Structural response rapid prediction method based on physical coding graph network

Abstract

The invention discloses a structure response rapid prediction method based on a physical coding graph network. The method comprises the following steps: constructing the physical coding graph network; inputting the physical parameters of the target structure into a physical coding graph network for coding; message passing is carried out on the coded physical parameters, message passing comprises node mapping and edge mapping which are carried out in sequence, and data flow of message passing conforms to mathematical logic; and decoding results of node mapping and edge mapping to obtain a displacement prediction result and an internal force prediction result. According to the invention, through constructing the physical coding graph network, multi-working-condition structure response rapid prediction is realized; the encoding, message passing and decoding processes are matched with complex structure parameters, accurate prediction of displacement and internal force is completed through node mapping and edge mapping in a labor division mode, the problem of repeated calculation of a traditional method is solved, the digital twinning real-time interaction requirement is met, rapid iterative design and real-time performance evaluation are efficiently supported, and a better solution is provided for structural engineering calculation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering data analysis, and in particular to a structure response rapid prediction method based on a physical encoding graph network. BACKGROUND

[0002] In the whole process of modern engineering construction, numerical simulation is an important means of calculation analysis and design.

[0003] Traditional finite element analysis can only perform single-condition calculation of a single structure model, and repeated operations are required in the face of similar load conditions. In the complex multi-condition scene of structure operation and maintenance, the calculation is complicated and time-consuming, and it is difficult to meet the rapid iterative design and real-time performance evaluation requirements. Further, with the application of digital twin technology in structural engineering, it is required to obtain load data in real time and quickly feedback the structure response. The traditional calculation method cannot meet this timeliness requirement.

[0004] The existing structure response prediction technology has many problems, which are embodied in: (1) Data-driven models rely on a large amount of high-quality data, which is difficult to meet in the engineering field; (2) Physical-driven models (such as physical information neural network PINN) are difficult to adapt to complex structures because the mapping network is only a multilayer perceptron, and the training is difficult due to the physical constraint integrated into the loss function, which causes convergence and stability problems; (3) Existing researches focus on the forward and inverse problems of a single condition, and lack of research on condition generalization. The prediction accuracy and range cannot meet the actual structural calculation requirements.

[0005] The Chinese invention patent with the authorized announcement number CN116432274B discloses a rod system structure optimization method and device, electronic equipment and storage medium, but the invention only targets a single rod system structure, and has low transferability.

[0006] In summary, there is an urgent need for a new efficient and rapid structure response rapid prediction technical solution. SUMMARY

[0007] The main purpose of the present application is to provide a structure response rapid prediction method based on a physical encoding graph network, which aims to solve the technical problems that the existing technology requires repeated operations in the face of similar load conditions in structure analysis, and the calculation is complicated and time-consuming in the complex multi-condition scene of structure operation and maintenance, which is difficult to meet the rapid iterative design and real-time performance evaluation requirements.

[0008] To achieve the above purpose, the present application provides a structure response rapid prediction method based on a physical encoding graph network, comprising the following steps: S1: constructing a physical encoding graph network, which is used to encode, message pass and decode physical parameters of a target structure to obtain structural responses, the structural responses at least including displacement prediction and internal force prediction; S2: inputting the physical parameters of the target structure into the physical encoding graph network for encoding; S3: performing the message passing on the encoded physical parameters, the message passing including sequentially performed node mapping and edge mapping, wherein data flow of the mapping network conforms to mathematical logic, the node mapping is used to perform the displacement prediction, and the edge mapping is used to perform the internal force prediction; S4: decoding results of the node mapping and the edge mapping to obtain displacement prediction results and internal force prediction results.

[0009] Preferably, the construction of the physical encoding graph network includes the following steps: A1: obtaining geometric relationships of the target structure and the physical parameters; wherein the physical parameters include node data and member data of physical properties of the target structure constrained by the geometric relationships; node features are obtained based on the node data and encoded, and edge features are obtained based on the member data and encoded; A2: obtaining a small amount of response data sets of the target structure; training the physical encoding graph network based on A1 and the response data sets, so that the physical encoding graph network encodes the node features and the edge features, performs the message passing based on different input load cases, and decodes and outputs different results of the node mapping and the edge mapping; wherein the response data sets store real responses of the target structure under different load cases, the load cases act on nodes corresponding to the node data, and the node data are taken as input; A3: outputting the trained physical encoding graph network.

