Electric power tower structure design and simulation system

Abstract

This invention relates to the field of digital twin and database data processing technology, and discloses a power transmission tower structure design and simulation system. The system decomposes historical power transmission tower digital twin models, extracts structural dimension parameters and material property parameters of components as entity nodes, and extracts extreme value features and gradient features of the finite element stress distribution matrix and deformation displacement matrix as attribute nodes. It then establishes directed edges to generate component-level feature graphs and stores them in a graph database. Upon receiving the structural parameters of the component to be designed, it converts them into feature vectors, executes a graph retrieval command based on subgraph isomorphism in the graph database, and directly outputs the matching attribute node data as the estimated simulation result. This invention constructs a graph association structure between structural parameters and historical simulation result features, enabling the design phase of new power transmission towers to directly call historical simulation data of standard components, eliminating the need to re-execute the finite element solution calculation process and reducing processor computing power consumption.

Description

Technical Field

[0001] This invention relates to the field of information retrieval and database data processing technology, and discloses a power transmission tower structure design and simulation system. Background Technology

[0002] Existing power transmission tower structural design and simulation systems typically employ a relational database in conjunction with a file system for persistent data storage when processing digital twin model data. Specifically, the system creates independent data tables for the power transmission tower's 3D geometric model files, material property parameter files, and finite element mesh files, categorized by project name, and allocates corresponding physical storage paths on the server disk. After completing structural mechanics simulation calculations for a specific load condition, the system converts the generated stress distribution matrix data and deformation displacement matrix data into result files, which are also archived according to project dimensions. When undertaking new power transmission tower design tasks, the system searches historical model files by traversing the project directory in the database, requiring manual comparison of the geometric dimensions and material parameters of tower components from different projects to find similar structures.

[0003] Based on the aforementioned existing technologies, when designing new power transmission towers, even if the tower head, body, or leg components to be designed are completely identical to those stored in historical engineering projects in terms of structural dimensions and material properties, conventional file systems and relational database architectures cannot establish a direct correlation between the structural parameters of the components and historical simulation results at the underlying physical data level. The system can only reload the matched historical model files into the simulation engine and repeat the entire process from geometry cleanup and mesh generation to iterative solution of the equations. This data silo-like storage and retrieval mechanism prevents historical simulation experience data from being directly reused across projects, resulting in excessive computational resource consumption due to repeated finite element method calculations during the design phase of new power transmission towers. Summary of the Invention

[0004] The purpose of this invention is to provide a power transmission tower structure design and simulation system, which can effectively solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The power tower structure design and simulation system includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it performs the following steps: receiving a set of historical power tower digital twin models, performing spatial analysis and topological decomposition on each digital twin model in the set of historical power tower digital twin models, and generating tower head components, tower body components, and tower leg components.

[0007] Extract the structural dimension parameters and material property parameters of each component, and map the structural dimension parameters and material property parameters to entity nodes in the component-level feature map;

[0008] Extract the finite element stress distribution matrix and finite element deformation displacement matrix of each component under the preset load conditions, extract extreme value features and gradient features from the finite element stress distribution matrix and finite element deformation displacement matrix, and map the extreme value features and gradient features to attribute nodes in the component-level feature map.

[0009] Establish directed edges between entity nodes and attribute nodes, generate component-level feature graphs, and store them in a graph database;

[0010] The system receives the structural dimensions and material properties of the power tower components to be designed, executes a graph retrieval command based on subgraph isomorphism in the graph database, and outputs attribute node data that matches the power tower components to be designed.

[0011] Preferably, spatial analysis and topological decomposition are performed on each digital twin model in the set of historical power tower digital twin models, including: reading the three-dimensional geometric patch data and finite element mesh node data of each digital twin model, and constructing a spatial bounding box hierarchical tree containing all geometric patch data and finite element mesh node data;

[0012] Traverse the nodes in the spatial bounding box hierarchy tree and extract the connection stiffness matrix between adjacent finite element mesh nodes;

[0013] Based on the values ​​of the main diagonal elements in the connection stiffness matrix, adjacent finite element mesh nodes with connection stiffness matrices greater than a preset stiffness threshold are assigned to the same initial connected domain.

[0014] Calculate the spatial geometric center coordinates of each initial connected domain, merge adjacent initial connected domains whose spatial geometric center coordinates are less than a preset distance threshold and have a shared geometric patch, and label the merged connected domains as tower head component, tower body component and tower leg component respectively.

[0015] Preferably, mapping structural dimension parameters and material property parameters to entity nodes in a component-level feature map includes: parsing the member cross-sectional length, cross-sectional width, wall thickness, and node spacing in the structural dimension parameters, and converting the member cross-sectional length, cross-sectional width, wall thickness, and node spacing into discrete numerical sequences according to a preset quantization interval;

[0016] The yield strength, elastic modulus, and density of the material properties are analyzed, and the yield strength, elastic modulus, and density are appended to the tail of the discrete numerical sequence to generate a multidimensional structural feature vector.

[0017] Create entity node objects corresponding to the tower head component, tower body component, and tower leg component in the graph database. Store the multidimensional structural feature vector as the first attribute field of the entity node object and assign a globally unique identifier to the entity node object.

[0018] Preferably, the extreme value features and gradient features are extracted from the finite element stress distribution matrix and the finite element deformation displacement matrix, including: dividing the finite element stress distribution matrix into multiple cubic mesh clusters according to spatial coordinates, calculating the maximum, minimum and average values ​​of stress values ​​in each cubic mesh cluster, and combining the maximum, minimum and average values ​​into a stress extreme value feature vector.

[0019] The stress change rate of each cubic mesh cluster in the three orthogonal directions in space is calculated respectively. The stress change rate in the three orthogonal directions is arranged into a three-dimensional stress gradient tensor. The three-dimensional stress gradient tensor is expanded into a one-dimensional vector and used as the stress gradient feature.

[0020] The finite element deformation displacement matrix is ​​processed using the same method as the stress extremum eigenvector and stress gradient feature to generate the deformation extremum eigenvector and deformation gradient feature.

[0021] Preferably, establishing directed edges between entity nodes and attribute nodes includes: extracting globally unique identifiers from entity nodes and component category labels from attribute nodes;

[0022] Establish the first directed edge in the direction from the entity node to the attribute node, and mark the edge type of the first directed edge as physical response association;

[0023] Extract the spatial assembly coordinates of the corresponding components of the entity nodes in the digital twin model, and calculate the spatial assembly coordinate difference between the corresponding components of adjacent entity nodes;

[0024] Establish a second directed edge between two adjacent entity nodes, mark the edge type of the second directed edge as spatial adjacency association, and store the spatial assembly coordinate difference as the edge weight attribute of the second directed edge.

