Geometric partition-based integrated die casting structure parameterization modeling method

Abstract

The application discloses a geometric partition-based integrated die casting structure parameterized modeling method, and relates to the technical field of three-dimensional modeling. The method comprises the following steps: performing feature aggregation on each node in a CAE model of an integrated die casting structure to obtain a plurality of supernodes of the CAE model; determining the nearest adjacent supernode of each supernode and establishing an edge index; constructing a graph network model corresponding to the integrated die casting structure according to each supernode and the edge index; taking the graph network model as a training sample and taking a target performance parameter corresponding to the CAE model as a training label to train a graph convolution network; and when a prediction error of the graph convolution network reaches a preset optimal value, determining the current graph network model as an optimal graph network model of the integrated die casting structure. According to the application, each node is compressed into a supernode through feature aggregation, and then the graph network model is optimized based on the supernode and the graph convolution network, so that efficient and accurate parameterized design can be realized.

Description

Technical Field

[0001] This application relates to the field of 3D modeling technology, and in particular to a parametric modeling method for integral die-cast parts based on geometric partitioning. Background Technology

[0002] Complex structures such as the die-cast rear floor of a unibody vehicle often have tens or even hundreds of thousands of nodes when represented using finite element meshes or parametric point clouds. Directly inputting these nodes into a graph convolutional network (GCNN) leads to the curse of node dimensionality in graph structures. The massive input means the GCNN needs to process extremely large adjacency and feature matrices, resulting in a huge number of model parameters, slow convergence, and extremely high memory and computational requirements. Furthermore, a single forward propagation is too time-consuming, failing to meet the demands of rapid iterative engineering design. In high-dimensional sparse data, the model is more likely to learn noise rather than true patterns, resulting in poor generalization ability. In addition, traditional dimensionality reduction methods or simple mesh coarsening can destroy the geometric and topological information of the structure, while random sampling may lose data from key feature points, neither of which is suitable for predicting the performance of geometrically sensitive structures. Summary of the Invention

[0003] The main objective of this application is to propose a parametric modeling method and related equipment for integral die-casting structures based on geometric partitioning, so as to achieve efficient and accurate parametric design of integral die-casting structures.

[0004] To achieve the above objectives, one aspect of this application proposes a parametric modeling method for integral die-cast parts based on geometric partitioning, the method comprising the following steps:

[0005] Feature aggregation is performed on each node in the CAE model of the integral die-cast part structure to obtain multiple super nodes of the CAE model;

[0006] Determine the nearest neighboring supernode for each supernode and establish an edge index;

[0007] A graph network model corresponding to the integrated die-casting structure is constructed based on each of the super nodes and the edge indices.

[0008] The graph network model is used as the training sample, and the target performance parameters corresponding to the CAE model are used as the training labels to train the graph convolutional network.

[0009] When the prediction error of the graph convolutional network reaches the preset optimal value, the current graph network model is determined as the optimal graph network model for the integrated die-casting structure.

[0010] In some embodiments, the step of feature aggregation of each node in the CAE model of the integral die-cast part structure to obtain multiple supernodes of the CAE model includes the following steps:

[0011] Extract the CAE model corresponding to different structural parameters from the parameterized database of the integrated die-cast part structure;

[0012] Based on the finite element mesh nodes of each CAE model, each CAE model is discretized into a node set containing multiple nodes;

[0013] Initial feature vectors are extracted from each of the node sets, and the initial feature vectors are converted into final feature matrices using a weighted coefficient matrix;

[0014] Using the final feature matrix as the clustering target, all nodes in each node set are clustered and assigned to the corresponding clusters.

[0015] Feature aggregation is performed on the nodes in each cluster to obtain the supernode corresponding to each cluster.

[0016] In some embodiments, the extraction of initial feature vectors for each of the node sets includes the following steps:

[0017] The initial feature vector is extracted from all nodes in each of the aforementioned node sets as follows:

[0018] ;

[0019] in, For nodes The initial feature vector, Let x, y, and z be the initial spatial coordinates in the x, y, and z directions, respectively. For nodes The initial thickness, For nodes Initial material property encoding;

[0020] The initial feature vector is normalized, and the initial spatial coordinates, the initial thickness, and the initial material property codes are normalized.