[0010] Preferably, the nodes include free nodes and constraint nodes to which loads are applied; The node data of the free nodes are defined as load node data; The node data of the constraint nodes are defined as constraint node data; The node data include the load node data and the constraint node data.

[0011] Preferably, the obtaining node features based on the node data and the encoding include: Based on the load node data, load node features are obtained; The load node features are encoded as: wherein, stiffness of a load node , the stiffness is defined as the load node data, , the stiffness of a load node , the stiffness of a load node , the stiffness of a load node , the stiffness of a load node , the stiffness of a load node , the stiffness of a load node

[0012] As preferred, the obtaining node features and encoding based on the node data further comprises: obtaining constraint node features based on the constraint node data; encoding the constraint node features into: wherein, , the stiffness of a constraint node , and setting the stiffness of the constraint node to be infinite to express the constraint attribute.

[0013] As preferred, the obtaining edge features and encoding based on the member data further comprises: obtaining load node edge features corresponding to the load node data based on the member data; encoding the load node edge features into: wherein, , the edge features of the first adjacent edge and the load node , the stiffness between the load node and its first adjacent node , the edge features of the second adjacent edge and the load node , the stiffness between the load node and its second adjacent node , the edge features of the third adjacent edge and the load node , the stiffness between the load node and its third adjacent node .

[0014] As preferred, the obtaining edge features and encoding based on the member data further comprises: ​​​​​obtaining member data corresponding to the constraint node data, to obtain constraint node edge features; Based on the encoding of the constraint node features and the constraint node edge features, the constraint node edge features are encoded as: wherein, is the edge feature of the first adjacent edge and the constraint node , is the edge feature of the second adjacent edge and the constraint node .

[0015] As a preferred, the displacement prediction comprises: inputting the encoded node features and the edge features into the physical encoding graph network, performing displacement prediction through a node mapping formula, and defining the result of the displacement prediction as the result of the node mapping; wherein the node mapping formula is: wherein, is the result of the node mapping, is the initial feature vector of node , is the set of neighbor nodes of node , is the node feature of the always updated neighbor node , is the edge feature between the node and its neighbor nodes , denotes the aggregation of the node and its neighbor nodes, and is a multi-layer perceptron.

[0016] As a preferred, the internal force prediction comprises: inputting the encoded node features and the edge features into the physical encoding graph network, performing internal force prediction through an edge mapping formula, and defining the result of the internal force prediction as the result of the edge mapping; wherein the edge mapping formula is: wherein, is the result of the edge mapping between node and its neighbor nodes , is the node feature of the always updated node , is the node feature of the always updated neighbor node . is an un-updated node and its neighbor nodes between the nodes, is a multilayer perceptron.

[0017] As preferred, the target structure at least comprises a truss structure; when the target structure is the truss structure, the nodes are the hinged points of the truss structure, and the members are the bars of the truss structure; wherein the structure calculation follows the mathematical logic, and the mathematical logic at least comprises a structural mechanics calculation logic.

[0018] Beneficial effects: the structure response rapid prediction method based on the physical coding graph network of the present application retains the advantages of data flow and training mode of the data-driven model, and the efficient structure attribute coding mode and the physical guided architecture design greatly enhance the physical interpretability and generalization performance of the model, realize maintaining high generalization prediction accuracy through small sample, thereby solving the technical problems that the prior art needs to repeat operation in the face of similar load working conditions in structure analysis, and in the complex multi-working condition scene of structure operation and maintenance, the calculation is complicated and time-consuming, and it is difficult to meet the requirements of rapid iterative design and real-time performance evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0020] Figure 1 the flowchart of the structure response rapid prediction method based on the physical coding graph network provided by the embodiment of the present application; Figure 2 the flowchart of the construction of the physical coding graph network of the structure response rapid prediction method based on the physical coding graph network provided by the embodiment of the present application; Figure 3 the schematic diagram of the node in the structure response rapid prediction method based on the physical coding graph network provided by the embodiment of the present application; Figure 4 the prediction situation based on GCN provided by the embodiment of the present application; Figure 5 the prediction situation based on GIN provided by the embodiment of the present application; Figure 6 the prediction situation based on GAT provided by the embodiment of the present application; Figure 7The prediction situation of the structural response fast prediction method based on the physical coding graph network provided by the embodiment of the present application; Figure 8 The error column contrast chart of the prediction situation based on the GCN, the prediction situation based on the GIN, the prediction situation based on the GAT and the prediction situation of the structural response fast prediction method based on the physical coding graph network provided by the embodiment of the present application; Figure 9 The truss structure and load schematic diagram for displacement prediction and internal force prediction provided by the embodiment of the present application; Figure 10 The real displacement situation based on the truss structure and load for displacement prediction provided by the embodiment of the present application; Figure 11 The predicted displacement situation based on the truss structure and load for displacement prediction provided by the embodiment of the present application; Figure 12 The real internal force situation based on the truss structure and load for internal force prediction provided by the embodiment of the present application; Figure 13 The predicted internal force situation based on the truss structure and load for internal force prediction provided by the embodiment of the present application.

