Building structure seismic dynamic response calculation method and device based on time integral graph network, equipment and medium

Abstract

The application discloses a building structure seismic dynamic response calculation method and device based on a time integral graph network, equipment and medium, and relates to the technical field of civil engineering. The method comprises the following steps: obtaining a finite element model of a target building structure and seismic time history data, constructing a graph structure and obtaining initial graph data. Then, the initial graph data is mapped to graph input data, which is input into a preset seismic response model for processing, and finally, a seismic response displacement sequence is obtained as a dynamic response result. The preset model comprises a graph encoder, a time integral calculation module and a graph decoder. The modules work cooperatively to realize efficient and accurate seismic response prediction, effectively reduce the consumption of computing resources and the calculation time, maintain high precision and generalization, and meet the real-time simulation demand.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of civil engineering, and in particular to a building structure seismic dynamic response calculation method, device and equipment based on a time integral graph network and a medium. BACKGROUND

[0002] In the field of civil engineering, application scenarios such as structural health monitoring, seismic performance evaluation, and rapid response after disasters require the ability to instantly obtain the response (such as displacement, acceleration) of large buildings or bridges under the action of seismic dynamics. The traditional numerical simulation method, represented by the finite element method, discretizes the continuous structure into a finite number of elements and nodes, establishes mass, damping, and stiffness matrices, and gradually solves the motion differential equation combined with the time integration algorithm to obtain the dynamic response of the structure. Although this method has high precision, it is time-consuming and resource-consuming, and it is difficult to meet the real-time requirements. The method based on neural networks uses models such as multilayer perceptron, convolutional neural network, or recurrent neural network to directly learn the mapping relationship from a large number of input (such as load history) - output (such as displacement response) data pairs. For example, using a long short-term memory network to process time series load data to predict the structure's response at future time, the accuracy of the model is highly dependent on the number and coverage of the training data. When encountering load conditions or structural states not included in the training set, the prediction performance of the model will drop sharply, and the generalization ability is poor. In addition, the completely "black box" model lacks interpretability and cannot guarantee that its prediction results meet the basic physical laws.

[0003] Therefore, there is an urgent need for a building structure seismic dynamic response calculation method to achieve high-precision, real-time dynamic response calculation of the structure while having lower requirements for training data. SUMMARY

[0004] The main purpose of the present application is to provide a building structure seismic dynamic response calculation method, device, equipment and medium based on a time integral graph network, aiming to solve the technical problem of how to achieve high-precision, real-time dynamic response calculation of the structure while having lower requirements for training data.

[0005] To achieve the above-mentioned purpose, the present application provides a building structure seismic dynamic response calculation method based on a time integral graph network, comprising:

[0006] obtaining a finite element model of a target building structure and seismic motion time history data within a prediction period, wherein the finite element model includes node information, element information, mass matrix, damping matrix, and stiffness matrix of the target building structure;

[0007] constructing a graph structure according to the node information and element information of the finite element model, and obtaining initial graph data;

[0008] mapping the mass matrix, the damping matrix and the stiffness matrix and the ground motion time history data based on the initial graph data to obtain graph input data;

[0009] inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integration module comprises a parameter calculation unit, an edge update unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer;

[0010] taking the seismic response displacement sequence as a seismic dynamic response result of the target building structure.

[0011] In an embodiment, the step of inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, comprises:

[0012] performing dimension reduction processing on the graph input data by the graph encoder to obtain a low-dimensional hidden graph;

[0013] performing multi-step recursion calculation on the low-dimensional hidden graph by the time integration calculation module to obtain updated hidden graph node features at each time step;

[0014] decoding and restoring the updated hidden graph node features by the graph decoder to obtain the seismic response displacement sequence of the target building structure.

[0015] In an embodiment, the step of performing dimension reduction processing on the graph input data by the graph encoder to obtain a low-dimensional hidden graph comprises:

[0016] inputting graph node features in the graph input data into a residual network module, performing feature compression and dimension reduction on the graph node features by the residual block to obtain low-dimensional node features, wherein the dimension of the low-dimensional node features is a preset proportion of the dimension of the graph node features;

[0017] inputting graph edge features in the graph input data into a convolution residual network module to perform convolution operation and dimension reduction on the graph edge features to obtain low-dimensional edge features, wherein the dimension of the low-dimensional edge features is a preset proportion of the dimension of the graph edge features;

[0018] The adjacency matrix in the graph input data is sparsified to obtain the processed adjacency matrix;

[0019] The low-dimensional node features, low-dimensional edge features, and the processed adjacency matrix are combined to obtain a low-dimensional hidden graph.

[0020] In one embodiment, the step of performing multi-step recursive calculations on the low-dimensional hidden graph through the time integration calculation module to obtain the updated hidden graph node features at each time step includes:

[0021] Extract low-dimensional edge features and low-dimensional node features from the low-dimensional hidden graph;

[0022] Set an edge update function and a node update function. The edge update function adopts a convolutional neural network, which includes a second preset number of convolutional layers and a third preset number of activation layers. The node update function adopts a fully connected neural network, which includes a second preset number of fully connected layers and a first preset number of activation layers.

[0023] The low-dimensional edge features are input into the edge update function to optimize the low-dimensional edge features, resulting in updated edge features.

[0024] The low-dimensional node features are recursively updated in the first step to obtain the hidden graph node features updated in the first time step.

[0025] Based on the updated edge features, the low-dimensional node features of adjacent graph nodes in the low-dimensional hidden graph are aggregated through the node update function to obtain the graph nodes for the next time step.

[0026] The updated hidden graph node features of the first time step are used as the initial node features of the graph nodes of the next time step. The update and aggregation steps are repeated until the calculation of all time steps in the prediction period is completed, and the updated hidden graph node features of each time step are obtained.

[0027] In one embodiment, the step of obtaining the seismic response displacement sequence of the target building structure by decoding and reconstructing the target building structure based on the updated hidden graph node features using the graph decoder includes:

[0028] The updated hidden graph node features at each time step are sequentially input into the fully connected layer of the graph decoder for dimension recovery and feature mapping to obtain the physical domain features corresponding to each time step.

[0029] Displacement information is extracted from the physical domain features corresponding to each time step. The displacement information includes the x-direction displacement, y-direction displacement, and z-direction displacement of the target building structure's corresponding structural degrees of freedom at the current time step.

[0030] validating the displacement information of each time step to obtain a validation result;

[0031] when the numerical range of the displacement information is within a preset displacement threshold, integrating the displacement information of all effective time steps in chronological order to obtain a seismic response displacement sequence of the target building structure.

[0032] In an embodiment, the step of constructing a graph structure according to the node information and element information of the finite element model and obtaining initial graph data comprises:

[0033] extracting structural degrees of freedom from the node information of the finite element model, and determining the physical coordinates and mechanical properties corresponding to each structural degree of freedom;

[0034] mapping each structural degree of freedom to a graph node, and associating the physical coordinates and mechanical properties of the corresponding structural degree of freedom with the graph node;

[0035] extracting structural element connection relationships from the element information of the finite element model, and determining the structural degree of freedom range connected by each structural element;

[0036] setting a graph edge between the corresponding graph nodes according to the structural degree of freedom range connected by the structural element, and associating the type and size of the corresponding structural element with the graph edge;

[0037] generating an adjacency matrix based on the stiffness matrix of the finite element model, wherein an element value of the first preset value in the adjacency matrix indicates that the corresponding graph nodes are mechanically associated, and an element value of the second preset value indicates that there is no mechanical association;

[0038] integrating the graph nodes, the graph edges, and the adjacency matrix to obtain initial graph data.