[0021] The process of converting the initial feature vector into the final feature matrix using a weighted coefficient matrix includes the following steps:

[0022] Construct the weighted coefficient matrix for:

[0023] ;

[0024] in, Represents the spatial nearest neighbor weight. Represents the thickness nearest neighbor weight. Indicates the nearest neighbor weight of materials;

[0025] Using the weighted coefficient matrix The normalized initial feature vector The feature information in the matrix is ​​weighted to obtain the final feature matrix. ,in These are the final spatial coordinates in the x, y, and z directions, respectively. For nodes The final thickness, For nodes The final material property code.

[0026] In some embodiments, before performing feature aggregation on the nodes in each cluster to obtain the supernode corresponding to each cluster, the method further includes the following steps:

[0027] The connectivity coefficient matrix is ​​constructed based on the edges between the finite element mesh nodes of the CAE model. ;in, Represents a node There is a connection between them;

[0028] For each node, check whether connected adjacent nodes belong to the same cluster. If the number of nodes belonging to another cluster reaches a first set percentage or the number of nodes with different material properties reaches a second set percentage, return to the step of using the final feature matrix as the clustering target to cluster all nodes in each node set and assign them to the corresponding cluster, until the number of nodes belonging to another cluster is less than the first set percentage or the number of nodes with different material properties is less than the second set percentage.

[0029] In some embodiments, performing feature aggregation on nodes in each cluster to obtain the supernode corresponding to each cluster includes the following steps:

[0030] For each cluster, the coordinates of the center point of all nodes are taken as the geometric features of the supernode. ;

[0031] The thickness of all nodes in each cluster is averaged, and the average thickness is taken as the thickness feature of the supernode. ;

[0032] For all nodes in each cluster, identify the node with the largest number of nodes and the same material properties, and then use the material properties as the material features of the supernode. Finally, the supernode is obtained. .

[0033] In some embodiments, constructing the graph network model corresponding to the integral die-casting structure based on each of the supernodes and the edge indices includes the following steps:

[0034] The graph network model is defined as follows: ;

[0035] in:

[0036] Represents the set of supernodes. Each of the supernodes serves as a node in the graph network model;

[0037] Denotes the set of edges. , representing the edge index between each of the supernodes, the edge index serving as the edge of the graph network model. ;

[0038] Represents the adjacency matrix. , This represents the topological relationship of the integral die-cast component structure. If there is no connection between two nodes, then... If the corresponding element is 0, then if a connection exists, then... The corresponding element is 1;

[0039] Represents a collection of node attributes. .

[0040] In some embodiments, before extracting the CAE models corresponding to different structural parameters from the parameterized database of the integral die-cast part structure, the method includes the following steps:

[0041] On the base plate of the floor after integral die casting of the integral die-cast structure, a variable topological mesh is defined, the mesh dimension of which is M*N, and key control nodes are defined in the topological mesh. ,in , which are parametric coordinates relative to the base edge;

[0042] Set activation parameters The activation parameters Key control nodes for characterizing the generation of ribs in the topological mesh. ;

[0043] A topology mesh is generated in the shell CAD file of the integral die-cast floor based on the topology mesh;

[0044] According to the activation parameters Generate rib paths;

[0045] The rib path is scanned and shaped to create a three-dimensional rib solid.

[0046] The ribbed entity and the shell of the integral die-cast floor are subjected to Boolean addition operation to form a ribbed integral die-cast floor model;

[0047] The ribbed integral die-cast floor model was meshed using full-shell elements, and the thickness of the ribs was defined in the CAE environment. ,thickness With the activation parameters Correspondingly;

[0048] A torsional stiffness condition is applied in a CAE environment, and the process of calling the CAE solver, setting analysis cards, applying loads and boundary conditions, and extracting results is scripted. The calculation results are the target performance parameters; wherein, the target performance parameters include the maximum displacement and the maximum equivalent stress.

[0049] Create the parameterized database, and store the corresponding structural parameters, target performance parameters, and CAE model as a data record in the parameterized database.