[0021] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0023] Traditional methods such as finite element analysis, although play a key role in solving structural mechanics problems, can only realize single working condition calculation of a structure model, and even if facing similar load working conditions, still need to carry out repeated calculation. In practical application, engineering structures are affected by various factors in the operation and maintenance stage, and the structure needs to bear various complex load working conditions, and each working condition needs to be re-calculated with tedious finite element calculation, resulting in serious repeated calculation problem, time-consuming and laborious, and it is difficult to meet the requirements of rapid iterative design and real-time performance evaluation. Further, with the rise of digital twinning and the application of digital twinning system in structure, structural digital twinning emphasizes real-time interaction between physical entity and virtual model, and puts forward higher requirements for structure calculation, that is, after real-time acquisition of load data, the structure response is quickly fed back. However, the traditional calculation method is difficult to meet this timeliness, so it is necessary to break through the repeated calculation dilemma and establish an efficient and fast structure response prediction method.

[0024] In view of the above technical difficulties, with reference to Figure 1 The embodiment discloses a structural response fast prediction method based on a physical coding graph network, comprising the following steps: S1: a physical encoding graph network is constructed, the physical encoding graph network is used for encoding, message passing and decoding of physical parameters of a target structure to perform structural response, the structural response at least includes displacement prediction and internal force prediction; S2: the physical parameters of the target structure are input into the physical encoding graph network for encoding; S3: message passing is performed on the encoded physical parameters, the message passing includes node mapping and edge mapping performed in sequence, wherein the data flow of the mapping network conforms to mathematical logic, the node mapping is used for displacement prediction, and the edge mapping is used for internal force prediction; S4: the results of the node mapping and the edge mapping are decoded to obtain displacement prediction results and internal force prediction results.

[0025] Through the above, by constructing a physical encoding graph network, rapid prediction of structural response under multiple working conditions is realized; wherein the encoding, message passing and decoding process adapt to complex structural parameters, and the node mapping and edge mapping complete accurate prediction of displacement and internal force, solve the problem of repeated calculation in traditional methods, meet the real-time interaction demand of digital twin, efficiently support rapid iterative design and real-time performance evaluation, and provide a better solution for structural engineering calculation.

[0026] With the innovation of artificial intelligence technology, agent models based on deep learning methods have been successfully applied to prediction tasks of complex engineering systems. Among them, graph neural networks have the ability to express the characteristics of graph structures composed of nodes and edges, and the data form of the graph structure is similar to the structure system. Based on the similar data form and structure system of the graph structure, it is very suitable to use graph neural networks as mapping modules for structural working condition generalization prediction tasks. However, if only simple structural similarity is used for mapping, the data-driven model often needs a large amount of high-quality data as training data set, and the data set in the engineering field, especially for structural calculation, cannot meet the data requirements. Further, in existing physical-driven models, the physical information neural network (PINN) uses a multilayer perceptron as a mapping network, which is difficult to adapt to the complex characteristics of engineering structures, and the physical-driven model makes the model training difficult by using the physical constraints as a loss function, and new convergence and stability problems are generated. It is particularly pointed out that the existing research on physical-driven models mainly focuses on the positive problem and inverse problem of a single working condition of the structure, and represents sample-free solution and parameter inversion. There is a lack of research on working condition generalization of structural calculation or the prediction accuracy and range are difficult to support structural calculation. In order to solve the current limitations of constructing structural generalization calculation to realize rapid prediction of structural response, a new physical encoding graph network is constructed in the embodiment.