[0039] In an embodiment, the step of mapping the mass matrix, the damping matrix, and the stiffness matrix and the ground motion time history data based on the initial graph data to obtain graph input data comprises:

[0040] extracting a graph edge list from the initial graph data, and determining the structural element corresponding to each graph edge;

[0041] mapping the submatrix corresponding to the structural element in the mass matrix, the submatrix corresponding to the structural element in the damping matrix, and the submatrix corresponding to the structural element in the stiffness matrix to the first edge feature component, the second edge feature component, and the third edge feature component of each graph edge;

[0042] combining the first edge feature component, the second edge feature component, and the third edge feature component to obtain graph edge features;

[0043] extracting a graph node list from the initial graph data, determining a structural degree of freedom corresponding to each of the graph nodes;

[0044] mapping a load time sequence corresponding to the structural degree of freedom in the ground motion time history data into a node feature component of each of the graph nodes;

[0045] obtaining initial displacement data of the target building structure at a starting moment of a prediction period;

[0046] adding the initial displacement data into the node feature component corresponding to the graph nodes to obtain graph node features;

[0047] integrating the graph edge features, the graph node features and an adjacency matrix in the initial graph data to obtain graph input data.

[0048] In addition, to achieve the above-mentioned purpose, the application further provides a building structure seismic dynamic response calculation device based on a time integral graph network, which comprises:

[0049] an acquisition module, configured to acquire a finite element model of a target building structure and ground motion time history data in a prediction period, wherein the finite element model comprises node information, element information, a mass matrix, a damping matrix and a stiffness matrix of the target building structure;

[0050] a construction module, configured to construct a graph structure according to the node information and the element information of the finite element model and obtain initial graph data;

[0051] a mapping module, configured to map the mass matrix, the damping matrix, the stiffness matrix and the ground motion time history data based on the initial graph data to obtain graph input data;

[0052] a processing module, configured to input the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integral calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integral module comprises a parameter calculation unit, an edge updating unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer;

[0053] a result module, configured to take the seismic response displacement sequence as a seismic dynamic response result of the target building structure.

[0054] In addition, to achieve the above object, the application further provides a storage medium, which is a computer readable storage medium, and a computer program is stored on the storage medium, and the computer program realizes the steps of the building structure seismic dynamic response calculation method based on a time integral graph network when executed by a processor.

[0055] In addition, to achieve the above object, the application further provides a computer program product, which comprises a computer program, and the computer program realizes the steps of the building structure seismic dynamic response calculation method based on a time integral graph network when executed by a processor.

[0056] The application obtains a finite element model of a target building structure and seismic time history data, constructs a graph structure and obtains initial graph data. Then, the initial graph data is mapped to graph input data, which is input into a preset seismic response model for processing, and finally, a seismic response displacement sequence is obtained as a dynamic response result. The preset model comprises a graph encoder, a time integral calculation module and a graph decoder, and the modules work cooperatively to realize efficient and accurate seismic response prediction, effectively reduce the consumption of computing resources and the calculation time, while maintaining high precision and generalization, and meet the real-time simulation demand. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor.

[0058] Figure 1 Flowchart of a first embodiment of the building structure seismic dynamic response calculation method based on a time integral graph network of the application;

[0059] Figure 2 21-layer finite element model frame structure diagram of the first embodiment of the building structure seismic dynamic response calculation method based on a time integral graph network of the application;

[0060] Figure 3 Training seismic load time history curve diagram of the first embodiment of the building structure seismic dynamic response calculation method based on a time integral graph network of the application;

[0061] Figure 4 Predicted seismic load time history curve diagram of the first embodiment of the building structure seismic dynamic response calculation method based on a time integral graph network of the application;

[0062] Figure 5 Flowchart of a second embodiment of the building structure seismic dynamic response calculation method based on a time integral graph network of the application;

[0063] Figure 6 A module structure schematic diagram of a building structure seismic dynamic response calculation device based on a time integral graph network according to the present application;

[0064] Figure 7 A device structure schematic diagram of a hardware running environment related to a building structure seismic dynamic response calculation method based on a time integral graph network according to an embodiment of the present application.

[0065] The object implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0066] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0067] In order to better understand the technical solutions of the present application, the following will be described in detail in conjunction with the drawings and specific embodiments of the specification.

[0068] In the era of Industry 4.0, the importance of digital twin technology in industrial production and operation is increasingly prominent, and the real-time evolution of the structure state becomes a key link. However, the traditional numerical method has the problems of high consumption of computing resources and long time-consuming in calculating the dynamic response of the structure, which is difficult to meet the real-time requirements. In recent years, the method based on neural network has gradually emerged, but the existing method is either highly dependent on data or inefficient in dealing with complex structures.

[0069] Therefore, the application provides a building structure seismic dynamic response calculation method based on a time integral graph network to solve the above problems. The main solution of the embodiment of the application is: obtaining a finite element model of a target building structure and seismic time history data in a prediction period, wherein the finite element model comprises node information, element information, a mass matrix, a damping matrix and a stiffness matrix of the target building structure; constructing a graph structure according to the node information and the element information of the finite element model, and obtaining initial graph data; mapping the mass matrix, the damping matrix, the stiffness matrix and the seismic time history data based on the initial graph data to obtain graph input data; inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integral calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integral module comprises a parameter calculation unit, an edge update unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer; and taking the seismic response displacement sequence as a seismic dynamic response result of the target building structure.

[0070] Based on the above, the application further provides a building structure seismic dynamic response calculation method based on a time integral graph network, which is described with reference to Figure 1 , Figure 1 The flowchart of the first embodiment of the building structure seismic dynamic response calculation method based on the time integral graph network of the application is shown in the figure. In this embodiment, the building structure seismic dynamic response calculation method based on the time integral graph network comprises steps S10-S50.

[0071] Step S10: obtaining a finite element model of a target building structure and seismic time history data in a prediction period.

[0072] It should be noted that the target building structure refers to various civil, public, or industrial building structures for which seismic dynamic response prediction is required, such as bridges and residential buildings. The finite element model includes the node information, element information, mass matrix, damping matrix, and stiffness matrix of the target structure. First, in terms of acquiring the finite element model, the computing system needs to retrieve the corresponding finite element model file from the design database or engineering simulation platform of the target building structure. This finite element model must fully cover the geometric features and mechanical properties of the structure. The node information includes the three-dimensional spatial coordinates (accurate to the millimeter level) of all key nodes of the structure and the structural degrees of freedom (such as translational and rotational degrees of freedom) corresponding to each node, and the node number and the structural component to which it belongs (such as beam node and column node) must be labeled. The element information must clearly specify the finite element type (such as beam element, column element, floor slab element), cross-sectional dimensions (such as the cross-sectional height and width of beam element), material parameters (such as the elastic modulus of steel and Poisson's ratio of concrete), and the connection relationship between elements and nodes for each structural component, ensuring that it can accurately reflect the actual topology of the structure. Simultaneously, the finite element model must also include three core mechanical matrices: the mass matrix, the damping matrix, and the stiffness matrix. The mass matrix, using either a uniform mass matrix or a lumped mass matrix, is generated based on the mass distribution characteristics of the structural components, reflecting the inertial response of the structure under dynamic loads. The damping matrix, using Rayleigh damping, is generated through a pre-defined... , coefficients (e.g.) =0.01、 =0.01) is calculated by combining the mass matrix and stiffness matrix, reflecting the energy dissipation effect during structural vibration; the stiffness matrix is ​​generated by assembling the element stiffness matrix, reflecting the structure's ability to resist deformation, and its dimension is consistent with the total number of degrees of freedom of the structure.