[0025] Traverse all entity nodes and attribute nodes in the component-level feature graph, and repeat the directed edge building step until the component-level feature graph is constructed.

[0026] Preferably, the graph retrieval instruction based on subgraph isomorphism is executed in the graph database, including: converting the structural dimension parameters and material property parameters of the power tower component to be designed into a multidimensional structural feature vector to be retrieved;

[0027] Calculate the cosine similarity between the multidimensional structural feature vector to be retrieved and the multidimensional structural feature vectors of each entity node in the graph database, and extract entity nodes whose cosine similarity is greater than a preset similarity threshold to form a candidate entity node set;

[0028] Starting from each candidate entity node in the candidate entity node set, extract connected attribute nodes along directed edges to construct multiple candidate subgraphs;

[0029] The components of the power tower to be designed are converted into subgraphs to be matched, and the subgraph isomorphic matching degree between the subgraph to be matched and multiple candidate subgraphs is calculated.

[0030] Preferably, based on the values ​​of the main diagonal elements in the connection stiffness matrix, adjacent finite element mesh nodes with a connection stiffness matrix greater than a preset stiffness threshold are divided into the same initial connected domain, including: extracting the heat transfer coefficient at non-zero, non-diagonal positions in the connection stiffness matrix;

[0031] Take any unassigned finite element mesh node as the starting node and add the starting node to the current traversal queue;

[0032] Extract the first finite element mesh node in the current traversal queue, and query the adjacent finite element mesh nodes that are directly connected to the first finite element mesh node;

[0033] Determine whether the heat transfer coefficient between the first finite element mesh node and its adjacent finite element mesh nodes is greater than a preset stiffness threshold. If it is greater than the preset stiffness threshold, then divide the adjacent finite element mesh nodes into the same initial connected domain as the starting node and add the adjacent finite element mesh nodes to the current traversal queue until the current traversal queue is empty.

[0034] Preferably, the finite element stress distribution matrix is ​​divided into multiple cubic mesh clusters according to spatial coordinates, including: obtaining the three-dimensional spatial coordinates of all finite element mesh nodes in the finite element stress distribution matrix, and calculating the maximum and minimum boundary values ​​of all three-dimensional spatial coordinates in the three coordinate axis directions;

[0035] Starting from the minimum boundary value, the three coordinate axes are divided at equal intervals according to the preset fixed side length to generate multiple virtual cube meshes covering the entire space of the finite element stress distribution matrix.

[0036] Finite element mesh nodes whose three-dimensional spatial coordinates fall within the same virtual cube mesh are divided into the same cube mesh cluster;

[0037] When there are no finite element mesh nodes within the virtual cube mesh, delete the virtual cube mesh and record the relative spatial position index of adjacent cube mesh clusters.

[0038] Preferably, calculating the spatial assembly coordinate difference between components corresponding to adjacent entity nodes includes: extracting all finite element mesh nodes contained in each component corresponding to the adjacent entity nodes, calculating the average three-dimensional spatial coordinates of all finite element mesh nodes in the tower head component, tower body component, and tower leg component respectively, and setting the average three-dimensional spatial coordinates as the spatial assembly coordinates of the components.

[0039] Calculate the absolute values ​​of the spatial assembly coordinate differences between two adjacent components in the X-axis direction, Y-axis direction, and Z-axis direction, respectively.

[0040] The absolute values ​​of the coordinate differences along the X-axis, Y-axis, and Z-axis are concatenated to form a three-dimensional spatial distance vector, which is then used as the edge weight attribute of the second directed edge between adjacent entity nodes.

[0041] Preferably, calculating the subgraph isomorphic matching degree between the subgraph to be matched and multiple candidate subgraphs includes: extracting the set of entity nodes and the set of directed edges in the subgraph to be matched and the candidate subgraphs;

[0042] Traverse the set of entity nodes, calculate the Euclidean distance between the multidimensional structural feature vector of the entity node in the subgraph to be matched and the multidimensional structural feature vector of the corresponding entity node in the candidate subgraph, and sum the reciprocals of the Euclidean distances to generate the node matching degree factor.

[0043] Traverse the set of directed edges, calculate the vector difference magnitude between the edge weight attribute of the second directed edge in the subgraph to be matched and the edge weight attribute of the corresponding second directed edge in the candidate subgraph, and sum the reciprocals of the vector difference magnitudes to generate the edge matching degree factor.

[0044] Multiply the node matching degree factor by the edge matching degree factor to generate the subgraph isomorphic matching degree value.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. This invention decomposes a digital twin model of a historical power transmission tower into tower head, tower body, and tower leg components. Structural dimensional parameters and material property parameters are extracted as entity nodes, and extreme value and gradient features of the finite element stress distribution matrix and deformation displacement matrix are extracted as attribute nodes. A component-level feature graph is constructed using directed edges and stored in a graph database. When receiving a component to be designed, the parameters are converted into feature vectors, and a subgraph isomorphic search is performed in the graph database. The matching attribute node data is directly output as the estimated simulation result. This technique abstracts physical properties and simulation result features into a graph-related structure, changing the retrieval logic of storing files in isolation by project. This allows the design phase of new power transmission towers to directly call historical simulation results of standard structural components, eliminating the need to re-execute the mesh generation and finite element solution calculations, and reducing the computational load and processor power consumption for finite element solutions of standard structural components.

[0047] 2. In constructing the component-level feature map, a spatial bounding box hierarchical tree combined with a connection stiffness matrix is ​​used to divide the connected domains. Mesh nodes with physical connection stiffness are merged into components, ensuring that the disassembled components conform to the actual mechanical force transmission path. Structural dimensional parameters are converted into discrete numerical sequences and concatenated with material parameters to generate multi-dimensional structural feature vectors as entity node attributes, providing a standardized vector comparison basis for map retrieval. Stress and deformation matrices are divided into cubic mesh clusters, and extreme values ​​and gradient features are extracted as attribute nodes, preserving the local features of the simulation results in spatial distribution. When establishing directed edges, the difference in spatial assembly coordinates of components is calculated as the weight of spatially adjacent edges, integrating the geometric positional relationships between components into the map structure. During retrieval, candidate entity nodes are screened using cosine similarity, and connected attribute nodes are extracted to construct candidate subgraphs. The isomorphic matching degree of the subgraph is calculated by combining node matching degree factors and edge matching degree factors, ensuring that the retrieval process simultaneously considers the structural similarity of components and the matching degree of spatial assembly relationships, guaranteeing the accuracy of the output simulation results. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the overall execution process of the present invention.

[0049] Figure 2 This is a flowchart of the spatial analysis and topology decomposition process of the present invention;

[0050] Figure 3 This is a flowchart illustrating the construction process of entity nodes and attribute nodes in this invention.