[0027] Specifically, referring to Figure 2 The construction of the physical encoding graph network includes the following steps: A1: Obtain the geometric relationship and physical parameters of the target structure; wherein, the physical parameters include node data and member data of physical attributes of the target structure constrained by the geometric relationship; obtain node features and encode based on the node data, and obtain edge features and encode based on the member data; A2: Obtain a small amount of response data set of the target structure; train the physical encoding graph network based on A1 and the response data set, so that the physical encoding graph network encodes the node features and the edge features, performs message passing based on the input of different load cases, and decodes the output of different node mappings and edge mappings; wherein, the response data set stores the real response of the target structure under different load cases, the load case acts on the node corresponding to the node data, and the node data is taken as the input; A3: Output the trained physical encoding graph network.

[0028] In the specific application of the embodiment, the target structure at least includes a truss structure; When the target structure is a truss structure, the node is a hinged point of the truss structure, and the member is a rod of the truss structure.

[0029] The truss structure and the graph structure of the graph neural network have similarities, so the hinged point of the truss structure and the rod can naturally be expressed by the node features and the edge features of the graph neural network respectively.

[0030] Specifically, referring to Figure 3 , the node includes a free node and a constraint node to which a load is applied; The node data of the free node is defined as load node data; The node data of the constraint node is defined as constraint node data; The node data includes the load node data and the constraint node data.

[0031] Generally, the rod system attribute tensile stiffness of the truss structure is expressed by the edge features. The load case applied to the node can use the node features as input, but the constraint condition of the node as the node feature input affects the generalization of the load.

[0032] The problem faced by the static force calculation of the structure in the finite element can be expressed as an algebraic equation of , wherein is the overall stiffness matrix of the structure, and if the displacement vector is solved under the condition that the load vector is known, the overall stiffness matrix needs to be processed according to the boundary constraint condition, then the inverse matrix of is calculated, and finally the value of the vector is obtained. Taking the displacement calculation of the truss structure as an example, each node involves two degrees of freedom of direction and direction, and if the number of nodes is , the number of degrees of freedom is , and the number of equations is . then is a matrix of rows columns. The global processing way of directly inverting the global stiffness matrix is not consistent with the node discrete processing way of graph neural networks, so we establish a new structure feature encoding method combined with numerical iterative theory. The iterative formula of numerical iterative theory is: The iterative formula is used to calculate the algebraic equation , is the stiffness matrix of node , is the stiffness relationship matrix between node and node , is the approximate solution of node in the th iteration, is the approximate solution of a certain node around node in the th iteration.

[0033] In order to simultaneously express the unit stiffness and constraint nodes of the truss structure in the neural network, we take the stiffness influence relationship matrix between nodes as the edge feature to represent the unit stiffness.

[0034] Specifically, referring to Figure 3 , the node features are obtained based on the node data and encoded, including: Based on the load node data, the load node features are obtained; The load node features are encoded as: Wherein, indicates the stiffness of the load node , and the stiffness is defined as the load node data, is the load node number, indicates the set of surrounding neighbor nodes indicates the stiffness influence on node on the surrounding nodes on node .

[0035] Based on the encoding of the load node features, specifically, referring to Figure 3 , the edge features are obtained based on the component data and encoded, including: Obtain the component data corresponding to the load node data to obtain the load node edge features; Based on the encoding of the load node features and the load node edge features, the load node edge features are encoded as: wherein, is the edge feature of the first adjacent edge and the load node , is the stiffness between the load node and its first adjacent node , is the edge feature of the second adjacent edge and the load node , is the stiffness between the load node and its second adjacent node , is the edge feature of the third adjacent edge and the load node , is the stiffness between the load node and its third adjacent node .

[0036] In addition, for the constraint node, the constraint attribute is expressed by directly setting the stiffness of the constraint node to infinity.

[0037] Specifically, referring to Figure 3 , based on the node data, the node features are obtained and encoded, and further comprising: based on the constraint node data, the constraint node features are obtained; the constraint node features are encoded as: wherein, represents the stiffness of the constraint node , and the stiffness of the constraint node is set to infinity to express the constraint attribute.

[0038] Based on the constraint node feature encoding, specifically, referring to Figure 3 , based on the member data, the edge features are obtained and encoded, and further comprising: obtaining the member data corresponding to the constraint node data to obtain the constraint node edge features; based on the encoding of the constraint node features and the constraint node edge features, the constraint node edge features are encoded as: wherein, is the edge feature of the first adjacent edge and the constraint node , is the edge feature of the second adjacent edge and the constraint node .