[0073] Furthermore, regarding the acquisition of seismic ground motion time history data, the calculation system needs to select seismic ground motion time history data that conforms to the local seismic fortification intensity (e.g., 7 degrees, 8 degrees) and design earthquake group, based on the seismic ground motion parameter zoning map of the area where the target building structure is located. This data needs to be presented in the form of acceleration time history curves, with a time span covering the preset response prediction period (e.g., 10s, 15s), and a time step set to a preset small step size (e.g., 0.02s) to ensure accurate reflection of the variation of seismic load over time.

[0074] Step S20: Construct a graph structure based on the node information and element information of the finite element model, and obtain initial graph data.

[0075] It should be noted that this process aims to transform the discretization concept in traditional finite element analysis into a topological representation that can be processed by graph neural networks.

[0076] Further, step S20 further comprises: first, extracting structural degrees of freedom from the node information of the finite element model, and determining the corresponding physical coordinates and mechanical properties of each structural degree of freedom. Specifically, the structural degree of freedom is an independent parameter that describes the possible deformation or movement of the structure under dynamic load, for example, in three-dimensional space, each node can have three translational degrees of freedom and three rotational degrees of freedom. The physical coordinates are used to determine the position of the node in space, while the mechanical properties include mass, damping and other parameters, which are essential for describing the response behavior of the structure under dynamic load.

[0077] Then, each structural degree of freedom is mapped to a graph node, and the graph node is associated with the corresponding physical coordinates and mechanical properties of the structural degree of freedom. Specifically, the graph node is the basic unit of the graph structure, and each graph node represents a structural degree of freedom, and its feature vector contains the physical coordinates and mechanical properties of the degree of freedom. This mapping method enables the graph neural network to directly process physical information related to the structural degree of freedom, providing a basis for subsequent dynamic response prediction.

[0078] Next, the structural element connection relationship is extracted from the element information of the finite element model, and the structural degree of freedom range connected by each structural element is determined. Specifically, the element information describes the connection relationship between nodes and the physical properties of the element, such as stiffness and damping. The structural element connection relationship defines which nodes are connected through the element, and the structural degree of freedom range specifies the specific degrees of freedom involved in these connection relationships. For example, a beam element may connect two nodes and involve the translational degrees of freedom of these nodes.

[0079] Subsequently, according to the structural degree of freedom range connected by the structural element, the graph edges between the corresponding graph nodes are set, and the graph edges are associated with the type and size of the corresponding structural element. Specifically, the graph edge represents the connection relationship between the graph nodes, and its feature vector contains information such as the type and size of the structural element. These information is essential for describing the interaction between nodes, for example, different types and sizes of elements may have different effects on the force transmission between nodes.

[0080] Then, an adjacency matrix is generated based on the stiffness matrix of the finite element model, wherein the element value of the adjacency matrix is a first preset value indicating that there is a mechanical correlation between the corresponding graph nodes, and the element value is a second preset value indicating no mechanical correlation. Specifically, the stiffness matrix is an important component of the finite element model, which describes the deformation characteristics of the structure under force. By generating an adjacency matrix from the stiffness matrix, it can be ensured that the graph structure can accurately reflect the mechanical behavior of the structure. The adjacency matrix is a two-dimensional matrix, and the element value is used to indicate whether there is a mechanical correlation between the graph nodes. For example, if there is a mechanical correlation between two nodes, the corresponding adjacency matrix element value is 1; if there is no mechanical correlation between two nodes, the corresponding adjacency matrix element value is 0.

[0081] Finally, the graph nodes, the graph edges, and the adjacency matrix are integrated to obtain initial graph data. Specifically, the initial graph data contains the topological information and physical properties of the structure, providing necessary input for subsequent graph neural network processing. This process not only preserves the physical characteristics of the finite element model, but also provides a structured data form for the learning and prediction of graph neural networks. Through this way of constructing the graph structure, the mechanical behavior of the structure can be effectively represented, laying a solid foundation for efficient and low-data-dependent structural dynamic response prediction.

[0082] In step S30, the mass matrix, the damping matrix, and the stiffness matrix are mapped based on the initial graph data, and the ground motion time history data is obtained.

[0083] It should be noted that this process not only needs to accurately convert physical information into a form that can be processed by a graph neural network, but also needs to ensure that this information is reasonably represented in the graph structure, so that subsequent calculations and predictions can accurately reflect the behavior of the structure under seismic action.

[0084] Further, step S30 further includes: first, extracting a graph edge list from the initial graph data to determine the corresponding structure unit of each graph edge. The graph edge list is a collection of all edges in the graph structure, and each graph edge represents the connection relationship between two graph nodes, corresponding to the two nodes connected by the element in the finite element model.

[0085] Next, the sub-matrices corresponding to the structural elements in the mass matrix, the sub-matrices corresponding to the structural elements in the damping matrix, and the sub-matrices corresponding to the structural elements in the stiffness matrix are mapped to the first edge feature component, the second edge feature component, and the third edge feature component of each edge in the graph. The mass matrix describes the mass distribution of the structure, the damping matrix describes the energy dissipation characteristics of the structure, and the stiffness matrix describes the elastic properties of the structure. The sub-matrices in these matrices are associated with specific structural elements, reflecting the physical properties of the elements. By mapping these sub-matrices to the feature components of the graph edges, the physical properties of the structural elements can be converted into a form that can be processed by the graph neural network.

[0086] Next, the first edge feature component, the second edge feature component, and the third edge feature component are combined to obtain the graph edge features. The graph edge features are the attribute vectors of the graph edges, containing all the physical information related to the edges. This combination allows the graph neural network to consider multiple physical properties such as mass, damping, and stiffness when processing the graph edges, thereby more accurately simulating the behavior of the structural elements.

[0087] Subsequently, a list of graph nodes is extracted from the initial graph data, and the corresponding structural degrees of freedom of each graph node are determined. The list of graph nodes is a collection of all nodes in the graph structure, and each graph node represents a structural degree of freedom. By extracting the list of graph nodes, the representation of each structural degree of freedom in the graph structure can be determined, providing a basis for subsequent feature mapping.

[0088] Then, the load time series corresponding to the structural degrees of freedom in the seismic time history data are mapped to the node feature components of each graph node. The seismic time history data describe the change of seismic action over time, and the load time series is the specific performance of seismic motion on a particular degree of freedom. By mapping the load time series to the node feature components, the external excitation information of the seismic motion can be converted into a form that can be processed by the graph neural network, so that the graph neural network can receive the corresponding seismic motion information at each time step.

[0089] Finally, the initial displacement data of the target building structure at the start time of the prediction period is obtained and added to the node feature component corresponding to the graph node to obtain the graph node feature. Then, the graph edge feature, the graph node feature, and the adjacency matrix in the initial graph data are integrated to obtain the graph input data. Specifically, the graph node feature is the attribute vector of the graph node, which contains all the physical information and initial state information related to the node. By adding the initial displacement data to the node feature component, the graph neural network can consider both the external excitation of ground motion and the initial state of the structure when processing the graph node, thereby more accurately simulating the behavior of the structure's degrees of freedom. In addition, the graph input data is the input of the graph neural network, which contains the topological information, physical properties, initial state, and external excitation information of the structure. By integrating these information, the graph neural network can provide a comprehensive perspective and make dynamic response prediction considering all relevant factors.

