Dynamic finite element agent model construction method and device based on random forest

Abstract

The invention belongs to the technical field of agent model construction, and particularly relates to a dynamic finite element agent model construction method and device based on a random forest. The method comprises the following steps: S1, acquiring finite element simulation data of an aircraft structure, wherein the finite element simulation data comprises physical quantities of grid units and node coordinates of four nodes surrounding the grid units; s2, determining the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; s3, correcting the physical quantity of each node based on the test process; and S4, training a plurality of preset random forest models of different specifications by taking node coordinates and a test process as expected input and taking the corrected physical quantity of the node as expected output, and selecting an optimal random forest model as a dynamic finite element agent model by using a preset evaluation index. According to the method, mutual mapping of physical quantities and finite element simulation results in the whole test process is realized, and the airplane structural strength test efficiency is improved.

Description

Technical Field

[0001] The present application belongs to the technical field of agent model construction, and in particular relates to a method and device for constructing a dynamic finite element agent model based on random forest. Background Art

[0002] Finite element simulation is a necessary step to evaluate the structural strength of test pieces, simulate the test process, and predict loading damage before the aircraft structural strength test. However, the finite element calculation results based on full-field simulation are difficult to map with the physical test results based on discrete strain gauge sensors. This is because the layout of discrete strain gauge sensors is difficult to completely cover the entire test component of the aircraft structure. Discrete strain gauge sensors generally only specify their spatial X, Y, and Z position coordinates. When mapping specific strain gauges to finite element model units, it is necessary to consider factors such as spatial coordinates, pasting position, detection direction, and model quality, and perform manual sheet selection. For the full-aircraft structural strength test, the workload is large, and the mapping process between discrete strain gauges and finite element simulation model units will consume a lot of time and manpower costs, resulting in the simulation process not fully serving the test verification process.

[0003] On the other hand, the finite element model cannot meet the demand for viewing dynamic simulation results according to the progress of physical tests. When using the finite element simulation method, the finite element model does not reflect the test condition progress and strain change process intuitively enough. It can only reflect the simulation results at the end of the test and lacks the ability to map the strain of the entire test process. Summary of the invention

[0004] In order to solve the above problems, the present application provides a method and device for constructing a dynamic finite element proxy model based on random forest, which uses a finite element proxy model to replace finite element simulation. The finite element proxy model is used to quickly obtain simulation calculation results of specific positions and specific loading times, avoiding the complicated process of manually searching for mapping units, and providing a more effective means for test data utilization, test site monitoring, and consistency evaluation of virtual and real data.

[0005] The first aspect of the present application provides a method for constructing a dynamic finite element proxy model based on random forest, which mainly includes: Step S1, obtaining finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; Step S2, determining the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; Step S3, correcting the physical quantity of each node based on the test progress; Step S4: Taking the node coordinates and the test process as the expected input and the corrected physical quantity of the node as the expected output, a plurality of preset random forest models of different specifications are trained, and the optimal random forest model is selected as the dynamic finite element proxy model using the preset evaluation index.

[0006] Preferably, in step S2, the physical quantity of the node is set to an average value of the physical quantities of each grid unit adjacent to the node.

[0007] Preferably, step S3 further comprises: When the finite element simulation data of the aircraft structure obtained is linear simulation data, the physical quantity of each node is directly corrected using the test process; When the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, a correction amount is first calculated according to the nonlinear relationship, and then the correction amount is used to correct the physical quantity of each node.

[0008] Preferably, in step S4, the evaluation index includes a mean square error index or an R-square index.

[0009] Preferably, the physical quantity includes strain.

[0010] The second aspect of the present application provides a dynamic finite element proxy model construction device based on random forest, which mainly includes: A finite element simulation data acquisition module, used to acquire finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; A node physical quantity determination module, used to determine the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; A physical quantity correction module is used to correct the physical quantity of each node based on the test process; The random forest model selection module is used to train multiple preset random forest models of different specifications with node coordinates and test progress as expected inputs and the corrected physical quantity of the node as expected outputs, and use preset evaluation indicators to select the optimal random forest model as the dynamic finite element proxy model.

