Method for simulating analysis of honeycomb panel peel strength by using digital twinning technology

Abstract

The application discloses a method for simulating and analyzing honeycomb plate peeling strength by using digital twin technology, and belongs to the technical field of electronic digital data processing, which comprises the following steps: obtaining parameter data of a three-dimensional geometric model of the honeycomb plate, performing mesh division on the three-dimensional geometric model of the honeycomb plate to obtain an initial mesh model; according to material distribution characteristics in the initial mesh model, obtaining the elastic modulus and shear strength parameters of each mesh from a preset material characteristic table to determine stress distribution data; for the stress distribution data, calculating the stress response of the three-dimensional geometric model of the honeycomb plate in the peeling process to obtain a stress-time related response data set; converting the response data set into a multi-dimensional feature vector representing the frequency domain characteristics of the stress, and outputting a predicted value of the peeling strength by using a random forest regression model. The method for simulating and analyzing the honeycomb plate peeling strength by using the digital twin technology solves the problem of low prediction accuracy of the current simulation analysis.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology. Background Technology

[0002] As a lightweight and high-strength new material, honeycomb panels play an irreplaceable and important role in aerospace, construction and transportation. Their unique structural design achieves an excellent balance between weight and strength, making them a key material that has attracted much attention in modern engineering.

[0003] However, despite the widespread application of honeycomb panels, existing methods for evaluating their performance still have significant shortcomings. The peel strength of honeycomb panels, as one of their core performance characteristics, is influenced by both their complex three-dimensional structure and material properties, making it difficult for traditional modeling methods to accurately reproduce their true stress state. This lack of modeling accuracy directly leads to discrepancies between simulation analysis and actual experimental results, thus significantly reducing the reliability of predictions. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides a method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology, thereby solving the aforementioned problems.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology, comprising the following steps:

[0006] S1: Obtain parameter data of the three-dimensional geometric model of the honeycomb panel through a pre-established three-dimensional geometric model of the honeycomb panel. The parameter data includes the size of the honeycomb unit and the material distribution characteristics. Then, perform meshing on the three-dimensional geometric model of the honeycomb panel to obtain an initial mesh model.

[0007] S2: Based on the material distribution characteristics in the initial mesh model, establish a layered index in the preset material property table, obtain the elastic modulus and shear strength parameters reflecting the peel strength in each mesh from the preset material property table, and determine the stress distribution data under the displacement constraint boundary conditions;

[0008] S3: For stress distribution data, ABAQUS is used to dynamically calculate the stress response of the three-dimensional geometric model of the honeycomb panel during the peeling process, and stress and time-related response datasets are obtained.

[0009] S4: Transform the response dataset into a multidimensional feature vector representing the frequency domain characteristics of stress. Use a trained random forest regression model, take the multidimensional feature vector as input, and output the predicted value of the peeling intensity.

[0010] Preferably, in step S1, a three-dimensional geometric model of a honeycomb panel is extracted from a preset honeycomb panel model database. After verifying the data integrity of the three-dimensional geometric model of the honeycomb panel, the size of the honeycomb unit and the material distribution characteristics of the three-dimensional geometric model of the honeycomb panel are extracted to obtain a classification parameter dataset.

[0011] The triangulation algorithm is driven by the classification parameter dataset to divide the geometric surface of the honeycomb panel into multiple triangular units. Each triangular unit is used as a grid unit, and each grid unit is assigned a unique grid number to generate the initial grid structure.

[0012] Optionally, in step S2, a material type identifier for each mesh is obtained from the material distribution characteristics in the initial mesh model, wherein the identifier is an integer value;

[0013] Based on the material type identifier, a hierarchical index is created in the material property table material_properties in the PostgreSQL database via the JDBC interface to establish a relationship between the material type identifier and the material ID in the material property table material_properties according to the material type.

[0014] Based on the material type identifier, query the parameters in the elastic modulus and shear strength fields of the material property table corresponding to the material ID corresponding to the material type identifier, convert the elastic modulus and shear strength parameters to MPa unit system, and obtain the material property dataset.