[0090] Step S40, inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure.

[0091] It should be noted that the preset seismic response model includes a graph encoder, a time integration calculation module, and a graph decoder. The graph encoder includes a residual network module and a convolution residual network module. The residual network module includes a first preset number of residual blocks, and the convolution residual network module includes a first preset number of convolution residual blocks. The time integration module includes a parameter calculation unit, an edge update unit, and a node recursion unit. The graph decoder includes a first preset number of fully connected layers and an output layer.

[0092] Specifically, the graph encoder is the first part of the model, and its task is to extract key features from the graph input data and encode these features into a more easily processed form. The graph encoder consists of two main modules: a residual network module and a convolution residual network module. The residual network module is composed of a first preset number of residual blocks. The design of these residual blocks allows the network to learn deeper feature representations during training while avoiding the problem of gradient vanishing, thereby improving the training efficiency and stability of the model. The convolution residual network module further enhances the model's ability to process graph structure data. It contains a first preset number of convolution residual blocks, which can effectively capture the spatial relationships between nodes and edges in the graph through convolution operations, further enriching the feature representation.

[0093] Then, the feature data obtained after the graph encoder processing is sent to the time integration calculation module. This module is the core of the entire model, which is responsible for calculating and updating the state of each node in the graph according to the time sequence, thereby simulating the dynamic response process of the structure under the action of the earthquake. The time integration calculation module is composed of a parameter calculation unit, an edge update unit and a node recursion unit. The parameter calculation unit is responsible for calculating various parameters required in the time integration process, which are crucial for accurately simulating the physical process. The edge update unit updates the edge information according to the characteristics of the graph edge and the state of the adjacent nodes, ensuring that the interaction between the nodes in the graph structure can be correctly reflected. The node recursion unit recursively calculates the node state at the next time step based on the updated edge information and the current node state, and in this way, the model can gradually build the node state change in the entire time sequence, i.e. the dynamic response process of the structure.

[0094] Finally, the node state data obtained by the time integration calculation module is passed to the graph decoder. The task of the graph decoder is to decode these encoded feature data into interpretable seismic response displacement sequences. The graph decoder is composed of a first preset number of fully connected layers and an output layer. The fully connected layer is responsible for further conversion and integration of the feature data to extract key information related to the seismic response displacement sequence. The output layer finally converts this information into a specific displacement sequence, i.e. the response result of the target building structure under the action of the earthquake.

[0095] By inputting the graph input data into such a sophisticated and functional preset seismic response model, we can obtain the seismic response displacement sequence of the target building structure. This process not only fully utilizes the advantages of graph neural networks in handling complex structural data, but also accurately simulates the dynamic response process of the structure through the time integration calculation module, and finally obtains a prediction result with high precision and reliability through the graph decoder.

[0096] Step S50, taking the seismic response displacement sequence as the seismic dynamic response result of the target building structure.

[0097] It should be noted that the seismic response displacement sequence obtained after the preset seismic response model processing is a detailed dynamic response description of the target building structure to the action of the earthquake in the prediction period. This displacement sequence not only contains the displacement change of the structure under the action of the earthquake, but also implicitly contains the dynamic information such as velocity and acceleration, because displacement is obtained from acceleration by time integration, and velocity is the time derivative of displacement. These information is crucial for evaluating the seismic performance of the structure.

[0098] Furthermore, using the seismic response displacement sequence as the seismic dynamic response result of the target building structure means that this sequence can be directly applied to various aspects such as seismic design, performance evaluation, and disaster prevention. For example, in seismic design, engineers can use this sequence to evaluate the structure's response under different seismic intensities, thereby determining whether the structure meets the requirements of seismic codes. In performance evaluation, this sequence can help identify weak points in the structure, providing a basis for structural reinforcement and modification. In disaster prevention, by analyzing this sequence, the potential damage to the structure during an earthquake can be predicted, allowing for the implementation of appropriate protective measures in advance.

[0099] Furthermore, this seismic dynamic response result can be used to validate and improve the finite element model. By comparing the predicted displacement sequence with actual observation data or experimental results, the accuracy and reliability of the finite element model can be evaluated. If there is a significant deviation between the predicted results and the actual data, the finite element model can be adjusted and optimized accordingly to improve the model's prediction accuracy.

[0100] Furthermore, the model proposed in this embodiment is verified, such as... Figure 2 The 21-story finite element model frame structure diagram shown is illustrated. The frame structure is 98m high and has a planar dimension of 48.6m × 48.6m. The structural finite element model uses full beam elements for simulation. The finite element model has 2984 nodes, 7656 elements, 17400 degrees of freedom, and a stiffness matrix dimension of 17400 × 17400. The main parameters of the finite element model are shown in Table 1, the finite element model characteristic table.

[0101] Table 1. Properties of the Finite Element Model

[0102]

[0103] The data from the first 15 seconds of a specific seismic load were applied to the aforementioned finite element model, such as... Figure 3 The training earthquake load time history curve is shown in the figure, and the training dataset is obtained.

[0104] In the preset seismic response model of this embodiment, the relevant parameters of the model are determined based on the degrees of freedom of the finite element model described above. Specific parameter settings are shown in Table 2, the preset seismic response model parameter table:

[0105] Table 2 Preset Seismic Response Model Parameter Table

[0106]

[0107] Using the same seismic load data from the first 15 seconds as described above, applied to the aforementioned preset seismic response model, such as... Figure 4 The predicted earthquake load time history curve is shown in the figure, and the prediction results are obtained.

[0108] By comparing the calculation method of the preset seismic response model of the present embodiment with the calculation method of the finite element model, the prediction result index table of Table 3 and the calculation efficiency result table of Table 4 are obtained as shown:

[0109] Table 3 Prediction result index table

[0110]

[0111] Table 4 Calculation efficiency result table

[0112]

[0113] The average and contrast values of the normalized mean square error (NMSE) and the determination coefficient (R²) are listed in Table 3, where the average of NMSE is 0.00129, the contrast is 0.00116, the average of R² is 0.986, and the contrast is 0.012, which shows that the patent method has high prediction accuracy and fitting degree. Table 4 compares the performance of the traditional finite element method and the patent method in terms of calculation time, time step number and efficiency improvement multiple. The calculation time of the patent method is only 8.2 seconds, which is 169 times faster than the 1385.7 seconds of the finite element method, and it handles 5736 time steps, significantly improving the calculation efficiency. These data show that the patent method not only has high accuracy in structure dynamic response prediction, but also has fast calculation speed, with significant improvement.

[0114] In this embodiment, the finite element model of the target building structure and the seismic time history data are obtained, the graph structure is constructed and the initial graph data is obtained. Then the initial graph data is mapped to the graph input data, which is input into the preset seismic response model for processing, and finally the seismic response displacement sequence is obtained as the dynamic response result. The preset model includes a graph encoder, a time integration calculation module and a graph decoder, which work cooperatively to realize efficient and accurate seismic response prediction, effectively reduce the consumption of computing resources and calculation time, while maintaining high precision and generalization, meeting the real-time simulation demand.