[0051] Figure 4 This is a flowchart of the directed edge construction process of the present invention;

[0052] Figure 5 This is a flowchart of the map retrieval process of the present invention;

[0053] Figure 6 This is a flowchart of the numerical calculation process for the subgraph isomorphic matching degree in this invention. Detailed Implementation

[0054] Please refer to the attached document. Figure 1 This implementation provides a power tower structure design and simulation system that runs on a computing device equipped with a processor and memory. The memory stores computer programs that can be executed by the processor, and when the computer programs are executed, they implement the following processing flow.

[0055] We receive a collection of historical power transmission tower digital twin models. This collection includes at least one set of fully parameterized digital twin models of power transmission towers that have completed structural mechanics simulation verification. Each digital twin model contains 3D geometric model data, finite element mesh generation data, material property assignment data, and simulation solution data under preset load conditions. The preset load conditions include, but are not limited to, self-weight load conditions, wind load conditions, icing load conditions, line breakage load conditions, and seismic load conditions. Each load condition corresponds to independent simulation solution data.

[0056] Please refer to the attached document. Figure 2 The system performs spatial analysis and topological decomposition on each digital twin model in the collection of historical power transmission tower digital twin models, generating tower head components, tower body components, and tower leg components. During the spatial analysis and topological decomposition process, the system reads the three-dimensional geometric data and finite element mesh data of the digital twin model. Based on the mechanical force transmission path and spatial structural characteristics, the complete tower model is decomposed into three core functional components: the tower head component, which undertakes the functions of conductor suspension and lightning protection; the tower body component, which undertakes the functions of vertical load transmission and lateral deformation resistance; and the tower leg component, which undertakes the functions of foundation connection and overall load distribution.

[0057] The system extracts the structural dimensions and material properties of each component and maps them to entity nodes in the component-level feature map. The structural dimensions include the geometric dimensions of each member and the connection parameters of each node within the component, while the material properties include the physical and mechanical properties of the steel used in the component. The system standardizes the extracted parameters, transforming them into structured data suitable for map storage and comparison. Based on this structured data, entity nodes are created in the component-level feature map. Each entity node corresponds to a single component and stores all the structural and material parameters of that component.

[0058] The system extracts the finite element stress distribution matrix and finite element deformation displacement matrix of each component under preset load conditions. Extremum and gradient features are then extracted from these matrices and mapped to attribute nodes in the component-level feature map. The finite element stress distribution matrix is ​​a two-dimensional matrix composed of the stress tensor values ​​and corresponding three-dimensional spatial coordinates of all finite element mesh nodes within the component under the preset load conditions. The finite element deformation displacement matrix is ​​a two-dimensional matrix composed of the three-dimensional displacement values ​​and corresponding three-dimensional spatial coordinates of all finite element mesh nodes within the component under the same load conditions. The system extracts spatial dimension features from these two matrices to obtain extremum and gradient features that characterize the mechanical response of the component. Attribute nodes are created in the component-level feature map based on these extracted features. Each attribute node corresponds to a component mechanical response feature under a specific load condition, and the attribute node stores the full simulation feature data of the corresponding component.

[0059] Directed edges are established between entity nodes and attribute nodes to generate a component-level feature graph, which is then stored in a graph database. Directed edges are used to establish relationships between entity nodes and attribute nodes, as well as relationships between different entity nodes, integrating scattered node data into a component-level feature graph with a topological structure. The graph database uses a native graph database that supports attribute graph models to store node data, edge data, and corresponding attribute fields of the component-level feature graph, supporting efficient retrieval and traversal operations based on the graph structure.

[0060] The system receives the structural dimensions and material properties of the power tower components to be designed, executes a graph retrieval command based on subgraph isomorphism in the graph database, and outputs attribute node data that matches the power tower components to be designed. The power tower components to be designed are tower head components, tower body components, or tower leg components that need to undergo mechanical simulation verification during the design process of new power towers. The system receives the structural dimensions and material properties of the components to be designed from the designer, converts the parameters into searchable data consistent with the node data format in the graph database, performs a search operation in the graph database based on the principle of subgraph isomorphism, and filters out the attribute node data with the highest matching degree to the components to be designed. The output attribute node data can be directly used as the estimated mechanical simulation results of the components to be designed without re-executing the complete finite element solution process.

[0061] Please refer to the attached document. Figure 3 In this embodiment, the core elements of the component-level feature map follow a unified storage specification, and the specific definitions are shown in Table 1.

[0062] Table 1. Attribute Definitions of Core Nodes and Edges in Component-Level Feature Graph

[0063] Entity Node Globally unique identifier String type 128-bit distributed unique identifier, with no duplicate values. Entity Node Multidimensional structural feature vector Numeric array type Standardized characteristic data of component structural dimensions and material properties Entity Node Component category tags Enumeration type They are divided into three categories: tower head, tower body, and tower legs. Entity Node Model Number String type The unique number corresponding to the historical digital twin model Attribute Node Component category tags Enumeration type Maintain consistency with the component category label of the corresponding entity node. Attribute Node Load condition label Enumeration type The load case type corresponding to the simulation calculation Attribute Node Stress extreme eigenvector Numeric array type Stress maximum, minimum, and average values ​​sequence for each cubic mesh cluster Attribute Node Stress gradient eigenvector Numeric array type 3D stress gradient tensor expansion sequence of each cubic mesh cluster Attribute Node Deformation extreme value eigenvector Numeric array type Sequence of maximum, minimum, and average deformation displacements for each cubic mesh cluster Attribute Node Deformation gradient eigenvector Numeric array type 3D Deformation Gradient Tensor Expansion Sequence of Each Cubic Mesh Cluster First directed edge Edge type identifier String type Fixed as "physical response correlation" First directed edge Starting Node ID String type The globally unique identifier of the corresponding entity node First directed edge Termination Node ID String type Unique identifier for the corresponding attribute node First directed edge Load condition matching code Numerical Used for quickly filtering associated attribute nodes under specified load conditions. Second directed edge Edge type identifier String type Fixed as "spatial adjacency association" Second directed edge Starting Node ID String type The globally unique identifier of the entity node corresponding to the component with a lower spatial position Second directed edge Termination Node ID String type The globally unique identifier of the entity node corresponding to the component with a higher spatial position. Second directed edge Edge weight vector Numeric array type Three-dimensional absolute value sequence of spatial assembly coordinates of adjacent components

[0064] Table 1 clarifies the storage specifications for various core elements in the component-level feature map, ensuring that the map data generated by decomposing different historical models have a unified format standard, and providing a standardized data foundation for subsequent cross-model map retrieval.