[0011] Preferably, the physical quantity of the node is set to an average value of the physical quantities of each grid unit adjacent to the node.

[0012] Preferably, the physical quantity correction module includes: A correction unit based on linear simulation data, used to directly use the test process to correct the physical quantity of each node when the finite element simulation data of the aircraft structure obtained is linear simulation data; The correction unit based on nonlinear simulation data is used to first calculate the correction amount according to the nonlinear relationship when the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, and then use the correction amount to correct the physical quantity of each node.

[0013] Preferably, the evaluation index includes a mean square error index or an R-square index.

[0014] Preferably, the physical quantity includes strain.

[0015] This application realizes the mutual mapping of physical quantities in the whole test process and finite element simulation results, and improves the efficiency of aircraft structure strength testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flowchart of a preferred embodiment of the method for constructing a dynamic finite element proxy model based on random forest in the present application.

[0017] Figure 2 It is a schematic diagram to determine the physical quantity of the node. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the implementation of this application clearer, the technical scheme in the implementation of this application will be described in more detail in combination with the drawings in the implementation of this application. In the drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described implementation is a part of the implementation of this application, not all of the implementations. The implementation described below with reference to the drawings is exemplary and is intended to be used to explain this application, and cannot be understood as a limitation on this application. Based on the implementation in this application, all other implementations obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The implementation of this application is described in detail below in combination with the drawings.

[0019] The first aspect of the present application provides a method for constructing a dynamic finite element proxy model based on random forest, such as Figure 1 As shown, it mainly includes: Step S1, obtaining finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; Step S2, determining the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; Step S3, correcting the physical quantity of each node based on the test progress; Step S4: Taking the node coordinates and the test process as the expected input and the corrected physical quantity of the node as the expected output, a plurality of preset random forest models of different specifications are trained, and the optimal random forest model is selected as the dynamic finite element proxy model using the preset evaluation index.

[0020] First, the finite element proxy model is explained. In recent years, with the rise of artificial intelligence algorithms, machine learning has a strong ability to fit data. It has realized the construction of proxy models in many fields to quickly respond to the data features of the pre-trained set. Among them, the integrated model of the decision tree is particularly effective, proving that machine learning has the possibility of becoming a fast-response proxy model for finite elements. The finite element proxy model refers to the use of data processing, proxy model and other technologies to fit important information such as finite element position and test results through the constructed proxy model, that is, to train a finite element proxy model, so that the finite element position can be directly input, and the stress and strain at the position can be calculated using the trained finite element proxy model. The so-called dynamic finite element proxy model refers to the process of training the proxy model, in which not only the finite element position but also the time information is input. The proxy model obtained in this way can output different results according to different test processes, so that it can synchronously respond to the physical test process at different stages of test verification and provide support for test command decision-making.

[0021] Since the purpose of this application is to quickly obtain simulation results in order to map them with physical test data, this application preprocesses the data used for training the finite element proxy model in steps S1 to S3, and then divides these data into a training set and a test set. In step S4, the finite element proxy model is trained and tested. In step S1, first prepare the aircraft structure finite element model and simulation result file corresponding to the physical test, that is, the finite element simulation data, and check the match between the simulation result file and the finite element model. Then, HyperWorks software is used to eliminate unexpected stress concentration units based on the finite element results, and extract the information of the units in the remaining finite element model and the physical quantities of the simulation results. In some optional embodiments, the physical quantities include strain, reference Figure 2 ,Each rectangular box represents a grid unit, each grid unit has strain information, and the rectangular box where each grid unit is located has four corner points, corresponding to four nodes.

[0022] Afterwards, in step S2, the strain of the grid element is converted into the strain of the node, as Figure 2 The strain of the node A shown is calculated based on the strains of the adjacent mesh units a, b, c, and d. In some optional embodiments, in step S2, the physical quantity of the node is set to the average value of the physical quantities of the mesh units adjacent to the node. That is, the average value of the strains of the mesh units a, b, c, and d is used as the strain of node A.