[0115] Based on the first embodiment of the present application, the same or similar contents as the above embodiment one can refer to the above introduction, and will not be repeated hereinafter. On this basis, please refer to Figure 5 , the building structure seismic dynamic response calculation method based on time integration graph network further includes steps S201-S203:

[0116] Step S201, dimension reduction processing of the graph input data is performed by the graph encoder to obtain a low-dimensional hidden graph.

[0117] It should be noted that the graph encoder compresses the high-dimensional features contained in the graph input data through the internal residual network module and the convolution residual network module, extracts key information, and maps it to a low-dimensional space to obtain a low-dimensional hidden graph. This dimension reduction process not only reduces the storage requirements and computational complexity of data, but also improves the model's ability to capture important features, which helps subsequent time integration calculations and seismic response prediction.

[0118] Further, step S201 includes: first, inputting the graph node features in the graph input data into the residual network module, performing feature compression and dimension reduction on the graph node features through the residual blocks to obtain low-dimensional node features, wherein the dimension of the low-dimensional node features is a preset proportion of the dimension of the graph node features. Specifically, the graph node features are input into the residual network module. The residual network module is composed of multiple residual blocks. Each residual block introduces a skip connection, enabling the network to learn the residual mapping between input features and output features. This design helps to alleviate the gradient vanishing problem in deep network training, while promoting the network to learn deeper feature representations. In the residual block, a series of linear transformations and nonlinear activation functions are used to compress the features and reduce the dimension of the graph node features, thereby obtaining low-dimensional node features. This process not only reduces the dimension of the data, but also enhances the expressiveness of the features, enabling the low-dimensional node features to effectively represent the original high-dimensional features with a preset proportion of the dimension.

[0119] Secondly, the graph edge features in the graph input data are input into the convolution residual network module to perform convolution operation and dimension reduction on the graph edge features to obtain low-dimensional edge features, wherein the dimension of the low-dimensional edge features is a preset proportion of the dimension of the graph edge features. Specifically, the processing of graph edge features is similar to that of graph node features, but it is processed using the convolution residual network module. The convolution residual network module uses convolution operation to capture the spatial dependency in the graph structure, which is crucial for understanding the interaction between nodes. By performing convolution operation and dimension reduction on the graph edge features, low-dimensional edge features are obtained. These low-dimensional edge features also retain the key information of the original graph edge features with a preset proportion of the dimension, providing a more compact data representation for subsequent time integration calculations.

[0120] Then, the adjacency matrix in the graph input data is sparsified to obtain a processed adjacency matrix. Specifically, after processing the graph node features and the graph edge features, the adjacency matrix in the graph input data is sparsified. The adjacency matrix describes the connection between nodes in the graph, and its sparsification can further reduce the storage requirements and computational complexity of data. Sparse processing usually involves identifying and removing connections that contribute less to model prediction, resulting in a more streamlined adjacency matrix that still accurately represents the topology of the graph.

[0121] Finally, the low-dimensional node features, low-dimensional edge features, and the processed adjacency matrix are combined to obtain a low-dimensional latent graph. Specifically, the low-dimensional latent graph is a reduced-dimensional data representation that not only contains key physical information of nodes and edges in the graph, but also preserves the topological structure of the graph. This data representation provides an efficient and compact input for subsequent time integration calculations and seismic response prediction, enabling the model to significantly reduce computational resource consumption while maintaining high prediction accuracy.

[0122] Step S202: performing multi-step recursive calculation on the low-dimensional latent graph by the time integration calculation module to obtain updated latent graph node features at each time step.

[0123] It should be noted that the time integration calculation module is the core part of the seismic response prediction, which is responsible for multi-step recursive calculation on the low-dimensional latent graph to simulate the evolution of structural dynamic response under seismic action. The module works collaboratively through the parameter calculation unit, edge update unit, and node recursion unit to achieve time series update of latent graph node features.

[0124] Further, step S202 further includes: first, extracting low-dimensional edge features and low-dimensional node features from the low-dimensional latent graph. The low-dimensional edge features and low-dimensional node features extracted from the low-dimensional latent graph are the basis for time integration calculation. These features contain key information of structural units and degrees of freedom, which are used to simulate the dynamic response of the structure under seismic action.

[0125] Secondly, the edge update function and the node update function are set. The edge update function adopts a convolutional neural network, which includes a second preset number of convolutional layers and a third preset number of activation layers. The node update function adopts a fully connected neural network, which contains a second preset number of fully connected layers and a first preset number of activation layers. Specifically, the convolutional layers are responsible for extracting local features in edge features, while the activation layers introduce nonlinearity, enabling the network to learn more complex feature representations. In this way, the edge update function can optimize the low-dimensional edge features, obtaining updated edge features that more accurately reflect the interactions between structural units. The node update function is implemented using a fully connected neural network, which contains a second preset number of fully connected layers and a first preset number of activation layers. The fully connected layers are responsible for mapping node features to a new high-dimensional space, while the activation layers further enhance the network's non-linear expression capabilities. The node update function aggregates the low-dimensional node features of adjacent graph nodes to obtain the graph node features at the next time step, a process that simulates the dynamic changes in structural degrees of freedom under seismic action. Then, the low-dimensional edge features are input into the edge update function for feature optimization, obtaining updated edge features. The low-dimensional node features are updated in the first step, obtaining the updated hidden graph node features at the first time step. Based on the updated edge features, the node update function aggregates the low-dimensional node features of adjacent graph nodes in the low-dimensional hidden graph, obtaining the graph nodes at the next time step. Specifically, the low-dimensional node features are updated in the first step, obtaining the updated hidden graph node features at the first time step. This step is the basis for subsequent recursive calculations, providing the model with initial node state information. Next, based on the updated edge features, the node update function aggregates the low-dimensional node features of adjacent graph nodes in the low-dimensional hidden graph, obtaining the graph node features at the next time step. This aggregation process takes into account the interactions between structural units, enabling the model to more accurately predict the dynamic response of the structure. Finally, the updated hidden graph node features at the first time step are used as the initial node features of the graph nodes at the next time step, and the update and aggregation steps are repeated until the calculation of all time steps within the prediction period is completed, obtaining the updated hidden graph node features at each time step. Specifically, the updated hidden graph node features at the first time step are used as the initial node features of the graph nodes at the next time step, and the update and aggregation steps are repeated until the calculation of all time steps within the prediction period is completed. This multi-step recursive calculation process simulates the evolution of structural dynamic response under seismic action, obtaining the updated hidden graph node features at each time step. These features not only contain the dynamic information such as displacement, velocity and acceleration of the structure at each time step, but also reflect the interactions between structural units and energy transfer.

[0126] In this way, the method of the embodiment can process a large number of time steps in a short time, providing a fast and accurate tool for seismic design and performance evaluation of structures. The efficiency of this method is reflected in its ability to significantly reduce the consumption of computing resources while maintaining high prediction accuracy. In addition, this method can adapt to different seismic scenarios and structure types, and has good generalization ability.

[0127] Step S203, the updated hidden graph node features are decoded by the graph decoder to obtain the seismic response displacement sequence of the target building structure.

[0128] It should be noted that the role of the graph decoder is to convert the updated hidden graph node features of the time integration calculation module back to the actual physical quantity, i.e., the seismic response displacement sequence of the target building structure. This decoding process is a key step for mapping low-dimensional hidden graph node features back to high-dimensional space, which involves reconstructing the spatial information and physical details lost in the encoding process. The graph decoder is usually composed of multiple fully connected layers, which can learn the complex mapping relationship from low-dimensional features to high-dimensional output, and introduce nonlinearity through activation functions to enhance the expression ability of the model.