[0039] Through the above structure coding, all truss structure attributes (especially the constraint nodes) are realized as input in the form of edge features, which are separated from the generalized load node features; and the data normalization is directly realized based on the coding mode of the stiffness influence relationship matrix.

[0040] It should be noted that the reason why the graph neural network is based on the numerical iteration theory is that: (1) the optimization theory and the numerical iteration theory are both based on optimization as the underlying theory, so the similarity is high; (2) the graph neural network is a discrete calculation based on nodes, which is similar to the processing method of sparse matrix, which is corresponding to the iterative calculation; (3) the numerical iteration theory can clearly express the strict mathematical logical relationship between physical variables, and the calculation logical relationship can provide clear design ideas for the data flow of the network model.

[0041] However, in general, whether it is matrix calculation or iterative calculation, it is a calculation method for a single working condition, and the significance of the neural network model is to break through the problem of repeated calculation of working conditions, realize the generalized calculation of structural working conditions, and establish a meta-model of proxy calculation. Therefore, we introduce the graph neural network, which mainly establishes the mapping relationship between node features and edge features, which is called the message passing mechanism, which is represented by the propagation function and the aggregation function. The propagation function corresponds to the message propagation, which updates the features of the adjacent nodes according to the main node. The aggregation function corresponds to the message aggregation, which updates the features of the main node according to the main node and the updated adjacent nodes. The message passing mechanism can be expressed as: wherein, and are fully connected layers of the neural network; and are feature vectors of data points of the layer and the next layer respectively; is an edge contact feature vector, is a corresponding adjacent data point feature vector.

[0042] It can be clearly seen that the structural load working condition generalization calculation method based on the physical coding graph network in the embodiment is based on the mathematical logical relationship of physical quantity updating in the numerical theory, and the network architecture is designed in combination with the message passing mechanism of the graph neural network. Based on the design, the physical coding graph network can perform node prediction tasks and edge prediction tasks for displacement prediction and internal force prediction of the truss structure.

[0043] Specifically, the displacement prediction includes: The encoded node features and edge features are input into the physical coding graph network, displacement prediction is performed through a node mapping formula, and the result of the displacement prediction is defined as the result of node mapping. wherein, is the result of node mapping, is the initial feature vector of the node , is the set of neighbor nodes of the node , is the node feature of the always-updated neighbor node , is the edge feature between the node and its neighbor nodes, denotes aggregation of the node and its neighbor nodes, and are multilayer perceptrons. It should be noted that the design of node mapping is based on numerical value iteration theory, i.e., an iteration formula. As can be seen from the iteration formula, the iteration process includes two node features, i.e., an un-updated load feature and an always-updated displacement feature, and the displacement feature is aggregated in the updating process, while the load feature is stacked.

[0044] Specifically, internal force prediction includes:

[0045] The encoded node features and edge features are input into the physical coding graph network, internal force prediction is performed through an edge mapping formula, and the result of the internal force prediction is defined as the result of edge mapping. The edge mapping formula is: wherein, is the result of edge mapping between the node and its neighbor nodes, is the node feature of the always-updated node , is the node feature of the always-updated neighbor node , is the edge feature between the node and its neighbor nodes, and are multilayer perceptrons.

[0046] ​​It should be noted that in a specific application, the node mapping formula is embedded in the node mapping module, the edge mapping formula is embedded in the edge mapping module, and the node mapping module is connected to the edge mapping module in the physical encoding graph network, so as to realize the node mapping and edge mapping in sequence.

[0047] The design principle of the physical encoding graph network is to focus on the causal relationship of the physical quantity calculation of the physical formula, and combine the causal relationship to design the update order of the node features and the edge features in the network, such as message aggregation and vector superposition. In addition, it should be specially pointed out that the advantage of the physical guide network design is that for the vectors appearing in the iterative formula such as , if not as input of the network, the features can be expressed by a mapping function such as a multi-layer perceptron, which has better flexibility.

[0048] To analyze the application effect of the embodiment, the same training data, test data and optimal value are used to compare the GCN (graph convolutional neural network), GIN (graph isomorphism neural network), GAT (graph attention network) and the structure response fast prediction method based on the physical encoding graph network of the embodiment. Specifically, Figure 4 shows the prediction based on GCN, Figure 5 shows the prediction of GIN, Figure 6 shows the prediction of GAT, Figure 7 shows the prediction of the embodiment. The Figures 4 to 7 comparison, combined with Figure 8 the error bar comparison shown, we find that the prediction of the structure response fast prediction method based on the physical encoding graph network of the embodiment is significantly better than the prediction of GCN, GIN and GAT.