[0050] To achieve the above objectives, another aspect of this application proposes a parametric modeling device for integral die-casting structures based on geometric partitioning, the device comprising:

[0051] The node aggregation unit is used to perform feature aggregation on each node in the CAE model of the integral die-cast part structure to obtain multiple super nodes of the CAE model.

[0052] An edge index building unit is used to determine the nearest neighboring supernode for each supernode and to build an edge index;

[0053] The graph network model construction unit is used to construct a graph network model corresponding to the integral die-casting structure based on each of the super nodes and the edge indices.

[0054] The graph convolutional network training unit is used to train the graph convolutional network using the graph network model as training samples and the target performance parameters corresponding to the CAE model as training labels.

[0055] The model determination unit is used to determine the current graph network model as the optimal graph network model for the integrated die-casting structure when the prediction error of the graph convolutional network reaches a preset optimal value.

[0056] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0057] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0058] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0059] The embodiments of this application include at least the following beneficial effects:

[0060] This application provides a parametric modeling method and related equipment for integrated die-casting structures based on geometric partitioning. The scheme involves feature aggregation of each node in the CAE model of the integrated die-casting structure to obtain multiple supernodes; determining the nearest neighboring supernodes of each supernode and establishing edge indices; constructing a graph network model corresponding to the integrated die-casting structure based on each supernode and edge index; training a graph convolutional network using the graph network model as training samples and the target performance parameters corresponding to the CAE model as training labels; and determining the optimal graph network model for the integrated die-casting structure when the prediction error of the graph convolutional network reaches a preset optimal value. This application solves the curse of dimensionality problem in graph convolutional networks caused by an excessive number of nodes, while preserving the original geometric and topological key information of the integrated die-casting structure to the maximum extent, thus achieving efficient and accurate parametric design. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 A flowchart illustrating a parametric modeling method for an integral die-cast part structure based on geometric partitioning, provided in an embodiment of this application;

[0063] Figure 2An example flowchart of a parametric modeling method for an integral die-cast part structure based on geometric partitioning, provided in an embodiment of this application;

[0064] Figure 3 A schematic diagram of a parametric modeling device for an integral die-casting part based on geometric partitioning, provided in an embodiment of this application;

[0065] Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0068] This application provides a method and related equipment for parametric modeling of integrated die-cast parts based on geometric partitioning, relating to the field of 3D modeling technology. The method and equipment provided in this application can be applied to terminals, servers, or software running on either terminal or server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing a parametric modeling method for integrated die-cast parts based on geometric partitioning, but is not limited to the above forms.

[0069] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0070] Reference Figure 1 This application provides a parametric modeling method for integral die-cast parts based on geometric partitioning. This method may include, but is not limited to, steps S100 to S140, as follows:

[0071] S100: Perform feature aggregation on each node in the CAE model of the integrated die-cast part structure to obtain multiple super nodes of the CAE model;

[0072] S110: Determine the nearest neighboring supernode for each supernode and establish an edge index;

[0073] S120: Construct a graph network model corresponding to the integrated die-casting structure based on each of the super nodes and the edge indices;

[0074] S130: Using the graph network model as training samples and the target performance parameters corresponding to the CAE model as training labels, train the graph convolutional network.

[0075] S140: When the prediction error of the graph convolutional network reaches the preset optimal value, the current graph network model is determined as the optimal graph network model of the integrated die-casting structure.

[0076] Optionally, the step of performing feature aggregation on each node in the CAE model of the integral die-cast part structure to obtain multiple supernodes of the CAE model includes the following steps:

[0077] Extract the CAE model corresponding to different structural parameters from the parameterized database of the integrated die-cast part structure;

[0078] Based on the finite element mesh nodes of each CAE model, each CAE model is discretized into a node set containing multiple nodes;

[0079] Initial feature vectors are extracted from each of the node sets, and the initial feature vectors are converted into final feature matrices using a weighted coefficient matrix;

[0080] Using the final feature matrix as the clustering target, all nodes in each node set are clustered and assigned to the corresponding clusters.

[0081] Feature aggregation is performed on the nodes in each cluster to obtain the supernode corresponding to each cluster.