[0129] Further, step S203 includes: first, input the updated hidden graph node features of each time step in time sequence to the fully connected layer of the graph decoder for dimension recovery and feature mapping, to obtain the physical domain features corresponding to each time step. Specifically, the fully connected layer performs dimension recovery and feature mapping on the input low-dimensional features through a series of linear combinations of weights and biases, and subsequent nonlinear activation functions. This process not only recovers the high-dimensional characteristics of the original data, but also ensures that the mapped features are reasonable in physical sense. Processing the updated hidden graph node features of each time step in time sequence can ensure that the decoding process matches the time evolution of seismic action, thereby obtaining the physical domain features corresponding to each time step.

[0130] Secondly, displacement information is extracted from the physical domain features corresponding to each time step, including the x-direction displacement, y-direction displacement and z-direction displacement of the corresponding structural degrees of freedom of the target building structure at the current time step. Specifically, extracting displacement information from physical domain features is the process of converting decoded data into specific physical quantities. Displacement information includes the x, y and z direction displacements of the target building structure at each time step, which are important parameters for evaluating structural response and designing structures.

[0131] Then, the displacement information at each time step is verified for validity, obtaining a verification result. Specifically, this step usually involves checking the reasonableness of the displacement values, for example, by comparing the displacement values with a preset displacement threshold, abnormal values that may be caused by calculation errors or model inaccuracies can be identified. Validity verification helps to improve the reliability of the prediction result, ensuring that only reasonable displacement information is used to construct the final seismic response displacement sequence of the target building structure.

[0132] Finally, when the numerical range of the displacement information is within the preset displacement threshold, the displacement information of all valid time steps is integrated in chronological order to obtain the seismic response displacement sequence of the target building structure. Specifically, when the numerical range of the displacement information is within the preset displacement threshold, the information is considered valid and is integrated into the seismic response displacement sequence. This integration process is performed in chronological order to ensure the time continuity and consistency of the displacement sequence. The final seismic response displacement sequence obtained is a complete dynamic response description of the target building structure under the action of earthquakes, which can be used for seismic performance evaluation, design optimization, and disaster prevention of the structure.

[0133] The embodiment generates a low-dimensional hidden graph by dimension reduction processing of the initial graph data through the graph encoder, updates the node features of the hidden graph by multi-step recursion using the time integral calculation module, and restores the seismic response displacement sequence of the target structure using the graph decoder, which improves the calculation efficiency while ensuring high precision and good generalization ability, greatly reducing the dependence on large-scale training data.

[0134] Based on the first embodiment of the present application, the present application also provides a building structure seismic dynamic response calculation device based on a time integral graph network, please refer to Figure 6 , the device comprises:

[0135] The acquisition module 10 is configured to acquire a finite element model of a target building structure and seismic time history data within a prediction period, wherein the finite element model comprises node information, element information, a mass matrix, a damping matrix, and a stiffness matrix of the target building structure.

[0136] The construction module 20 is configured to construct a graph structure according to the node information and element information of the finite element model, and obtain initial graph data.

[0137] The mapping module 30 is configured to map the mass matrix, the damping matrix, the stiffness matrix, and the seismic time history data based on the initial graph data to obtain graph input data.

[0138] The processing module 40 is configured to input the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integration module comprises a parameter calculation unit, an edge update unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer.

[0139] The result module 50 is configured to take the seismic response displacement sequence as a seismic dynamic response result of the target building structure.

[0140] The building structure seismic dynamic response calculation device based on the time integration graph network provided in the application adopts the building structure seismic dynamic response calculation method based on the time integration graph network in the above embodiment, and can solve the technical problem of how to realize high-precision and real-time dynamic response calculation of the structure while requiring less training data. Compared with the prior art, the building structure seismic dynamic response calculation device based on the time integration graph network provided in the application has the same beneficial effects as the building structure seismic dynamic response calculation method based on the time integration graph network provided in the above embodiment, and other technical features of the building structure seismic dynamic response calculation device based on the time integration graph network are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0141] In an embodiment, the processing module 40 is further configured to perform dimension reduction processing on the graph input data through the graph encoder to obtain a low-dimensional hidden graph, perform multi-step recursion calculation on the low-dimensional hidden graph through the time integration calculation module to obtain updated hidden graph node features at each time step, and restore the updated hidden graph node features through the graph decoder to obtain the seismic response displacement sequence of the target building structure.

[0142] In an embodiment, the processing module 40 is further configured to input graph node features in the graph input data into a residual network module, perform feature compression and dimension reduction on the graph node features through the residual block to obtain low-dimensional node features, wherein the dimension of the low-dimensional node features is a preset proportion of the dimension of the graph node features; input graph edge features in the graph input data into a convolution residual network module to perform convolution operation and dimension reduction on the graph edge features to obtain low-dimensional edge features, wherein the dimension of the low-dimensional edge features is a preset proportion of the dimension of the graph edge features; perform sparse processing on an adjacency matrix in the graph input data to obtain a processed adjacency matrix; and combine the low-dimensional node features, the low-dimensional edge features and the processed adjacency matrix to obtain a low-dimensional hidden graph.

[0143] In an embodiment, the processing module 40 is further configured to extract low-dimensional edge features and low-dimensional node features from the low-dimensional latent graph; set an edge update function and a node update function, the edge update function adopts a convolutional neural network, the convolutional neural network includes a second preset number of convolutional layers and a third preset number of activation layers, the node update function adopts a fully connected neural network, the fully connected neural network includes a second preset number of fully connected layers and a first preset number of activation layers; input the low-dimensional edge features into the edge update function to optimize the low-dimensional edge features, and obtain updated edge features; perform first-step recursive update on the low-dimensional node features to obtain updated latent graph node features at a first time step; aggregate low-dimensional node features of adjacent graph nodes in the low-dimensional latent graph based on the updated edge features through the node update function, and obtain graph nodes at a next time step; take the updated latent graph node features at the first time step as initial node features of the graph nodes at the next time step, repeatedly perform the updating and aggregating steps until the calculation of all time steps in the prediction period is completed, and obtain updated latent graph node features at each time step.

[0144] In an embodiment, the processing module 40 is further configured to input the updated latent graph node features at each time step in time sequence into the fully connected layers of the graph decoder for dimension recovery and feature mapping, to obtain physical domain features corresponding to each time step; extract displacement information from the physical domain features corresponding to each time step, the displacement information including x-direction displacement, y-direction displacement and z-direction displacement of the structure degrees of freedom of the target building structure at the current time step; perform validity verification on the displacement information at each time step, to obtain a verification result; when the numerical range of the displacement information is within a preset displacement threshold, integrate the displacement information of all valid time steps in time sequence, to obtain a seismic response displacement sequence of the target building structure.

[0145] In an embodiment, the construction module 20 is further configured to extract structural degrees of freedom from the node information of the finite element model, determine physical coordinates and mechanical properties corresponding to each of the structural degrees of freedom, map each of the structural degrees of freedom to a graph node, and associate the physical coordinates and the mechanical properties of the corresponding structural degrees of freedom with the graph node; extract structural element connection relationships from the element information of the finite element model, and determine the range of structural degrees of freedom connected by each of the structural elements; set a graph edge between the corresponding graph nodes according to the range of structural degrees of freedom connected by the structural elements, and associate the type and size of the corresponding structural elements with the graph edge; generate an adjacency matrix based on the stiffness matrix of the finite element model, wherein the element values in the adjacency matrix are first preset values indicating that the corresponding graph nodes are mechanically associated, and the element values are second preset values indicating that there is no mechanical association; and integrate the graph nodes, the graph edges, and the adjacency matrix to obtain initial graph data.

