermined geological structure type; Constructing a feature set according to the geological structure type characteristics of each undetermined geological structure type and the learnable feature representation; Determine a candidate module description feature under a target data dimension from the module description feature under the at least one data dimension; Performing significance detection on the module description features and the feature set under the target data dimension to obtain a significance feature vector of the target geological structure module and a significance feature vector of each undetermined geological structure type; Determining an iterative feature set according to the significant feature vector of the target geological structure module and the significant feature vector of each undetermined geological structure type; Repeat the step of determining the candidate module description features under the target data dimension from the module description features under the at least one data dimension until the module description features under all data dimensions are executed, and obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type according to the current iterative feature set.

3. The method according to claim 1, characterized in that The step of performing feature extraction processing on the target geological structure module and each undetermined geological structure type to obtain an overall feature vector of the target geological structure module and a category feature vector of each undetermined geological structure type includes: Through the dam body geological structure classification model, feature extraction processing is performed on the target geological structure module and each undetermined geological structure type respectively to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type.

4. The method according to claim 3, characterized in that Before the target geological structure module and each undetermined geological structure type are subjected to feature extraction processing by using the dam geological structure classification model to obtain the overall feature vector of the target geological structure module and the category feature vector of each undetermined geological structure type, the method further includes: Acquire a sample data set, and determine a set of sample geological structure modules for each undetermined geological structure type in the sample data set according to the actual geological structure type of the sample geological structure module in the sample data set, wherein each set of sample geological structure modules for the undetermined geological structure type includes the sample geological structure modules of the corresponding undetermined geological structure type in the sample data set; For each undetermined geological structure type, counting a first number of sample geological structure modules in the sample geological structure module set of the undetermined geological structure type; Determining, from the set of sample geological structure modules of the undetermined geological structure type, a second number of sample geological structure modules that also include the remaining undetermined geological structure types; According to the first number and the second number, determining that when the sample geological structure module includes the undetermined geological structure type, the sample geological structure module also includes matching content of the remaining undetermined geological structure types, and the matching content indicates the correlation between the undetermined geological structure type and the remaining undetermined geological structure types; According to the matching content, the correlation between type entities corresponding to the undetermined geological structure type in the network embedding model of each preset geological structure module is set to obtain a network embedding model corresponding to each preset geological structure module after the setting, wherein the sample data set includes multiple sample geological structure modules and actual geological structure types of the sample geological structure modules; Through the dam body geological structure classification model, feature extraction processing is performed on the sample geological structure module and each undetermined geological structure type respectively to obtain a sample overall feature vector of the sample geological structure module and a sample category feature vector of each undetermined geological structure type; According to the overall feature vector of the sample, predicting the matching status of the sample geological structure module with each preset geological structure module; According to the matching situation and the network embedding model corresponding to each preset geological structure module, node vector feature processing is performed on the sample category feature vector of each undetermined geological structure type to obtain the undetermined sample category feature vector corresponding to each undetermined geological structure type; Determine the matching confidence of each undetermined geological structure type with the sample geological structure module according to the undetermined sample category feature vector corresponding to each undetermined geological structure type; According to the matching confidence and the actual geological structure type of the sample geological structure module, the model parameters in the dam geological structure classification model are optimized to obtain a trained dam geological structure classification model.