[0015] Specifically, in step S2, the Scikit-learn SVC classifier is used on the material property dataset, with the kernel function being the radial basis function, and ±2σ is set as the outlier threshold. Mesh cells that exceed the threshold are marked, and the classified_parameters.csv file is generated.

[0016] Based on the classified_parameters.csv file, the displacement constraint boundary of the initial mesh model is set. The Cauchy stress tensor is calculated using the second-order central difference scheme through the ANSYS Mechanical APDL platform, with the time step satisfying the CFL condition, to obtain the initial stress distribution data, which is then output to the stress_initial.rst file.

[0017] It is worth noting that in step S3, the complete mesh file and step size control file are loaded in the ABAQUS explicit dynamics module, the simulation calculation is performed, and the simulated Cauchy stress tensor of the mesh element integration point at each time step is output.

[0018] Principal stress components are extracted from the simulated Cauchy stress tensor. A sixth-order polynomial model is fitted to the principal stress components using the leastsq function of SciPy. The stress-time curve is obtained based on the fitting results of the sixth-order polynomial model, and the relationship between stress and time is used as the response dataset.

[0019] Preferably, in step S3, the node coordinate matrix and element connection table are obtained from the stress distribution data in the stress_initial.rst file, and the spatial location data corresponding to the node number is extracted from the node coordinate matrix;

[0020] Based on spatial location data and element connection table, generate solid mesh elements for the honeycomb core layer and panel in ABAQUS, and output a complete mesh file containing the element type, panel type and material properties of the honeycomb core layer.

[0021] Specifically, in step S3, the Newmark-β method is used to set the time integration parameters, the load step increment is determined by dividing the material yield strength by a preset ratio, and the time step for dynamic simulation calculation is generated based on the load step increment.

[0022] Optionally, in step S4, a response dataset is obtained, and the time-domain signal in the response dataset is processed by a fast Fourier transform to obtain a frequency-domain feature set. The frequency-domain feature set is then integrated into a multi-dimensional feature vector for characterizing the frequency-domain properties of stress.

[0023] It is worth noting that in step S4, the multidimensional feature vector is input into a pre-trained random forest regression model, which is trained using historical data.

[0024] The dynamic response feature set is analyzed using the random forest regression model to obtain the predicted value of the stripping intensity.

[0025] The beneficial effects of this invention are as follows: In the method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology, a three-dimensional geometric model of the honeycomb panel is established and meshed using digital twin technology, a hierarchical index is constructed, and elastic modulus and shear strength parameters are obtained to determine stress distribution data. The stress response of the honeycomb panel during the peeling process is dynamically calculated using ABAQUS software to obtain a stress-time response dataset. The response dataset is transformed into a multi-dimensional feature vector and used as input to a random forest regression model to predict the peel strength. This invention achieves accurate prediction of the peel strength of honeycomb panels by establishing a correlation with the three-dimensional structure and material properties. Attached Figure Description

[0026] Figure 1This is a flowchart illustrating a method for simulating and analyzing the peel strength of a honeycomb panel using digital twin technology, as described in one embodiment of the present invention.

[0027] Figure 2 This is a flowchart illustrating the steps of step S2 in one embodiment of the present invention;

[0028] Figure 3 This is a flowchart illustrating step S3 in one embodiment of the present invention. Detailed Implementation

[0029] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that these descriptions are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0030] like Figure 1-3 As shown, a method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology includes the following steps:

[0031] S1: Obtain parameter data of the three-dimensional geometric model of the honeycomb panel through the pre-established three-dimensional geometric model of the honeycomb panel. The parameter data includes the size of the honeycomb unit and the material distribution characteristics. Mesh the three-dimensional geometric model of the honeycomb panel to obtain an initial mesh model for subsequent stress simulation.