[0146] In an embodiment, the mapping module 30 is further configured to extract a graph edge list from the initial graph data, determine the structural elements corresponding to each of the graph edges, map the sub-matrix of the mass matrix corresponding to the structural elements, the sub-matrix of the damping matrix corresponding to the structural elements, and the sub-matrix of the stiffness matrix corresponding to the structural elements to the first edge feature component, the second edge feature component, and the third edge feature component of each of the graph edges, combine the first edge feature component, the second edge feature component, and the third edge feature component to obtain a graph edge feature, extract a graph node list from the initial graph data, determine the structural degrees of freedom corresponding to each of the graph nodes, map the load time sequence corresponding to the structural degrees of freedom in the ground motion time history data to the node feature component of each of the graph nodes, obtain initial displacement data of the target building structure at the start time of the prediction period, add the initial displacement data to the node feature component corresponding to the graph nodes to obtain a graph node feature, and integrate the graph edge feature, the graph node feature, and the adjacency matrix in the initial graph data to obtain graph input data.

[0147] The present application provides a building structure seismic dynamic response calculation device based on a time integral graph network, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the building structure seismic dynamic response calculation method based on the time integral graph network in the above-mentioned embodiment one.

[0148] Reference will be made to the following description Figure 7This document illustrates a structural schematic diagram of a building structure seismic dynamic response calculation device based on a time integral graph network, suitable for implementing embodiments of this application. The building structure seismic dynamic response calculation device based on a time integral graph network in this application embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 7 The illustrated seismic dynamic response calculation device for building structures based on time integral graph networks is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0149] like Figure 7 As shown, the seismic dynamic response calculation device for building structures based on time integral graph networks may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in read-only memory (ROM) 1002 or programs loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the seismic dynamic response calculation device for building structures based on time integral graph networks. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the time integral graph network-based building structure seismic dynamic response calculation device to wirelessly or wiredly communicate with other devices to exchange data. Although various time integral graph network-based building structure seismic dynamic response calculation devices are shown in the figures, it should be understood that implementation or possession of all of them is not required. More or fewer may be implemented alternatively.

[0150] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable storage medium, the computer program containing program codes for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiments disclosed in the present application are executed.

[0151] The building structure seismic dynamic response calculation device based on the time integral graph network provided by the present application adopts the building structure seismic dynamic response calculation method based on the time integral graph network in the above-mentioned embodiments, and can solve the technical problem of how to realize high-precision and real-time dynamic response calculation of the structure while having low requirements for training data. Compared with the prior art, the building structure seismic dynamic response calculation device based on the time integral graph network provided by the present application has the same beneficial effects as the building structure seismic dynamic response calculation method based on the time integral graph network provided by the above-mentioned embodiments, and other technical features in the building structure seismic dynamic response calculation device based on the time integral graph network are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0152] It should be understood that parts of the present application can be realized by hardware, software, firmware or a combination thereof. In the description of the above-mentioned embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0153] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0154] The present application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e. computer program) for executing the building structure seismic dynamic response calculation method based on the time integral graph network in the above-mentioned embodiments.

[0155] The computer readable storage medium provided in the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor, or device, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read only memory (ROM), an erasable programmable read only memory (EPROM or flash memory), an optical fiber, a portable compact disk read only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present embodiment, the computer readable storage medium can be any tangible storage medium containing or storing a program that can be used by or in conjunction with an instruction execution device. The program code contained on the computer readable storage medium can be transmitted by any suitable storage medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), etc., or any suitable combination of the above.

[0156] The above computer readable storage medium can be contained in a building structure seismic dynamic response calculation device based on a time integral graph network, or can exist separately without being assembled into the building structure seismic dynamic response calculation device based on the time integral graph network.

[0157] The computer-readable storage medium described above can bear one or more programs, when the one or more programs are executed by the time-integral graph network-based building structure seismic dynamic response calculation device, the time-integral graph network-based building structure seismic dynamic response calculation device can be written in one or more program design languages or combinations thereof for executing computer program codes of the operations of the present application, the program design languages include object-oriented program design languages such as Java, Smalltalk, C++, and also include conventional procedural program design languages such as the 'C' language or similar program design languages. The program code can be executed completely on a user computer, partially on a user computer, as an independent software package, partially on a user computer and partially on a remote computer, or completely on a remote computer or server. In the case involving a remote computer, the remote computer can be connected to the user computer through any kind of network including a local area network (LAN: Local Area Network) or a wide area network (WAN: Wide Area Network), or can be connected to an external computer (for example, through the Internet by using an Internet service provider).

[0158] The flowcharts and block diagrams in the drawings illustrate the possible implementation architecture, function and operation of the methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment, or a part of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders than those noted in the drawings. For example, two blocks represented in succession can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system for executing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0159] The modules involved in the embodiments of the present application can be implemented in a software manner or in a hardware manner. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0160] The readable storage medium provided by the application is a computer readable storage medium, and the computer readable storage medium stores computer readable program instructions (i.e., a computer program) for executing the building structure seismic dynamic response calculation method based on the time integral graph network, and can solve the technical problem of how to realize high-precision and real-time dynamic response calculation of the structure while having low requirements on training data. Compared with the prior art, the computer readable storage medium provided by the application has the same beneficial effects as the building structure seismic dynamic response calculation method based on the time integral graph network provided by the above-mentioned embodiments, and will not be repeated here.

[0161] The application further provides a computer program product comprising a computer program, which, when executed by a processor, implements the steps of the building structure seismic dynamic response calculation method based on the time integral graph network as described above.

[0162] The computer program product provided by the application can solve the technical problem of how to realize high-precision and real-time dynamic response calculation of the structure while having low requirements on training data. Compared with the prior art, the computer program product provided by the application has the same beneficial effects as the building structure seismic dynamic response calculation method based on the time integral graph network provided by the above-mentioned embodiments, and will not be repeated here.

[0163] The above-mentioned is only part of the embodiments of the application, and does not limit the patent scope of the application, and any equivalent structural transformation, direct / indirect application in other related technical fields made by using the content of the application specification and drawings under the technical concept of the application are included in the patent protection scope of the application.

Claims

1. A method for calculating seismic dynamic response of a building structure based on a time-integration graph network, characterized in that, The method comprises the following steps: obtaining a finite element model of a target building structure and time history data of ground motion in a prediction period, wherein the finite element model comprises node information, element information, a mass matrix, a damping matrix and a stiffness matrix of the target building structure; constructing a graph structure according to the node information and the element information of the finite element model, and obtaining initial graph data; mapping the mass matrix, the damping matrix and the stiffness matrix and the time history data of ground motion based on the initial graph data to obtain graph input data; inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integration calculation module comprises a parameter calculation unit, an edge update unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer; taking the seismic response displacement sequence as a seismic dynamic response result of the target building structure; the step of mapping the mass matrix, the damping matrix and the stiffness matrix and the time history data of ground motion based on the initial graph data to obtain graph input data comprises: extracting a graph edge list from the initial graph data to determine a structure element corresponding to each graph edge; mapping a sub-matrix corresponding to the structure element in the mass matrix, a sub-matrix corresponding to the structure element in the damping matrix and a sub-matrix corresponding to the structure element in the stiffness matrix into a first edge feature component, a second edge feature component and a third edge feature component of each graph edge; combining the first edge feature component, the second edge feature component and the third edge feature component to obtain a graph edge feature; extracting a graph node list from the initial graph data to determine a structure degree of freedom corresponding to each graph node; mapping a load time sequence corresponding to the structure degree of freedom in the time history data of ground motion into a node feature component of each graph node; obtaining initial displacement data of the target building structure at the start time of the prediction period; adding the initial displacement data to the node feature component corresponding to the graph node to obtain graph node features; integrating the graph edge features, the graph node features and an adjacency matrix in the initial graph data to obtain graph input data.