[0082] Optionally, extracting initial feature vectors for each of the node sets includes the following steps:

[0083] The initial feature vector is extracted from all nodes in each of the aforementioned node sets as follows:

[0084] ;

[0085] in, For nodes The initial feature vector, Let x, y, and z be the initial spatial coordinates in the x, y, and z directions, respectively. For nodes The initial thickness, For nodes Initial material property encoding;

[0086] The initial feature vector is normalized, and the initial spatial coordinates, the initial thickness, and the initial material property codes are normalized.

[0087] The process of converting the initial feature vector into the final feature matrix using a weighted coefficient matrix includes the following steps:

[0088] Construct the weighted coefficient matrix for:

[0089] ;

[0090] in, Represents the spatial nearest neighbor weight. Represents the thickness nearest neighbor weight. Indicates the nearest neighbor weight of materials;

[0091] Using the weighted coefficient matrix The normalized initial feature vector The feature information in the matrix is ​​weighted to obtain the final feature matrix. ,in These are the final spatial coordinates in the x, y, and z directions, respectively. For nodes The final thickness, For nodes The final material property code.

[0092] Optionally, before performing feature aggregation on the nodes in each of the clusters to obtain the supernode corresponding to each cluster, the method further includes the following steps:

[0093] The connectivity coefficient matrix is ​​constructed based on the edges between the finite element mesh nodes of the CAE model. ;in, Represents a node There is a connection between them;

[0094] For each node, check whether connected adjacent nodes belong to the same cluster. If the number of nodes belonging to another cluster reaches a first set percentage or the number of nodes with different material properties reaches a second set percentage, return to the step of using the final feature matrix as the clustering target to cluster all nodes in each node set and assign them to the corresponding cluster, until the number of nodes belonging to another cluster is less than the first set percentage or the number of nodes with different material properties is less than the second set percentage.

[0095] Optionally, the step of performing feature aggregation on the nodes in each cluster to obtain the supernode corresponding to each cluster includes the following steps:

[0096] For each cluster, the coordinates of the center point of all nodes are taken as the geometric features of the supernode. ;

[0097] The thickness of all nodes in each cluster is averaged, and the average thickness is taken as the thickness feature of the supernode. ;

[0098] For all nodes in each cluster, identify the node with the largest number of nodes and the same material properties, and then use the material properties as the material features of the supernode. Finally, the supernode is obtained. .

[0099] Optionally, the step of constructing the graph network model corresponding to the integral die-casting structure based on each of the supernodes and the edge indices includes the following steps:

[0100] The graph network model is defined as follows: ;

[0101] in:

[0102] Represents the set of supernodes. Each of the supernodes serves as a node in the graph network model;

[0103] Denotes the set of edges. , representing the edge index between each of the supernodes, the edge index serving as the edge of the graph network model. ;

[0104] Represents the adjacency matrix. , This represents the topological relationship of the integral die-cast component structure. If there is no connection between two nodes, then... If the corresponding element is 0, then if a connection exists, then... The corresponding element is 1;

[0105] Represents a collection of node attributes. .

[0106] Optionally, before extracting the CAE models corresponding to different structural parameters from the parameterized database of the integral die-cast part structure, the method includes the following steps:

[0107] On the base plate of the floor after integral die casting of the integral die-cast structure, a variable topological mesh is defined, the mesh dimension of which is M*N, and key control nodes are defined in the topological mesh. ,in , which are parametric coordinates relative to the base edge;

[0108] Set activation parameters The activation parameters Key control nodes for characterizing the generation of ribs in the topological mesh. ;

[0109] A topology mesh is generated in the shell CAD file of the integral die-cast floor based on the topology mesh;

[0110] According to the activation parameters Generate rib paths;

[0111] The rib path is scanned and shaped to create a three-dimensional rib solid.

[0112] The ribbed entity and the shell of the integral die-cast floor are subjected to Boolean addition operation to form a ribbed integral die-cast floor model;

[0113] The ribbed integral die-cast floor model was meshed using full-shell elements, and the thickness of the ribs was defined in the CAE environment. ,thickness With the activation parameters Correspondingly;

[0114] A torsional stiffness condition is applied in a CAE environment, and the process of calling the CAE solver, setting analysis cards, applying loads and boundary conditions, and extracting results is scripted. The calculation results are the target performance parameters; wherein, the target performance parameters include the maximum displacement and the maximum equivalent stress.