[0032] S2: Based on the material distribution characteristics in the initial mesh model, establish a layered index in the preset material property table, obtain the elastic modulus and shear strength parameters reflecting the peel strength in each mesh from the preset material property table, and determine the stress distribution data under the displacement constraint boundary conditions;

[0033] S3: For the stress distribution data, ABAQUS (finite element analysis software) is used to dynamically calculate the stress response of the three-dimensional geometric model of the honeycomb panel during the peeling process, and obtain the stress and time-related response dataset;

[0034] S4: Transform the response dataset into a multidimensional feature vector representing the frequency domain characteristics of stress. Use a trained random forest regression model, take the multidimensional feature vector as input, and output the predicted value of the peeling intensity.

[0035] In the method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology, a three-dimensional geometric model of the honeycomb panel is established and meshed using digital twin technology. A hierarchical index is constructed, and elastic modulus and shear strength parameters are obtained to determine stress distribution data. The stress response of the honeycomb panel during the peeling process is dynamically calculated using ABAQUS software, resulting in a stress-time response dataset. This response dataset is then transformed into a multi-dimensional feature vector, which is used as input to a random forest regression model to predict the peel strength. This invention achieves accurate prediction of the peel strength of honeycomb panels by establishing a correlation with the three-dimensional structure and material properties.

[0036] It is worth noting that in step S1, the three-dimensional geometric model of the honeycomb panel is extracted from the preset honeycomb panel model database. After verifying the data integrity of the three-dimensional geometric model of the honeycomb panel, the size of the honeycomb unit and the material distribution characteristics of the three-dimensional geometric model of the honeycomb panel are extracted to obtain the classification parameter dataset.

[0037] The triangulation algorithm is driven by the classification parameter dataset to divide the geometric surface of the honeycomb panel into multiple triangular units. Each triangular unit is used as a grid unit, and each grid unit is assigned a unique grid number to generate the initial grid structure.

[0038] A pre-defined honeycomb panel model database stores 3D geometric data of various honeycomb panels. During extraction, the integrity of the model is checked first, such as ensuring the closure of the geometry, the continuity of vertex data, and the correctness of the facet connections. If a model is found to be missing some vertex data, it will be marked as incomplete and discarded. After verifying the data integrity of the 3D geometric model of the honeycomb panel, a dataset of geometric parameters is obtained, which includes the overall side length, overall thickness, overall height, size of the honeycomb cells, and material distribution characteristics formed by the arrangement of each honeycomb cell. Since the honeycomb panel is composed of multiple honeycomb cells of the same size, the cell size of each honeycomb cell is the same. In one possible implementation, the cell size and material distribution characteristics can be extracted from the geometric parameter set. Taking the cell size of the honeycomb panel as an example, assuming that the side length of a standard hexagonal honeycomb cell is 5 mm and the wall thickness is 0.2 mm, the material distribution characteristics are reflected in the uniform distribution of aluminum alloy in the honeycomb core and the composite material coverage in the panel. The classification parameter dataset is thus formed.

[0039] In one possible implementation, since each cellular cell is a polygon, the triangulation algorithm driving the initial mesh structure can divide the geometric surface of the cellular panel into a large number of triangular cells based on a classification parameter dataset. Assuming each cellular cell surface is divided into approximately 100 triangles, this ensures the mesh captures geometric details.

[0040] Preferably, in step S2, a material type identifier for each grid is obtained from the material distribution characteristics in the initial grid model. This identifier is an integer value. When processing the material type identifiers of each grid in the initial grid model, it can be understood as a digital labeling method for the material properties of grid cells. When the initial grid model is formed, each grid is assigned an integer value as a material type identifier; for example, a value of 1 indicates aluminum alloy, and a value of 2 indicates composite material. This labeling method provides a basis for subsequent material property queries.

[0041] Based on the material type identifier, a hierarchical index is created in the material property table material_properties in the PostgreSQL database via the JDBC interface to establish a relationship between the material type identifier and the material ID in the material property table material_properties according to the material type.

[0042] Based on the material type identifier, the parameters in the elastic modulus and shear strength fields of the material property table `material_properties` corresponding to the material ID of the material type identifier are retrieved. The elastic modulus and shear strength parameters are then converted to MPa units to obtain the material property dataset. When querying the PostgreSQL database via the JDBC interface, the `material_properties` table stores property data for various materials. A hierarchical index quickly locates the target material corresponding to the material type identifier using the material ID and material type field. For example, querying for an aluminum alloy with a material type identifier of 1 yields an elastic modulus of 70000 MPa and a shear strength of 200 MPa. This data will be directly used for subsequent analysis to ensure the accuracy of the material properties.