2. The method of claim 1, wherein, the step of inputting the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, comprises: performing dimension reduction processing on the graph input data through the graph encoder to obtain a low-dimensional hidden graph; performing multi-step recursion calculation on the low-dimensional hidden graph through the time integration calculation module to obtain updated hidden graph node features at each time step; The graph decoder is used to decode and restore the updated hidden graph node features to obtain a seismic response displacement sequence of the target building structure.

3. The method of claim 2, wherein, The step of dimension reduction processing of the graph input data by the graph encoder to obtain a low-dimensional hidden graph comprises: inputting graph node features in the graph input data into a residual network module, performing feature compression and dimension reduction on the graph node features by the residual block to obtain low-dimensional node features, wherein the dimension of the low-dimensional node features is a preset proportion of the dimension of the graph node features; inputting graph edge features in the graph input data into a convolution residual network module to perform convolution operation and dimension reduction on the graph edge features to obtain low-dimensional edge features, wherein the dimension of the low-dimensional edge features is a preset proportion of the dimension of the graph edge features; performing sparse processing on the adjacency matrix in the graph input data to obtain a processed adjacency matrix; combining the low-dimensional node features, the low-dimensional edge features and the processed adjacency matrix to obtain a low-dimensional hidden graph.

4. The method of claim 2, wherein, The step of multi-step recursive calculation of the low-dimensional hidden graph by the time integration calculation module to obtain updated hidden graph node features at each time step comprises: extracting low-dimensional edge features and low-dimensional node features from the low-dimensional hidden graph; setting an edge update function and a node update function, the edge update function adopts a convolutional neural network comprising a second preset number of convolutional layers and a third preset number of activation layers, and the node update function adopts a fully connected neural network comprising a second preset number of fully connected layers and a first preset number of activation layers; inputting the low-dimensional edge features into the edge update function to optimize the low-dimensional edge features to obtain updated edge features; performing first-step recursive update on the low-dimensional node features to obtain updated hidden graph node features at a first time step; based on the updated edge features, aggregating low-dimensional node features of adjacent graph nodes in the low-dimensional hidden graph by the node update function to obtain graph nodes at a next time step; taking the updated hidden graph node features at the first time step as initial node features of the graph nodes at the next time step, repeating the updating and aggregating steps until the calculation of all time steps in the prediction period is completed to obtain updated hidden graph node features at each time step.

5. The method of claim 2, wherein, The step of decoding and restoring the updated hidden graph node features by the graph decoder to obtain a seismic response displacement sequence of the target building structure comprises: inputting the updated hidden graph node features at each time step into the fully connected layers of the graph decoder in time sequence to restore the dimension and map the features, to obtain corresponding physical domain features at each time step; extracting displacement information from the physical domain features corresponding to each time step, the displacement information comprising x-direction displacement, y-direction displacement and z-direction displacement of the target building structure corresponding to the structure freedom degree at the current time step; performing validity verification on the displacement information at each time step to obtain a verification result; When the value range of the displacement information is within a preset displacement threshold, the displacement information of all effective time steps is integrated in time sequence to obtain a seismic response displacement sequence of the target building structure.

6. The method of claim 1, wherein, The step of constructing a graph structure according to the node information and the element information of the finite element model and obtaining initial graph data comprises: extracting structural degrees of freedom from the node information of the finite element model, determining the physical coordinates and mechanical properties corresponding to each structural degree of freedom; mapping each structural degree of freedom to a graph node, and associating the physical coordinates and mechanical properties of the corresponding structural degree of freedom with the graph node; extracting structural element connection relationships from the element information of the finite element model, and determining the structural degree of freedom range connected by each structural element; setting a graph edge between the corresponding graph nodes according to the structural degree of freedom range connected by the structural element, and associating the type and size of the corresponding structural element with the graph edge; generating an adjacency matrix based on the stiffness matrix of the finite element model, wherein the element value in the adjacency matrix is a first preset value indicating that the corresponding graph nodes are mechanically associated, and the element value is a second preset value indicating that there is no mechanical association; integrating the graph node, the graph edge and the adjacency matrix to obtain initial graph data.

7. A time integration graph network-based building structure seismic force response calculation apparatus characterized by comprising: The device comprises: an acquisition module configured to acquire a finite element model of a target building structure and ground motion time history data in a prediction period, wherein the finite element model comprises node information, element information, a mass matrix, a damping matrix and a stiffness matrix of the target building structure; a construction module configured to construct a graph structure according to the node information and the element information of the finite element model, and obtain initial graph data; a mapping module configured to map the mass matrix, the damping matrix and the stiffness matrix and the ground motion time history data based on the initial graph data to obtain graph input data; further configured to extract a graph edge list from the initial graph data, determine the structural element corresponding to each graph edge; map the sub-matrix corresponding to the structural element in the mass matrix, the sub-matrix corresponding to the structural element in the damping matrix and the sub-matrix corresponding to the structural element in the stiffness matrix to the first edge feature component, the second edge feature component and the third edge feature component of each graph edge; combine the first edge feature component, the second edge feature component and the third edge feature component to obtain graph edge features; extract a graph node list from the initial graph data, determine the structural degree of freedom corresponding to each graph node; map the load time sequence corresponding to the structural degree of freedom in the ground motion time history data to the node feature component of each graph node; acquire initial displacement data of the target building structure at the start time of the prediction period; add the initial displacement data to the node feature component corresponding to the graph node to obtain graph node features; integrate the graph edge features, the graph node features and the adjacency matrix in the initial graph data to obtain graph input data; The processing module is configured to input the graph input data into a preset seismic response model for processing to obtain a seismic response displacement sequence of the target building structure, wherein the preset seismic response model comprises a graph encoder, a time integration calculation module and a graph decoder, the graph encoder comprises a residual network module and a convolution residual network module, the residual network module comprises a first preset number of residual blocks, the convolution residual network module comprises a first preset number of convolution residual blocks, the time integration calculation module comprises a parameter calculation unit, an edge update unit and a node recursion unit, and the graph decoder comprises a first preset number of fully connected layers and an output layer. The result module is configured to take the seismic response displacement sequence as a seismic dynamic response result of the target building structure.

8. A time integration graph network-based building structure seismic dynamic response calculation device, characterized by, The device comprises a memory, a processor and a building structure seismic dynamic response calculation program based on a time integration graph network stored on the memory and running on the processor, and the building structure seismic dynamic response calculation program based on the time integration graph network is configured to implement the steps of the building structure seismic dynamic response calculation method based on the time integration graph network.

9. A storage medium, characterized by The storage medium stores a building structure seismic dynamic response calculation program based on a time integration graph network, and the building structure seismic dynamic response calculation program based on the time integration graph network implements the steps of the building structure seismic dynamic response calculation method based on the time integration graph network when executed by the processor.

Images