5. The method according to claim 4, characterized in that The method optimizes the model parameters in the dam body geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module to obtain a trained dam body geological structure classification model, including: Calculating a label error parameter according to the matching confidence and the actual geological structure type of the sample geological structure module; Calculating the conditional entropy of the sample geological structure module according to the matching condition between the sample geological structure module and each preset geological structure module; According to the matching condition between the sample geological structure module and each preset geological structure module, the target geological structure module category to which the sample geological structure module is classified is determined, and according to the target geological structure module category to which each sample geological structure module is classified, the corresponding sample quantity information of each preset geological structure module is determined; Calculate the category error parameter according to the conditional entropy of each sample geological structure module and the corresponding sample number information of each preset geological structure module; Determining error parameters of the dam body geological structure classification model according to the label error parameters and the category error parameters; When the error parameter does not reach a preset error parameter threshold, optimizing the model parameters in the dam body geological structure classification model according to the error parameter; According to the matching condition between the sample geological structure module and each preset geological structure module, determining the target geological structure module category to which the sample geological structure module is classified from each preset geological structure module; According to the target geological structure module category to which each sample geological structure module is classified, the type of each sample geological structure module is identified to obtain a sample geological structure module set corresponding to each preset geological structure module; For each preset geological structure module, according to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module is adjusted to obtain the adjusted network embedding model corresponding to the preset geological structure module; According to the optimized dam body geological structure classification model and the adjusted network embedding model corresponding to each preset geological structure module, the steps of performing feature extraction processing on the sample geological structure module and each undetermined geological structure type through the dam body geological structure classification model are repeated until the error parameter of the dam body geological structure classification model reaches the preset error parameter threshold, thereby obtaining a trained dam body geological structure classification model.

6. The method according to claim 5, characterized in that The method further comprises: adjusting the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module according to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, and obtaining the adjusted network embedding model corresponding to the preset geological structure module, including: According to the actual geological structure type of the sample geological structure module in the sample geological structure module set corresponding to the preset geological structure module, a sample geological structure module subset of each undetermined geological structure type is determined in the sample geological structure module set corresponding to the preset geological structure module, and each sample geological structure module subset of the undetermined geological structure type includes a sample geological structure module corresponding to the corresponding undetermined geological structure type; For each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type to determine that when the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes target matching content of the remaining undetermined geological structure types, and the target matching content indicates the correlation between the undetermined geological structure type and the remaining undetermined geological structure types; According to the target matching content, the correlation of each undetermined geological structure type in the network embedding model corresponding to the preset geological structure module is adjusted to obtain the adjusted network embedding model corresponding to the preset geological structure module.

7. The method according to claim 6, characterized in that For each undetermined geological structure type, a quantitative analysis is performed on the sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type to determine that when the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes target matching content of the remaining undetermined geological structure types, including: For each undetermined geological structure type, counting a first number of sample geological structure modules in the sample geological structure module subset of the undetermined geological structure type; Determining, from the subset of sample geological structure modules of the undetermined geological structure type, a second number of sample geological structure modules that also include the remaining undetermined geological structure types; When it is determined according to the first number and the second number that the sample geological structure module in the preset geological structure module includes the undetermined geological structure type, the sample geological structure module also includes target matching content of the remaining undetermined geological structure types.

8. The method according to claim 4, characterized in that The dam body geological structure classification model is used to perform feature extraction processing on the sample geological structure module and each undetermined geological structure type, respectively, to obtain a sample overall feature vector of the sample geological structure module and a sample category feature vector of each undetermined geological structure type, including: By using the dam body geological structure classification model, a feature extraction operation is performed on the sample geological structure module to obtain a sample module description feature of the sample geological structure module; Through the dam body geological structure classification model, feature extraction operations are performed on each undetermined geological structure type to obtain sample geological structure type features of each undetermined geological structure type; according to the sample geological structure type features of each undetermined geological structure type and the learnable feature representation, a feature set is constructed; Performing significance detection on the sample module description features and the feature set to obtain the sample overall feature vector of the sample geological structure module and the sample category feature vector of each undetermined geological structure type; The method optimizes the model parameters in the dam body geological structure classification model according to the matching confidence and the actual geological structure type of the sample geological structure module to obtain a trained dam body geological structure classification model, including: According to the matching confidence and the actual geological structure type of the sample geological structure module, the model parameters in the dam geological structure classification model and the learnable feature representation are optimized to obtain a trained dam geological structure classification model.

9. A server system, characterized in that: The method comprises a server, wherein the server is used to execute the method described in any one of claims 1 to 8.

Images