[0065] In this embodiment, please refer to the appendix. Figure 4 By decomposing historical digital twin models of power transmission towers into independent functional components, constructing entity nodes based on the structural and material parameters of the components, and constructing attribute nodes based on the simulated mechanical response characteristics of the components, and establishing relationships between nodes through directed edges to generate component-level feature maps, a structured association storage of power transmission tower structural parameters and simulation result data is achieved. Based on a subgraph isomorphic retrieval mechanism using a graph database, historical simulation result data corresponding to the component to be designed can be directly obtained, breaking the data silo limitations of isolated storage by project in existing technologies and avoiding the processor power consumption caused by repeatedly performing finite element solution calculations.

[0066] In one optional embodiment, the process of spatial analysis and topological decomposition of each digital twin model in the set of historical power tower digital twin models is achieved by using a connected domain partitioning method based on spatial bounding box hierarchical tree and connection stiffness matrix.

[0067] The system reads the 3D geometric patch data and finite element mesh node data from each digital twin model, and constructs a spatial bounding box hierarchy tree containing all geometric patch data and finite element mesh node data. The 3D geometric patch data is a set of triangular or quadrilateral patches that make up the 3D geometric model of the tower, including the vertex coordinates, patch normal vector, and component identifier of each patch. The finite element mesh node data is a set of nodes in the finite element model of the tower, including the 3D spatial coordinates, node number, and element number of each node. The spatial bounding box hierarchy tree adopts an axis-aligned bounding box structure. The root node of the tree corresponds to the maximum bounding box covering the entire spatial range of the tower model. The bounding box corresponding to each parent node can be divided into multiple child bounding boxes corresponding to child nodes, with each child bounding box completely contained within the spatial range of the parent bounding box. The leaf nodes of the tree correspond to the minimum bounding box corresponding to a single geometric patch or a single finite element mesh node.

[0068] For any set of finite element mesh nodes ,in For the first The 3D spatial coordinates of each node, and the corresponding boundary values ​​of the axis-aligned bounding box are calculated as follows:

[0069]

[0070] in, , The enclosure is in Minimum and maximum boundary values ​​in the axial direction. , The enclosure is in Minimum and maximum boundary values ​​in the axial direction. , The enclosure is in Minimum and maximum boundary values ​​in the axial direction.

[0071] Traverse the nodes in the bounding box hierarchy tree and extract the connection stiffness matrix between adjacent finite element mesh nodes. The connection stiffness matrix is ​​a submatrix of the global stiffness matrix obtained by integrating the element stiffness matrices in the finite element model. It corresponds to the stiffness relationship between adjacent finite element mesh nodes. The order of the connection stiffness matrix is ​​consistent with the number of adjacent nodes. The numerical values ​​of the elements in the matrix represent the linear relationship between force and displacement between the corresponding two nodes. The numerical values ​​of the main diagonal elements represent the linear stiffness of the corresponding node itself, and the numerical values ​​of the off-diagonal elements represent the coupling stiffness between the two nodes.

[0072] Based on the values ​​of the main diagonal elements in the connection stiffness matrix, adjacent finite element mesh nodes with a connection stiffness matrix greater than a preset stiffness threshold are assigned to the same initial connected domain. The preset stiffness threshold is pre-set based on the elastic modulus of the steel used in the tower and the cross-sectional dimensions of the members, used to distinguish adjacent nodes with direct mechanical force transmission relationships from those without. The heat transfer coefficient at non-zero, non-diagonal positions in the connection stiffness matrix is ​​extracted. The heat transfer coefficient is a value in the 0-1 range obtained after standardization of the coupling stiffness values, used to quantify the mechanical connection strength between two adjacent nodes. Any unassigned finite element mesh node is used as the starting node and added to the current traversal queue, which is managed using a first-in, first-out (FIFO) queue structure. The first finite element mesh node in the current traversal queue is extracted, and adjacent finite element mesh nodes directly connected to the first finite element mesh node are queried. Directly connected adjacent nodes refer to nodes belonging to the same finite element element as the current node. Determine whether the heat transfer coefficient between the first finite element mesh node and its adjacent finite element mesh nodes is greater than a preset stiffness threshold. If it is, assign the adjacent finite element mesh nodes to the same initial connected component as the starting node and add them to the current traversal queue until the current traversal queue is empty. Repeat the above traversal process until all finite element mesh nodes have been assigned to their corresponding initial connected components.

[0073] For any two adjacent finite element mesh nodes With nodes The corresponding heat transfer coefficient is The preset stiffness threshold is The node affiliation determination condition is:

[0074]

[0075] in, For node affiliation identifier, when At that time, node With nodes Partitioned to the same initial connected component; when At that time, node With nodes They are not partitioned into the same initial connected component.

[0076] The spatial geometric center coordinates of each initial connected domain are calculated. Adjacent initial connected domains whose spatial geometric center coordinates are less than a preset distance threshold and share a common geometric surface are merged. The merged connected domains are then labeled as tower head components, tower body components, and tower leg components, respectively. The spatial geometric center coordinates of an initial connected domain are the average of the three-dimensional spatial coordinates of all finite element mesh nodes within that domain. The preset distance threshold is pre-set based on the overall dimensions of the tower model and is used to determine whether adjacent initial connected domains belong to the same functional component.

[0077] Shared geometric patches refer to geometric patches that are simultaneously contained within two adjacent initial connected components, representing the physical connection between the two initial connected components. After merging, based on the spatial location and functional attributes of the connected components in the tower model, they are labeled as tower head components, tower body components, and tower leg components, respectively, with their spatial locations within the tower model. The connected components of the highest interval along the axis are labeled as tower head components, and their spatial location is within the tower model. The connected domains in the middle section along the axis are marked as tower components, and their spatial location is within the tower model. The connected domain of the lowest interval in the axial direction is marked as the tower leg component.

[0078] The effects of different preset stiffness thresholds on the initial connected component partitioning results are shown in Table 2.

[0079] Table 2 Comparison of initial connected component partitioning results under different preset stiffness thresholds

[0080] Preset stiffness threshold Initial number of connected components Maximum number of nodes in a singly connected component Minimum number of nodes in a simply connected domain Number of components after merger 0.1 226 12854 2 17 0.3 89 24106 4 8 0.5 37 38921 8 4 0.7 12 65839 15 3 0.9 5 112547 32 2

[0081] Table 2 illustrates the impact of different preset stiffness thresholds on the initial connected component partitioning results. The value of the preset stiffness threshold directly determines the granularity of the initial connected component partitioning. A threshold that is too low will result in overly fine connected component partitioning, increasing the computational load of subsequent merging operations; a threshold that is too high will cause connected components of different functional components to be incorrectly merged, affecting the accuracy of component disassembly. Based on the comparative data in Table 2, the optimal preset stiffness threshold can be selected according to the mesh density of the tower model to ensure that the component disassembly results conform to the actual mechanical force transmission path.