[0115] Create the parameterized database, and store the corresponding structural parameters, target performance parameters, and CAE model as a data record in the parameterized database.

[0116] The following sections will provide a detailed description and explanation of some optional embodiments of this application, using specific application examples.

[0117] This embodiment aims to solve the curse of dimensionality in graph convolutional networks caused by an excessive number of nodes, while preserving the key geometric and topological information of the original structure to the greatest extent possible, thereby achieving efficient and accurate parametric design. Specifically, it provides a parametric modeling method for integrated die-cast parts based on geometric partitioning.

[0118] Reference Figure 2 This embodiment may include the following technical solutions:

[0119] Step 1: On the base plate of the integral die-cast floor, define a variable topology mesh with dimensions M*N, and define a set of key control nodes in the topology mesh. ,in These are parametric coordinates relative to the base edge. Adjusting the positions of these nodes allows control over the density of the rib layout.

[0120] Step 2, set activation parameters The activation parameters are used to characterize which key control nodes are in the topology mesh. It will generate ribs, specifically a set of binary vector matrices, for example This indicates that ribs are generated at control points 2 and 3, and control points 1 and 3.

[0121] Step 3: Generate a topology mesh in the CAD file of the floor shell after the foundation is die-cast, based on the topology mesh parameters defined in Step 1.

[0122] Step 4, according to the activation parameters defined in Step 2 Generate rib paths;

[0123] Step 5: Scan and shape the rib path generated in Step 4 to create a three-dimensional rib solid;

[0124] Step 6: Perform an automatic Boolean addition operation on all the generated ribbed entities and the base integral die-cast floor shell to form a complete ribbed integral die-cast floor model;

[0125] Step 7: Automatic mesh generation is performed on the ribbed, integrally die-cast floor model using full-shell elements. Since the ribs are integral with the matrix, there is no need to handle contact issues, resulting in fast and stable mesh generation. The thickness of the ribs is then defined in the CAE environment. ,thickness With the activation parameters defined in step 2 Correspondingly;

[0126] Step 8: Automatically apply torsional stiffness conditions in the CAE environment, and script the process of calling the CAE solver, setting analysis cards, applying loads and boundary conditions, and extracting results. The calculation results include the maximum displacement and the maximum equivalent stress. After generating the geometric model and finite element model, the calculation is automatically submitted, and key target performance parameters are extracted from the result file.

[0127] Step 9: Create a parameterized database for the integrated die casting. The table structure in the database corresponds to the parameters defined in Step 1 and Step 2. After completing Steps 3-7, the combination of input parameters, output performance results, and model file are automatically stored in the database as a complete data record.

[0128] Step 10: Key control nodes of the integrated die-cast part Activation parameters Thickness of the reinforcing bars Perform full factorial design (DOE) to automatically generate a massive number of input parameter combinations, drive steps 3-8, and obtain a complete integrated die-casting parametric database.

[0129] Step 11: Extract the CAE model of the die-casting part corresponding to different structural parameters from the parameterized database of the integrated die-casting part. Based on the finite element mesh nodes, discretize it into a set containing N nodes. Each node contains its spatial coordinates and is appended with the physical properties (thickness, material) of its element, which together constitute the initial feature vector of the node. The initial feature vector of each node is shown in the following formula:

[0130] ;

[0131] In the formula, For nodes eigeninitial vectors, These are the coordinates in the x, y, and z directions, respectively. For the thickness of the node, Encode the material properties of the nodes;

[0132] Step 12, the initial feature vectors of the nodes obtained in Step 11 Normalization is performed to normalize the spatial coordinates and physical attribute information to... Within the range;

[0133] Step 13, construct the weighted coefficient matrix ,in, Represents the spatial nearest neighbor weight. Thickness nearest neighbor weight, Represents the nearest neighbor weight of materials, if A height indicates that spatial proximity is dominant. A high value indicates that the thickness is dominated by the nearest neighbor property. A high value indicates that the material's proximity is dominant;

[0134] Step 14, using For the normalized initial eigenvector The feature information in the matrix is ​​weighted to obtain the final feature matrix. ,in These are the coordinates in the x, y, and z directions, respectively. For the thickness of the node, Encode the material properties of the nodes;

[0135] Step 15, using the weighted final feature matrix To determine the clustering target, the K-means clustering algorithm is applied to all nodes of the integral die-cast part, dividing the N nodes into K clusters.