[0049] In a specific application of the embodiment, as shown in Figure 9 , in a truss structure, the black circle solid circle represents a free node, the red solid triangle represents a constraint node, the red arrow represents the load applied on the corresponding node, and the length of the red arrow represents the size of the applied load. Figure 10 shows the true displacement based on the load condition of Figure 9 , Figure 11 shows the predicted displacement based on the displacement prediction of the embodiment. By Figure 10 and Figure 11 , we can clearly see that the displacement prediction of the embodiment still has a high precision prediction result under the condition of a small amount of response data set, and at least provides a better technical solution for the structure response in the field of civil engineering.

[0050] Still based on the application shown in Figure 9 , in the truss structure, Figure 12 shows the predicted displacement based on Figure 9The true internal force obtained under the load condition of Figure 13 The predicted internal force based on the internal force prediction of the present embodiment is shown. By Figure 12 And Figure 13 Through comparison, it can be determined that the internal force prediction of the present embodiment still has a high-precision prediction result under the condition of a small amount of response data set, and at least provides a more optimal technical solution for structural response in the field of civil engineering.

[0051] Based on the above application, it can be deduced that in the field of civil engineering, when analyzing truss structures as a representative of the truss structure, the mechanical calculation logic followed by the truss structure corresponds to the mathematical logic of the present embodiment. Based on this corresponding relationship, the present embodiment is suitable for engineering component analysis in the field of civil engineering to achieve rapid prediction of structural response.

[0052] Based on the above description, the structural response rapid prediction method based on the physical coding graph network of the present embodiment realizes the following technical innovations: (1) The numerical simulation theory is used as the theoretical support to design the physical coding graph network, which greatly enhances the physical interpretability and generalization performance of the model, and has obvious advantages compared with conventional graph networks.

[0053] (2) The stiffness influence matrix is used as the normalized coding of the structural attribute. This method not only realizes efficient extraction of the structural attribute features, but also promotes the generalization of the load point features as an edge feature processing method.

[0054] Based on the above technical innovations, the present embodiment realizes the following technical effects: The physical coding graph network retains the data flow and training method of the data-driven model, and the efficient structural attribute coding method and physically guided architecture design greatly enhance the physical interpretability and generalization performance of the model, achieving high generalization prediction accuracy with small samples, thereby solving the technical problems that the existing technology needs to be repeatedly operated in the face of similar load conditions in structural analysis, and the calculation is complicated and time-consuming in the complex multi-condition scene of structural operation and maintenance, which is difficult to meet the demand for rapid iterative design and real-time performance evaluation, especially suitable for analysis of engineering components in the field of civil engineering that meet the mechanical calculation logic.

[0055] It should be understood that the above is only for illustration and does not constitute any limitation on the technical solutions of the present application. In specific applications, those skilled in the art can set it up according to the needs, and the present application does not limit it.

[0056] It should be noted that the above-described workflow is only illustrative and does not limit the scope of protection of the present application. In actual application, those skilled in the art can select part or all of them to achieve the purpose of the present embodiment scheme according to actual needs, which is not limited here.

[0057] It is to be understood that the terminology "including", "comprising", or any other variation thereof, is intended to cover a non-exclusive inclusion such that process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0058] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and necessary general hardware platform, of course, also can be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of software product, and the computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk), including a plurality of instructions to make a terminal device (may be a mobile phone, computer, server, air conditioner, or network equipment, etc.) execute the method described in various embodiments of the present application.

[0059] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A fast prediction method for structural response based on physical coding graph networks, characterized in that, Includes the following steps: S1: Construct a physical coding graph network, which is used to encode, message pass and decode the physical parameters of the target structure to perform structural response, the structural response including at least displacement prediction and internal force prediction; S2: Input the physical parameters of the target structure into the physical coding graph network encoding; S3: The encoded physical parameters are processed by message passing, which includes sequential node mapping and edge mapping, wherein the data flow of the mapping network conforms to mathematical logic, the node mapping is used to perform displacement prediction, and the edge mapping is used to perform internal force prediction. S4: Decode the results of the node mapping and the edge mapping to obtain the displacement prediction result and the internal force prediction result.