[0082] In this embodiment, spatial indexing and fast traversal of the 3D data of the tower model are achieved by constructing a spatial bounding box hierarchical tree. A breadth-first traversal method based on the connection stiffness matrix and heat transfer coefficient is used to automatically partition the initial connected domains. Combined with a connected domain merging mechanism based on spatial geometric center distance and shared geometric patches, the tower model is accurately disassembled into tower head, tower body, and tower leg components. This disassembly method is based on mechanical force transmission relationships, ensuring that the disassembled components possess independent mechanical functional properties, providing accurate foundational data for the subsequent construction of component-level feature maps.

[0083] In one alternative embodiment, the mapping process of structural dimension parameters and material property parameters to solid nodes, and the mapping process of finite element stress distribution matrix and finite element deformation displacement matrix to property nodes, are implemented using a standardized eigenvector construction and spatial mesh generation method.

[0084] The structural dimensional parameters and material property parameters are mapped to solid nodes in the component-level feature map. Specifically, the process involves parsing the member cross-sectional length, width, wall thickness, and node spacing from the structural dimensional parameters. The member cross-sectional length and width represent the outer contour dimensions of the main and diagonal members within the component; the wall thickness represents the tube wall thickness of the member; and the node spacing represents the axial distance between the connecting nodes of two adjacent members. The member cross-sectional length, width, wall thickness, and node spacing are then converted into discrete numerical sequences according to a preset quantization interval. This preset quantization interval is based on the dimensional accuracy requirements in the power transmission tower design specifications and is used to convert continuous dimensional parameters into standardized discrete values, eliminating dimensional accuracy differences between different design models.

[0085] The yield strength, elastic modulus, and density of the material properties are analyzed. Yield strength is the critical stress value at which the material undergoes plastic deformation, elastic modulus is the stress-to-strain ratio during the linear elastic deformation stage, and density is the mass per unit volume of the material. The yield strength, elastic modulus, and density are appended to the end of the discrete numerical sequence to generate a multidimensional structural feature vector. Entity node objects corresponding to the tower head component, tower body component, and tower leg component are created in the graph database. The multidimensional structural feature vector is stored as the first attribute field of the entity node object, and a globally unique identifier (a 128-bit distributed unique identifier) ​​is assigned to each entity node object to ensure that no two entity nodes in the graph database have duplicate identifiers.

[0086] For any component, the discrete numerical sequence of its member dimension parameters is as follows: ,in to These are the discrete values ​​of the quantized bar section length, section width, wall thickness, and node spacing, respectively. The material property parameter sequence is as follows: ,in For yield strength, For elastic modulus, The generated multidimensional structural feature vector is the density. for:

[0087]

[0088] in, For vector concatenation operators, multidimensional structural feature vectors The dimension is .

[0089] Extreme value features and gradient features are extracted from the finite element stress distribution matrix and finite element deformation displacement matrix. These features are then mapped to attribute nodes in a component-level feature map. Specifically, the finite element stress distribution matrix is ​​divided into multiple cubic mesh clusters according to spatial coordinates. The maximum, minimum, and average stress values ​​within each cubic mesh cluster are calculated, and these values ​​are combined to form a stress extreme value feature vector. The stress change rate in the three orthogonal directions of each cubic mesh cluster is calculated, and these rates are arranged into a three-dimensional stress gradient tensor. This three-dimensional stress gradient tensor is then expanded into a one-dimensional vector and used as the stress gradient feature.

[0090] The finite element deformation displacement matrix is ​​processed using the same method as the stress extreme value eigenvector and stress gradient feature, generating deformation extreme value eigenvectors and deformation gradient features. These features are then combined into feature data for attribute nodes. Corresponding attribute node objects are created in the graph database, and the combined feature data is stored as the second attribute field of the attribute node objects. A corresponding component category label is added to each attribute node object, consistent with the component type of the corresponding entity node, categorized as tower head, tower body, and tower leg.

[0091] The specific process of dividing the finite element stress distribution matrix into multiple cubic mesh clusters according to spatial coordinates is as follows: First, obtain the three-dimensional spatial coordinates of all finite element mesh nodes in the finite element stress distribution matrix. Calculate the maximum and minimum boundary values ​​of all three-dimensional spatial coordinates along the three coordinate axes. The boundary value calculation method is consistent with the boundary calculation method of the spatial bounding box. Starting from the minimum boundary value, divide the matrix at equal intervals along the three coordinate axes according to a preset fixed side length, generating multiple virtual cubic meshes covering the entire spatial range of the finite element stress distribution matrix. The preset fixed side length is pre-set based on the overall size of the component and the average size of the finite element mesh, ensuring that each virtual cubic mesh contains a sufficient number of finite element mesh nodes. Finite element mesh nodes whose three-dimensional spatial coordinates fall within the same virtual cubic mesh are assigned to the same cubic mesh cluster. When no finite element mesh nodes exist within a virtual cubic mesh, delete the virtual cubic mesh and record the relative spatial position index of adjacent cubic mesh clusters. The relative spatial position index is used to characterize the spatial adjacency relationship between different cubic mesh clusters, providing a spatial positional basis for subsequent gradient feature calculations.

[0092] For any cubic mesh cluster Its interior contains There are finite element mesh nodes, and the corresponding von Mises equivalent stress value for each node is: The stress extremum eigenvector corresponding to this cubic mesh cluster for:

[0093]

[0094] in, This represents the maximum stress value within the mesh cluster. This represents the minimum stress value within the mesh cluster. This represents the average stress value within the grid cluster.

[0095] For any cubic mesh cluster Its spatial center coordinates are , and the grid cluster in The average stress value of adjacent mesh clusters in the positive direction of the axis is , The average stress value of adjacent mesh clusters in the negative axial direction is , The average stress value of adjacent mesh clusters in the positive direction of the axis is , The average stress value of adjacent mesh clusters in the negative axial direction is , The average stress value of adjacent mesh clusters in the positive direction of the axis is , The average stress value of adjacent mesh clusters in the negative axial direction is The preset fixed side length is The three-dimensional stress gradient tensor corresponding to this mesh cluster for:

[0096] in, for The rate of change of stress in the axial direction, The stress change rate in the Y-axis direction is... for The rate of change of stress along the axial direction. The three-dimensional stress gradient tensor. Expanding in row-major order yields a one-dimensional stress gradient eigenvector.

[0097] For any cubic mesh cluster Its interior contains Each node in the finite element mesh has 10 finite element mesh nodes. , , The deformation displacement values ​​in the axial direction are respectively The average deformation displacement of the mesh cluster in the three coordinate axes is respectively , , Using the same calculation method as the stress gradient tensor, the three-dimensional deformation gradient tensor is obtained. ,Will Expanding in row-major order yields a one-dimensional deformation gradient eigenvector.

[0098] The effects of different fixed side lengths on the cubic mesh cluster division and feature extraction results are shown in Table 3.