[0136] Step 16: Extract the CAE model of the die-casting part corresponding to different structural parameters from the parameterized database of the integrated die-casting part, and use the edges between the finite element mesh nodes as the connectivity coefficient matrix. ,in Represents a node There is a connection between them;

[0137] Step 17: For each node, check whether its connected adjacent nodes belong to the same cluster. If most connected nodes belong to another cluster or have different material properties, repeat step 15 to re-cluster the nodes, thereby ensuring that the clustering results are feasible in engineering.

[0138] Step 18: Perform feature aggregation on the nodes in each cluster to establish supernodes. The spatial coordinates of all nodes in the cluster are averaged, and the coordinates of the center point are taken as the geometric features of the supernode. The average thickness of all nodes in the cluster is taken as the thickness feature of the supernodes. The material properties of the majority of nodes in the cluster are used as the material features of the supernodes. ;

[0139] Step 19, for each supernode Find the nearest neighbor supernode using the KNN algorithm. and establish edge indexes. ;

[0140] Step 20: Analyze the edge index established in step 19. Perform an inspection to ensure the super node is active. With adjacent supernodes The connections between them are bidirectional, and duplicate edge indices are removed;

[0141] Step 21: Aggregate the established supernodes and edge indices to define a graph network model of the integrated die-casting part structure. The structure of this graph network model is as follows: ;

[0142] Its specific meaning is defined as follows:

[0143] Represents the set of supernodes. In this embodiment, the supernodes aggregated in step 18 are extracted as nodes of the graph network;

[0144] Denotes the set of edges. In this embodiment, the connections between the supernode vertices in step 19 are extracted as edges of the graph network. ;

[0145] Represents the adjacency matrix. ,matrix This represents the topological relationship of a single die-cast part. If there is no connection between two nodes, the element is 0; if there is a connection, the element is 1.

[0146] Represents a collection of node attributes. ;

[0147] Step 22: Use the graph network model generated in step 21 as the input of the graph convolutional network, and at the same time search the parameterized database of the integrated die casting for the target performance parameters (maximum displacement, maximum equivalent stress) of the corresponding structure as the output.

[0148] Step 23, network parameter initialization;

[0149] Step 24: Input the training samples (the graph network model generated in step 21) into the graph convolutional network for training, and calculate the prediction error through layer-by-layer forward propagation;

[0150] Step 25: Backpropagate the prediction error generated in Step 24 using the regression function, and iteratively update the network parameters of the graph convolutional network using the Adam optimization algorithm to minimize the cross-entropy loss function. If the preset optimal value is reached, proceed to Step 26; otherwise, return to Step 24, and repeat this process until the parameters of the graph convolutional network reach the optimal value, and store the optimal graph convolutional network.

[0151] Step 26: Output the graph network model optimized by the graph convolutional network. Due to the significant reduction in input dimension, the graph convolutional network can be lightweight, thereby enabling efficient training and inference, and completing the performance prediction and shape optimization of the integrated die-cast part.

[0152] The advantages of this embodiment are:

[0153] 1. Using a graph network to parametrically characterize integral die castings is more effective in representing the complex structural features of integral die castings compared to the traditional method of using a limited set of manually defined parameters;

[0154] 2. By constructing supernodes, the input dimension of the performance prediction model for integrated die-cast parts is reduced, thereby significantly reducing the computational complexity of the convolution process in the graph convolutional network framework;

[0155] 3. The geometry-based partitioning results are easy for engineers to understand and have strong interpretability.