[0023] In step S3, it is necessary to superimpose time information on the physical quantity of the node, that is, the strain value of the same node may be different in different test processes. In some optional implementations, step S3 further includes: When the finite element simulation data of the aircraft structure obtained is linear simulation data, the physical quantity of each node is directly corrected using the test process; When the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, a correction amount is first calculated according to the nonlinear relationship, and then the correction amount is used to correct the physical quantity of each node.

[0024] In this embodiment, for the strength test process T, if the simulation result is a linear simulation result, the test process T can be set to (0, 5%, 10%...100%), and then the strain R1 (strain value at the end of the simulation) obtained in step S2 is multiplied by the test process T, and the dynamic simulation data is indirectly calculated, that is, the strain R corresponding to a certain test process T is set to: R=T×R1. If the simulation result is a nonlinear simulation result, the test process T is set according to the process corresponding to the nonlinear simulation result. For example, if the test process T is 20%, theoretically, the ordinate value when the horizontal coordinate is 20% in the nonlinear relationship should be found first, and then the ordinate value is compared with the ordinate value when the nonlinear relationship is 100% to calculate the correction amount, and then use the correction amount to correct the physical quantity. In this embodiment, the nonlinear relationship is a nonlinear function formed with the horizontal coordinate as the process and the vertical coordinate as the physical quantity.

[0025] The training data set for machine learning is constructed in the above manner in the form of [X, Y, Z, T|R], where X, Y, Z, and T represent the X, Y, and Z coordinates of the node and the test process, respectively, and R is the corrected strain value. Finally, in step S4, the node coordinates and the test process are the expected inputs, and the corrected strain value is the expected output, so as to train multiple random forest models of different specifications to select the optimal random forest model.

[0026] When training a random forest model, the training data set is usually divided into a training set and a test set, with a ratio of 7:3, 5:5, and 3:7, respectively, to simulate three different scenarios with a large number of finite element units, a moderate number, and a small number. In this embodiment, setting up random forest models of various specifications mainly refers to assigning different parameter values ​​to the hyperparameters of the random forest model, such as the number of decision trees n_estimators, the learning rate learning_rate, the number of features considered when splitting each node (the node here refers to the model node, not the finite element simulation node) max_features, the maximum depth of the decision tree max_depth, the minimum number of samples required for node splitting min_samples_split, the minimum number of samples required for leaf nodes min_samples_leaf, the indicator criterion used to measure the quality of node splitting, etc., using the training set to train random forest models of various specifications, using the corresponding test set for verification, and calculating the mean square error MSE or R square indicator R between the predicted strain and the simulated strain (expected output) 2 , mean square error MSE is used to reflect the error between the predicted value and the actual simulation strain data, and R square index R 2 It is used to reflect the deviation between the predicted value and the actual simulated strain data, and then select the random forest model with the optimal specifications.

[0027] Finally, check the fitting effect. For the optimal random forest model trained for all training data sets, input all node coordinates of the finite element model, predict the strain value, form a cloud map under the same color spectrum, and compare the finite element simulation cloud map and the proxy model fitting cloud map. If the fitting effect is acceptable, complete the construction of the proxy model. If the fitting effect is poor, reset the random forest specifications for verification.

[0028] Through the above method, this application focuses on the position of the physical quantities of the strength test and the dynamic process of the test, ensuring that the strength test site can quickly obtain the simulation calculation results of the specific position and specific loading time through the finite element proxy model, avoiding the complicated process of manually searching for mapping units, and providing a more effective means for test data utilization, test site monitoring and consistency evaluation of virtual and real data.