[0099] Table 3. Correspondence between fixed side lengths of cubic mesh clusters and feature extraction results

[0100] Fixed side length (mm) Number of cubic mesh clusters Average number of nodes per grid cluster Stress eigenvector dimension Deformation feature vector dimension Feature extraction time (ms) 50 2864 12 8592 8592 1286 100 724 48 2172 2172 327 200 182 192 546 546 86 300 86 406 258 258 42 500 32 1089 96 96 18

[0101] Table 3 illustrates the impact of different fixed side lengths on the cubic mesh clustering and feature extraction results. The value of the fixed side length directly determines the granularity of the mesh clusters and the dimension of the feature vectors. Too small a side length leads to an excessive number of mesh clusters, increasing the dimension of the feature vectors and the computational cost of feature storage and retrieval. Too large a side length results in an excessive number of nodes within the mesh clusters, causing the loss of local details in stress and deformation distribution, thus affecting the accuracy of the simulation results. Based on the comparative data in Table 3, the optimal fixed side length can be selected according to the mesh density and feature representation requirements of the component, controlling computational and storage costs while ensuring the accuracy of feature representation.

[0102] In this embodiment, by quantizing structural dimensional parameters into discrete numerical sequences and concatenating them with material property parameters, a standardized multidimensional structural feature vector is generated. This provides a unified feature comparison basis for the construction of solid nodes, eliminating parameter format differences between different design models. Through a cubic mesh cluster spatial partitioning method, the continuous finite element stress and deformation matrices are transformed into discrete mesh cluster features. The extracted extreme value features and gradient features fully preserve the global statistical characteristics and spatial distribution local characteristics of the simulation results, providing accurate mechanical response feature data for the construction of attribute nodes. The standardized feature construction method ensures that the node data of different components have a unified format, providing a reliable data foundation for subsequent graph retrieval.

[0103] In one optional embodiment, the process of establishing directed edges between entity nodes and attribute nodes, as well as the graph retrieval process based on subgraph isomorphism, is implemented using an attribute graph association mechanism and a two-layer matching degree calculation method.

[0104] Directed edges are established between entity nodes and attribute nodes to generate a component-level feature graph and store it in a graph database. Specifically, globally unique identifiers are extracted from entity nodes and component category labels from attribute nodes. The globally unique identifiers uniquely identify entity nodes, and the component category labels identify the component type corresponding to the attribute node. A first directed edge is established in the direction from the entity node to the attribute node, and its edge type is marked as physical response association. This first directed edge establishes the association between the structural material parameters of the component and the corresponding mechanical simulation response characteristics of the component. Simulation feature data from the corresponding attribute nodes can be directly retrieved from the entity nodes through this first directed edge.

[0105] The spatial assembly coordinates of the components corresponding to the entity nodes in the digital twin model are extracted. The spatial assembly coordinate difference between the components corresponding to adjacent entity nodes is calculated. A second directed edge is established between two adjacent entity nodes, and the edge type of the second directed edge is marked as spatial adjacency association. The spatial assembly coordinate difference is stored as the edge weight attribute of the second directed edge. Adjacent entity nodes refer to entity nodes whose corresponding components have a direct physical connection relationship in the tower model, including entity nodes corresponding to the tower head component and entity nodes corresponding to the tower body component, and entity nodes corresponding to the tower body component and entity nodes corresponding to the tower leg component.

[0106] The specific process for calculating the spatial assembly coordinate difference between adjacent entity nodes and their corresponding components is as follows: Extract all finite element mesh nodes contained in each of the adjacent entity nodes and their corresponding components; calculate the average three-dimensional spatial coordinates of all finite element mesh nodes within the tower head assembly, tower body assembly, and tower leg assembly; and set this average three-dimensional spatial coordinate as the spatial assembly coordinate of the component. Then, calculate the spatial assembly coordinates of two adjacent components in... The absolute value of the coordinate difference along the axis The absolute value of the coordinate difference along the axis and The absolute value of the coordinate difference along the axis direction will The absolute value of the coordinate difference along the axis The absolute value of the coordinate difference along the axis and The absolute values ​​of the coordinate differences along the axes are concatenated to form a three-dimensional spatial distance vector, which is then used as the edge weight attribute of the second directed edge between adjacent entity nodes.

[0107] For any given component, its internal contents include There are finite element mesh nodes, and the three-dimensional spatial coordinates of each node are: , The spatial assembly coordinates of this component for:

[0108] in, For all nodes within the component The average value of the axis coordinates. For all nodes within the component The average value of the axis coordinates. For all nodes within the component The average value of the axis coordinates.

[0109] For two adjacent components With components Their spatial assembly coordinates are respectively and The edge weight vector of the corresponding second directed edge for:

[0110]

[0111] in, for The absolute value of the coordinate difference along the axis. for The absolute value of the coordinate difference along the axis. for The absolute value of the coordinate difference along the axis.

[0112] Traverse all entity nodes and attribute nodes in the component-level feature graph, repeatedly performing the directed edge construction step until the component-level feature graph is completed. The completed component-level feature graph is stored in a graph database, which supports directed traversal based on edge type, filtered queries based on node attributes, and subgraph matching operations.

[0113] Please refer to the attached document. Figure 5The system receives the structural dimension parameters and material property parameters of the power tower component to be designed, executes a graph retrieval command based on subgraph isomorphism in the graph database, and outputs attribute node data that matches the power tower component to be designed. Specifically, the structural dimension parameters and material property parameters of the power tower component to be designed are converted into a multi-dimensional structural feature vector to be retrieved. The conversion method is completely consistent with the multi-dimensional structural feature vector generation method of historical components, ensuring that the vector to be retrieved has the same dimension and quantification standard as the vector in the graph database.

[0114] Calculate the cosine similarity between the multidimensional structural feature vector to be retrieved and the multidimensional structural feature vectors of each entity node in the graph database. Extract entity nodes with a cosine similarity greater than a preset similarity threshold to form a candidate entity node set. The preset similarity threshold is set in advance according to the retrieval accuracy requirements and is used to filter entity nodes with low structural similarity, thereby narrowing the retrieval range of subsequent subgraph matching.

[0115] For the multidimensional structural feature vector to be retrieved The multidimensional structural feature vector of any entity node in the graph database Both vectors have dimensions. The corresponding cosine similarity for:

[0116]

[0117] in, The first of the vectors to be retrieved Dimensional value, For the first entity node vector in the database Dimensional value, Let L2 norm be the vector to be retrieved. It is the L2 norm of the entity node vector in the database.