[0156] Reference Figure 3 This application also provides a parametric modeling device for integral die-casting structures based on geometric partitioning, which can realize the above-mentioned parametric modeling method for integral die-casting structures based on geometric partitioning. The device includes:

[0157] The node aggregation unit is used to perform feature aggregation on each node in the CAE model of the integral die-cast part structure to obtain multiple super nodes of the CAE model.

[0158] An edge index building unit is used to determine the nearest neighboring supernode for each supernode and to build an edge index;

[0159] The graph network model construction unit is used to construct a graph network model corresponding to the integral die-casting structure based on each of the super nodes and the edge indices.

[0160] The graph convolutional network training unit is used to train the graph convolutional network using the graph network model as training samples and the target performance parameters corresponding to the CAE model as training labels.

[0161] The model determination unit is used to determine the current graph network model as the optimal graph network model for the integrated die-casting structure when the prediction error of the graph convolutional network reaches a preset optimal value.

[0162] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0163] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method of this application. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0164] It is understood that the content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the methods of this application, and the beneficial effects achieved are the same as those achieved by the methods of this application.

[0165] Figure 4 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0166] The processor 101 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0167] The memory 102 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 102 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 102 and is called and executed by the processor 101.

[0168] Input / output interface 103 is used to implement information input and output;

[0169] The communication interface 104 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0170] Bus 105 transmits information between various components of the device (e.g., processor 101, memory 102, input / output interface 103, and communication interface 104);

[0171] The processor 101, memory 102, input / output interface 103 and communication interface 104 are connected to each other within the device via bus 105.

[0172] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method of this application.

[0173] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0174] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0175] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0176] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0177] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0178] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0179] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0180] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0181] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0182] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0183] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0184] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0185] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0186] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A parametric modeling method for integral die-cast parts based on geometric partitioning, characterized in that, The method includes the following steps: Feature aggregation is performed on each node in the CAE model of the integral die-cast part structure to obtain multiple super nodes of the CAE model; Determine the nearest neighboring supernode for each supernode and establish an edge index; A graph network model corresponding to the integrated die-casting structure is constructed based on each of the super nodes and the edge indices. The graph network model is used as the training sample, and the target performance parameters corresponding to the CAE model are used as the training labels to train the graph convolutional network. When the prediction error of the graph convolutional network reaches the preset optimal value, the current graph network model is determined as the optimal graph network model of the integrated die-casting structure. The process of performing feature aggregation on each node in the CAE model of the integral die-cast part structure to obtain multiple supernodes of the CAE model includes the following steps: Extract the CAE model corresponding to different structural parameters from the parameterized database of the integrated die-cast part structure; Based on the finite element mesh nodes of each CAE model, each CAE model is discretized into a node set containing multiple nodes; Initial feature vectors are extracted from each of the node sets, and the initial feature vectors are converted into final feature matrices using a weighted coefficient matrix; Using the final feature matrix as the clustering target, all nodes in each node set are clustered and assigned to the corresponding clusters. Feature aggregation is performed on the nodes in each cluster to obtain the supernode corresponding to each cluster; The step of performing feature aggregation on nodes in each cluster to obtain the supernode corresponding to each cluster includes the following steps: For each cluster, the coordinates of the center point of all nodes are taken as the geometric features of the supernode. ; The thickness of all nodes in each cluster is averaged, and the average thickness is taken as the thickness feature of the supernode. ; For all nodes in each cluster, identify the node with the largest number of nodes and the same material properties, and then use the material properties as the material features of the supernode. Finally, the supernode is obtained. .

2. The method for parametric modeling of integral die-cast parts based on geometric partitioning according to claim 1, characterized in that, The step of extracting initial feature vectors for each of the node sets includes the following steps: The initial feature vector is extracted from all nodes in each of the aforementioned node sets as follows: ； in, For nodes The initial feature vector, Let x, y, and z be the initial spatial coordinates in the x, y, and z directions, respectively. For nodes The initial thickness, For nodes Initial material property encoding; The initial feature vector is normalized, and the initial spatial coordinates, the initial thickness, and the initial material property codes are normalized. The process of converting the initial feature vector into the final feature matrix using a weighted coefficient matrix includes the following steps: Construct the weighted coefficient matrix for: ； in, Represents the spatial nearest neighbor weight. Represents the thickness nearest neighbor weight. Indicates the nearest neighbor weight of materials; Using the weighted coefficient matrix The normalized initial feature vector The feature information in the matrix is ​​weighted to obtain the final feature matrix. ,in These are the final spatial coordinates in the x, y, and z directions, respectively. For nodes The final thickness, For nodes The final material property code.