[0029] The second aspect of the present application provides a dynamic finite element proxy model construction device based on random forest corresponding to the above method, mainly comprising: A finite element simulation data acquisition module, used to acquire finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; A node physical quantity determination module, used to determine the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; A physical quantity correction module is used to correct the physical quantity of each node based on the test process; The random forest model selection module is used to train multiple preset random forest models of different specifications with node coordinates and test progress as expected inputs and the corrected physical quantity of the node as expected outputs, and use preset evaluation indicators to select the optimal random forest model as the dynamic finite element proxy model.

[0030] In some optional implementations, the physical quantity of the node is set to an average value of the physical quantities of each grid unit adjacent to the node.

[0031] In some optional implementations, the physical quantity correction module includes: A correction unit based on linear simulation data, used to directly use the test process to correct the physical quantity of each node when the finite element simulation data of the aircraft structure obtained is linear simulation data; The correction unit based on nonlinear simulation data is used to first calculate the correction amount according to the nonlinear relationship when the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, and then use the correction amount to correct the physical quantity of each node.

[0032] In some optional implementations, the evaluation index includes a mean square error index or an R-square index.

[0033] In some optional embodiments, the physical quantity includes strain.

[0034] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for constructing a dynamic finite element proxy model based on random forest, characterized in that: include: Step S1, obtaining finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; Step S2, determining the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; Step S3, correcting the physical quantity of each node based on the test progress; Step S4: Taking the node coordinates and the test process as the expected input and the corrected physical quantity of the node as the expected output, a plurality of preset random forest models of different specifications are trained, and the optimal random forest model is selected as the dynamic finite element proxy model using the preset evaluation index.

2. The method for constructing a dynamic finite element proxy model based on random forest according to claim 1, characterized in that: In step S2, the physical quantity of the node is set to the average value of the physical quantities of each grid unit adjacent to the node.

3. The method for constructing a dynamic finite element proxy model based on random forest according to claim 1, characterized in that: Step S3 further comprises: When the finite element simulation data of the aircraft structure obtained is linear simulation data, the physical quantity of each node is directly corrected using the test process; When the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, a correction amount is first calculated according to the nonlinear relationship, and then the correction amount is used to correct the physical quantity of each node.

4. The method for constructing a dynamic finite element proxy model based on random forest according to claim 1, characterized in that: In step S4, the evaluation index includes a mean square error index or an R-square index.

5. The method for constructing a dynamic finite element proxy model based on random forest according to any one of claims 1 to 4, characterized in that: The physical quantity includes strain.

6. A dynamic finite element proxy model construction device based on random forest, characterized in that: include: A finite element simulation data acquisition module, used to acquire finite element simulation data of the aircraft structure, including physical quantities of each grid unit and node coordinates of four nodes surrounding the grid unit; A node physical quantity determination module, used to determine the physical quantity of each node according to the physical quantities of a plurality of grid units adjacent to the node; A physical quantity correction module is used to correct the physical quantity of each node based on the test process; The random forest model selection module is used to train multiple preset random forest models of different specifications with node coordinates and test progress as expected inputs and the corrected physical quantity of the node as expected outputs, and use preset evaluation indicators to select the optimal random forest model as the dynamic finite element proxy model.

7. The random forest-based dynamic finite element proxy model construction device according to claim 6, characterized in that: The physical quantity of the node is set to the average value of the physical quantities of each grid unit adjacent to the node.

8. The random forest-based dynamic finite element proxy model construction device according to claim 6, characterized in that: The physical quantity correction module comprises: A correction unit based on linear simulation data, used to directly use the test process to correct the physical quantity of each node when the finite element simulation data of the aircraft structure obtained is linear simulation data; The correction unit based on nonlinear simulation data is used to first calculate the correction amount according to the nonlinear relationship when the acquired finite element simulation data of the aircraft structure is nonlinear simulation data, and then use the correction amount to correct the physical quantity of each node.

9. The random forest-based dynamic finite element proxy model construction device according to claim 6, characterized in that: The evaluation index includes a mean square error index or an R-square index.

10. The random forest-based dynamic finite element proxy model construction device according to any one of claims 6 to 9, characterized in that: The physical quantity includes strain.

Images