[0118] Please refer to the attached document. Figure 6 Starting from each candidate entity node in the candidate entity node set, connected attribute nodes are extracted along directed edges to construct multiple candidate subgraphs. Each candidate subgraph contains a candidate entity node, attribute nodes connected to the entity node via a first directed edge, adjacent entity nodes connected to the entity node via a second directed edge, and attribute nodes corresponding to the adjacent entity nodes. The power tower component to be designed is transformed into a matching subgraph, which contains the entity node to be retrieved corresponding to the component to be designed, virtual attribute nodes corresponding to the entity node to be retrieved, virtual entity nodes corresponding to adjacent components that have spatial adjacency with the entity node to be retrieved, and virtual attribute nodes corresponding to the adjacent virtual entity nodes. The subgraph isomorphism matching degree between the matching subgraph and the multiple candidate subgraphs is calculated.

[0119] The specific process for calculating the subgraph isomorphic matching degree between the subgraph to be matched and multiple candidate subgraphs is as follows: Extract the sets of entity nodes and directed edges from both the subgraph to be matched and the candidate subgraphs. Traverse the set of entity nodes, calculate the Euclidean distance between the multidimensional structural feature vectors of the entity nodes in the subgraph to be matched and the corresponding multidimensional structural feature vectors of the entity nodes in the candidate subgraphs, and sum the reciprocals of these Euclidean distances to generate the node matching degree factor. Traverse the set of directed edges, calculate the magnitude of the vector difference between the edge weight attribute of the second directed edge in the subgraph to be matched and the edge weight attribute of the corresponding second directed edge in the candidate subgraph, and sum the reciprocals of these vector difference magnitudes to generate the edge matching degree factor. Multiply the node matching degree factor and the edge matching degree factor to generate the subgraph isomorphic matching degree.

[0120] For the set of entity nodes in the subgraph to be matched The set of entity nodes corresponding to the candidate subgraph ,in The number of entity nodes in the subgraph. The corresponding multidimensional structural feature vector is ,node The corresponding multidimensional structural feature vector is Node matching factor for:

[0121]

[0122] in, Let be the Euclidean distance between the two vectors. This is a preset minimum constant used to avoid cases where the denominator is 0.

[0123] For the second set of directed edges in the subgraph to be matched The second directed edge set corresponding to the candidate subgraph ,in The number of the second directed edges in the subgraph, edge The corresponding edge weight vector is ,side The corresponding edge weight vector is Edge matching factor for:

[0124]

[0125] in, The magnitude of the difference between the two edge weight vectors. This is a preset minimum constant used to avoid cases where the denominator is 0.

[0126] For the subgraph to be matched and any candidate subgraph, the corresponding subgraph isomorphic matching degree value for:

[0127]

[0128] in, For node matching degree factor, The edge matching degree factor is the matching degree value. The larger the value, the higher the isomorphism between the candidate subgraph and the subgraph to be matched, and the better the matching degree between the corresponding component and the component to be designed.

[0129] Based on the calculated subgraph isomorphism matching degree, all candidate subgraphs are sorted in descending order. The attribute node data corresponding to the top-ranked candidate subgraph is extracted and output as the attribute node data matching the power transmission tower component to be designed. The output attribute node data includes stress extreme value feature vectors, stress gradient features, deformation extreme value feature vectors, and deformation gradient features, which can be directly used for the mechanical performance evaluation of the component to be designed without re-executing the complete process of finite element model construction, mesh generation, and equation solving.

[0130] The results of the isomorphic matching degree calculation between different candidate subgraphs and the subgraph to be matched are shown in Table 4.

[0131] Table 4. Calculation results of isomorphic matching degree between the subgraph to be matched and the candidate subgraph.

[0132] Candidate subgraph number Node matching factor Edge matching factor Subgraph isomorphic matching degree value Matching order 001 28.642 12.387 354.721 1 002 26.915 10.246 275.769 2 003 22.158 11.863 262.874 3 004 18.426 9.752 179.691 4 005 15.739 8.624 135.736 5

[0133] Table 4 shows the isomorphic matching degree calculation results of different candidate subgraphs and the subgraph to be matched. The matching degree values ​​consider both the node matching degree of the component structural parameters and the edge matching degree of the component spatial assembly relationship. By sorting the matching degrees, the candidate subgraphs with the highest matching degree to the component to be designed can be quickly screened, ensuring that the output simulation results data have sufficient accuracy. In this embodiment, two different types of directed edges are used to establish the physical response correlation between the component structural parameters and the mechanical simulation results, as well as the spatial adjacency correlation between adjacent components, thus fully integrating the component's physical properties, mechanical response, and spatial positional relationship into the topological structure of the component-level feature graph. The candidate node screening mechanism based on cosine similarity effectively narrows the retrieval range of subgraph matching and improves retrieval efficiency. The subgraph isomorphic matching degree calculation method combining node matching degree factor and edge matching degree factor considers both the similarity of the component's own structure and the matching degree of the spatial assembly relationship between components, ensuring the accuracy of the retrieval results and providing reliable historical simulation result data for the component to be designed.

Claims

1. A power transmission tower structure design and simulation system, characterized in that, It includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it completes the following steps: receiving a set of historical power tower digital twin models, performing spatial analysis and topological decomposition on each digital twin model in the set of historical power tower digital twin models, and generating tower head components, tower body components, and tower leg components. Extract the structural dimension parameters and material property parameters of each component, and map the structural dimension parameters and material property parameters to entity nodes in the component-level feature map; Extract the finite element stress distribution matrix and finite element deformation displacement matrix of each component under the preset load conditions, extract extreme value features and gradient features from the finite element stress distribution matrix and finite element deformation displacement matrix, and map the extreme value features and gradient features to attribute nodes in the component-level feature map. Establish directed edges between entity nodes and attribute nodes, generate component-level feature graphs, and store them in a graph database; The system receives the structural dimensions and material properties of the power tower components to be designed, executes a graph retrieval command based on subgraph isomorphism in the graph database, and outputs attribute node data that matches the power tower components to be designed.

2. The power transmission tower structure design and simulation system according to claim 1, characterized in that, Spatial analysis and topological decomposition are performed on each digital twin model in the collection of historical power tower digital twin models, including: reading the three-dimensional geometric patch data and finite element mesh node data of each digital twin model, and constructing a spatial bounding box hierarchical tree containing all geometric patch data and finite element mesh node data; Traverse the nodes in the spatial bounding box hierarchy tree and extract the connection stiffness matrix between adjacent finite element mesh nodes; Based on the values ​​of the main diagonal elements in the connection stiffness matrix, adjacent finite element mesh nodes with connection stiffness matrices greater than a preset stiffness threshold are assigned to the same initial connected domain. Calculate the spatial geometric center coordinates of each initial connected domain, merge adjacent initial connected domains whose spatial geometric center coordinates are less than a preset distance threshold and have a shared geometric patch, and label the merged connected domains as tower head component, tower body component and tower leg component respectively.