3. The method for parametric modeling of integral die-cast parts based on geometric partitioning according to claim 1, characterized in that, Before performing feature aggregation on the nodes in each cluster to obtain the supernode corresponding to each cluster, the method further includes the following steps: The connectivity coefficient matrix is ​​constructed based on the edges between the finite element mesh nodes of the CAE model. ;in, Represents a node There is a connection between them; For each node, check whether connected adjacent nodes belong to the same cluster. If the number of nodes belonging to another cluster reaches a first set percentage or the number of nodes with different material properties reaches a second set percentage, return to the step of using the final feature matrix as the clustering target to cluster all nodes in each node set and assign them to the corresponding cluster, until the number of nodes belonging to another cluster is less than the first set percentage or the number of nodes with different material properties is less than the second set percentage.

4. The method for parametric modeling of integral die-cast parts based on geometric partitioning according to claim 1, characterized in that, The process of constructing the graph network model corresponding to the integral die-casting structure based on each of the supernodes and edge indices includes the following steps: The graph network model is defined as follows: ; in: Represents the set of supernodes. Each of the supernodes serves as a node in the graph network model; Denotes the set of edges. , representing the edge index between each of the supernodes, the edge index serving as the edge of the graph network model. ; Represents the adjacency matrix. , This represents the topological relationship of the integral die-cast component structure. If there is no connection between two nodes, then... If the corresponding element is 0, then if a connection exists, then... The corresponding element is 1; Represents a collection of node attributes. .

5. A method for parametric modeling of an integral die-cast part structure based on geometric partitioning according to any one of claims 1 to 4, characterized in that, Before extracting the CAE models corresponding to different structural parameters from the parameterized database of the integral die-cast part structure, the method includes the following steps: On the base plate of the floor after integral die casting of the integral die-cast structure, a variable topological mesh is defined, the mesh dimension of which is M*N, and key control nodes are defined in the topological mesh. ,in , which are parametric coordinates relative to the base edge; Set activation parameters The activation parameters Key control nodes for characterizing the generation of ribs in the topological mesh. ; A topology mesh is generated in the shell CAD file of the integral die-cast floor based on the topology mesh; According to the activation parameters Generate rib paths; The rib path is scanned and shaped to create a three-dimensional rib solid. The ribbed entity and the shell of the integral die-cast floor are subjected to Boolean addition operation to form a ribbed integral die-cast floor model; The ribbed integral die-cast floor model was meshed using full-shell elements, and the thickness of the ribs was defined in the CAE environment. ,thickness With the activation parameters Correspondingly; A torsional stiffness condition is applied in a CAE environment, and the process of calling the CAE solver, setting analysis cards, applying loads and boundary conditions, and extracting results is scripted. The calculation results are the target performance parameters; wherein, the target performance parameters include the maximum displacement and the maximum equivalent stress. Create the parameterized database, and store the corresponding structural parameters, target performance parameters, and CAE model as a data record in the parameterized database.

6. A parametric modeling device for integrated die-casting structures based on geometric partitioning, characterized in that, The device is used to implement the parametric modeling of an integral die-casting structure based on geometric partitioning as described in claim 1, and the device includes: The node aggregation unit is used to perform feature aggregation on each node in the CAE model of the integral die-cast part structure to obtain multiple super nodes of the CAE model. An edge index building unit is used to determine the nearest neighboring supernode for each supernode and to build an edge index. The graph network model construction unit is used to construct a graph network model corresponding to the integrated die-casting structure based on each of the super nodes and the edge indices. The graph convolutional network training unit is used to train the graph convolutional network using the graph network model as training samples and the target performance parameters corresponding to the CAE model as training labels. The model determination unit is used to determine the current graph network model as the optimal graph network model for the integrated die-casting structure when the prediction error of the graph convolutional network reaches a preset optimal value.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

Images