2. The fast structural response prediction method based on physical coding graph networks as described in claim 1, characterized in that, The construction of the physical coding graph network includes the following steps: A1: Obtain the geometric relationship and physical parameters of the target structure; wherein, the physical parameters include node data and component data of the physical properties of the target structure constrained by the geometric relationship; obtain and encode node features based on the node data, and obtain and encode edge features based on the component data; A2: Obtain a small amount of response dataset of the target structure; train the physical coding graph network based on A1 and the response dataset, so that the physical coding graph network encodes the node features and the edge features and then performs the message passing based on different input load conditions, and decodes the different output node mappings and edge mappings; wherein, the response dataset stores the actual response of the target structure under different load conditions, the load conditions act on the nodes corresponding to the node data, and the node data is used as input; A3: Output the trained physical coding graph network.

3. The fast structural response prediction method based on physical coding graph networks as described in claim 2, characterized in that, The nodes include free nodes and constrained nodes to which loads are applied; Define the node data of the free node as load node data; Define the node data of the constraint node as constraint node data; The node data includes the load node data and the constraint node data.

4. The fast structural response prediction method based on physical coding graph networks as described in claim 3, characterized in that, The process of obtaining and encoding node features based on the node data includes: Based on the load node data, the load node characteristics are obtained; The load node features are encoded as follows: in, Indicates load nodes The stiffness is defined as the stiffness of the load node data. Number the load nodes. Represents a node The set of surrounding neighboring nodes, Indicates the relationship with surrounding nodes For the node The effect of stiffness.

5. The fast structural response prediction method based on physical coding graph networks as described in claim 3, characterized in that, The process of obtaining and encoding node features based on the node data further includes: Based on the constraint node data, the constraint node features are obtained; The constraint node features are encoded as follows: in, Represents constraint nodes The stiffness of the constraint node is set to infinity to express the constraint property.

6. The fast structural response prediction method based on physical coding graph networks as described in claim 4, characterized in that, The step of obtaining and encoding edge features based on the component data includes: Obtain the component data corresponding to the load node data to obtain the load node edge features; Based on the encoding of the load node features and the load node edge features, the load node edge features are encoded as follows: in, The first adjacent edge and load node after encoding edge features, For load nodes The surrounding neighbor nodes corresponding to its first adjacent edge Stiffness between The second adjacent edge and load node after encoding edge features, For load nodes The surrounding neighbor nodes corresponding to its second adjacent edge Stiffness between The third adjacent edge and load node after encoding edge features, For load nodes The surrounding neighbor nodes corresponding to its third adjacent edge The stiffness between them.

7. The fast structural response prediction method based on physical coding graph networks as described in claim 5, characterized in that, The step of obtaining and encoding edge features based on the component data further includes: Obtain the component data corresponding to the constraint node data to obtain the constraint node edge features; Based on the encoding of the constraint node features and the constraint node edge features, the constraint node edge features are encoded as follows: in, The first adjacent edge and constraint node after encoding edge features, The second adjacent edge and constraint node after encoding Edge features.

8. The fast structural response prediction method based on physical coding graph networks as described in any one of claims 6 or 7, characterized in that, The displacement prediction includes: The encoded node features and edge features are input into the physical coding graph network, and displacement prediction is performed using a node mapping formula. The result of this displacement prediction is defined as the result of the node mapping; wherein, the node mapping formula is: in, The result of the node mapping, It is a node The initial eigenvectors, It is a node neighboring nodes The set, It is a continuously updated neighbor node Node characteristics, It is an unupdated node. and its neighboring nodes Edge features between them This represents the aggregation of a node and its neighboring nodes. and It is a multilayer perceptron.

9. The fast prediction method for structural response based on physical coding graph networks as described in any one of claims 6 or 7, characterized in that, The internal force prediction includes: The encoded node features and edge features are input into the physical coding graph network, and internal force prediction is performed using an edge mapping formula. The result of this internal force prediction is defined as the result of the edge mapping; wherein, the edge mapping formula is: in, For nodes and its neighboring nodes The result of the edge mapping between them It is a continuously updated node. Node characteristics, It is a continuously updated neighbor node Node characteristics, It is an unupdated node. and its neighboring nodes Edge features between them It is a multilayer perceptron.

10. The fast prediction method for structural response based on physical coding graph networks as described in claim 2, characterized in that, The target structure includes at least a rod system structure; When the target structure is the rod system structure, the node is the connection point of the rod system structure, and the component is the rod of the rod system structure; The structural calculations follow the mathematical logic described above, which includes at least the structural mechanics calculation logic.

Images