3. The power tower structure design and simulation system according to claim 1, characterized in that, Mapping structural dimension parameters and material property parameters to entity nodes in component-level feature maps includes: parsing the member cross-sectional length, cross-sectional width, wall thickness, and node spacing in the structural dimension parameters, and converting the member cross-sectional length, cross-sectional width, wall thickness, and node spacing into discrete numerical sequences according to a preset quantization interval; The yield strength, elastic modulus, and density of the material properties are analyzed, and the yield strength, elastic modulus, and density are appended to the tail of the discrete numerical sequence to generate a multidimensional structural feature vector. Create entity node objects corresponding to the tower head component, tower body component, and tower leg component in the graph database. Store the multidimensional structural feature vector as the first attribute field of the entity node object and assign a globally unique identifier to the entity node object.

4. The power tower structure design and simulation system according to claim 1, characterized in that, Extracting extreme value features and gradient features from the finite element stress distribution matrix and finite element deformation displacement matrix includes: dividing the finite element stress distribution matrix into multiple cubic mesh clusters according to spatial coordinates, calculating the maximum, minimum and average values ​​of stress values ​​in each cubic mesh cluster, and combining the maximum, minimum and average values ​​into a stress extreme value feature vector; The stress change rate of each cubic mesh cluster in the three orthogonal directions in space is calculated respectively. The stress change rate in the three orthogonal directions is arranged into a three-dimensional stress gradient tensor. The three-dimensional stress gradient tensor is expanded into a one-dimensional vector and used as the stress gradient feature. The finite element deformation displacement matrix is ​​processed using the same method as the stress extremum eigenvector and stress gradient feature to generate the deformation extremum eigenvector and deformation gradient feature.

5. The power tower structure design and simulation system according to claim 1, characterized in that, Establish directed edges between entity nodes and attribute nodes, including: extracting globally unique identifiers from entity nodes and component category labels from attribute nodes; Establish the first directed edge in the direction from the entity node to the attribute node, and mark the edge type of the first directed edge as physical response association; Extract the spatial assembly coordinates of the corresponding components of the entity nodes in the digital twin model, and calculate the spatial assembly coordinate difference between the corresponding components of adjacent entity nodes; Establish a second directed edge between two adjacent entity nodes, mark the edge type of the second directed edge as spatial adjacency association, and store the spatial assembly coordinate difference as the edge weight attribute of the second directed edge. Traverse all entity nodes and attribute nodes in the component-level feature graph, and repeat the directed edge building step until the component-level feature graph is constructed.

6. The power tower structure design and simulation system according to claim 1, characterized in that, Execute graph retrieval instructions based on subgraph isomorphism in the graph database, including: converting the structural dimension parameters and material property parameters of the power tower component to be designed into a multidimensional structural feature vector to be retrieved; Calculate the cosine similarity between the multidimensional structural feature vector to be retrieved and the multidimensional structural feature vectors of each entity node in the graph database, and extract entity nodes whose cosine similarity is greater than a preset similarity threshold to form a candidate entity node set; Starting from each candidate entity node in the candidate entity node set, extract connected attribute nodes along directed edges to construct multiple candidate subgraphs; The components of the power tower to be designed are converted into subgraphs to be matched, and the subgraph isomorphic matching degree between the subgraph to be matched and multiple candidate subgraphs is calculated.

7. The power tower structure design and simulation system according to claim 2, characterized in that, Based on the values ​​of the main diagonal elements in the connection stiffness matrix, adjacent finite element mesh nodes with connection stiffness matrices greater than a preset stiffness threshold are divided into the same initial connected domain, including: extracting the heat transfer coefficient at non-zero, non-diagonal positions in the connection stiffness matrix; Take any unassigned finite element mesh node as the starting node and add the starting node to the current traversal queue; Extract the first finite element mesh node in the current traversal queue, and query the adjacent finite element mesh nodes that are directly connected to the first finite element mesh node; Determine whether the heat transfer coefficient between the first finite element mesh node and its adjacent finite element mesh nodes is greater than a preset stiffness threshold. If it is greater than the preset stiffness threshold, then divide the adjacent finite element mesh nodes into the same initial connected domain as the starting node and add the adjacent finite element mesh nodes to the current traversal queue until the current traversal queue is empty.

8. The power tower structure design and simulation system according to claim 4, characterized in that, The finite element stress distribution matrix is ​​divided into multiple cubic mesh clusters according to spatial coordinates, including: obtaining the three-dimensional spatial coordinates of all finite element mesh nodes in the finite element stress distribution matrix, and calculating the maximum and minimum boundary values ​​of all three-dimensional spatial coordinates in the three coordinate axis directions; Starting from the minimum boundary value, the three coordinate axes are divided at equal intervals according to the preset fixed side length to generate multiple virtual cube meshes covering the entire space of the finite element stress distribution matrix. Finite element mesh nodes whose three-dimensional spatial coordinates fall within the same virtual cube mesh are divided into the same cube mesh cluster; When there are no finite element mesh nodes within the virtual cube mesh, delete the virtual cube mesh and record the relative spatial position index of adjacent cube mesh clusters.

9. The power tower structure design and simulation system according to claim 5, characterized in that, Calculate the spatial assembly coordinate difference between the components corresponding to adjacent entity nodes, including: extracting all finite element mesh nodes contained in each component corresponding to the adjacent entity nodes, calculating the average three-dimensional spatial coordinates of all finite element mesh nodes in the tower head component, tower body component and tower leg component respectively, and setting the average three-dimensional spatial coordinates as the spatial assembly coordinates of the components. Calculate the absolute values ​​of the spatial assembly coordinate differences between two adjacent components in the X-axis direction, Y-axis direction, and Z-axis direction, respectively. The absolute values ​​of the coordinate differences along the X-axis, Y-axis, and Z-axis are concatenated to form a three-dimensional spatial distance vector, which is then used as the edge weight attribute of the second directed edge between adjacent entity nodes.

10. The power tower structure design and simulation system according to claim 6, characterized in that, Calculate the subgraph isomorphic matching degree between the subgraph to be matched and multiple candidate subgraphs, including: extracting the set of entity nodes and the set of directed edges in the subgraph to be matched and the candidate subgraphs; Traverse the set of entity nodes, calculate the Euclidean distance between the multidimensional structural feature vector of the entity node in the subgraph to be matched and the multidimensional structural feature vector of the corresponding entity node in the candidate subgraph, and sum the reciprocals of the Euclidean distances to generate the node matching degree factor. Traverse the set of directed edges, calculate the vector difference magnitude between the edge weight attribute of the second directed edge in the subgraph to be matched and the edge weight attribute of the corresponding second directed edge in the candidate subgraph, and sum the reciprocals of the vector difference magnitudes to generate the edge matching degree factor. Multiply the node matching degree factor by the edge matching degree factor to generate the subgraph isomorphic matching degree value.

Images