[0043] Optionally, in step S2, the Scikit-learn SVC classifier is used on the material property dataset, with the kernel function being the radial basis function, and ±2σ is set as the outlier threshold. Mesh cells that exceed the threshold are marked, and the classified_parameters.csv file is generated.

[0044] Based on the classified_parameters.csv file, the displacement constraint boundary of the initial mesh model is set. The Cauchy stress tensor is calculated using the second-order central difference scheme through the ANSYS Mechanical APDL platform, with the time step satisfying the CFL condition, to obtain the initial stress distribution data, which is then output to the stress_initial.rst file. When setting the displacement constraint boundary of the initial mesh model, different boundary conditions can be applied to different material regions based on the classification results in the aforementioned CSV file. For example, in the edge region of the honeycomb panel, the aluminum alloy part is set as a fixed constraint, while the composite material part is allowed small displacements. This differentiated setting can better simulate actual working conditions. When calculating the Cauchy stress tensor using the ANSYS Mechanical APDL platform, the calculation accuracy can be improved using a second-order central difference scheme. Assuming the time step is set to 0.001 seconds according to the CFL condition, numerical stability is ensured. The CFL condition is an important criterion for ensuring the stability of numerical solutions in numerical analysis, requiring that the propagation speed of the numerical method does not exceed the signal propagation speed of the physical problem. Specifically, the initial mesh model is discretized using the set displacement constraint boundary of the initial mesh model. Combined with the second-order central difference scheme, a time step parameter is introduced, and the step size is adjusted to satisfy the CFL condition, resulting in an adjusted time step. Starting from the adjusted time step, the discretized initial mesh model is used as input to iteratively calculate the component data of the Cauchy stress tensor. By summarizing the component data of the Cauchy stress tensor at each time step, initial stress distribution data is generated. The obtained initial stress distribution data is output to the stress_initial.rst file to provide basic data for subsequent optimization.

[0045] Specifically, in step S3, the complete mesh file and step size control file are loaded in the ABAQUS explicit dynamics module, simulation calculations are performed, and the simulated Cauchy stress tensor of the mesh element integration points at each time step is output. When calculating the simulated Cauchy stress tensor in the ABAQUS explicit dynamics module, the stress component data of the mesh element integration points can be output for each time step, such as 0.001 seconds.

[0046] The principal stress components σ11, σ22, and σ33 are extracted from the simulated Cauchy stress tensor. A sixth-order polynomial model is fitted using SciPy's leastsq function for these principal stress components. The stress-time curve is obtained based on the fitting results of the sixth-order polynomial model, and the relationship between stress and time is used as the response dataset. Assuming that the peak value of the σ11 component is 150 MPa at a certain time step, these data are recorded to lay the foundation for subsequent analysis. When fitting the stress-time curve using SciPy's leastsq function, a sixth-order polynomial model can be constructed for the principal stress components corresponding to σ11, σ22, and σ33. The fitted stress variation parameters are obtained from the sixth-order polynomial model. By combining the fitted stress variation parameters with the time variation data, a stress-time curve is generated, showing the trend of stress change over time, such as a non-linear growth trend.

[0047] It is worth noting that in step S3, the node coordinate matrix and element connection table are obtained from the stress distribution data in the stress_initial.rst file, and the spatial position data corresponding to the node number is extracted from the node coordinate matrix. In this embodiment, the mesh elements are connected by nodes, the node coordinate matrix contains the x, y, z three-dimensional spatial position data of each node number, and the element connection table records the topological relationship between each mesh element and the node. In one possible implementation, it is assumed that the extracted node number range is 1 to 5000, and the corresponding spatial position data is stored in millimeters to ensure the accuracy of subsequent mesh reconstruction.

[0048] Based on spatial location data and element connection tables, solid mesh elements for the honeycomb core and panels are generated in ABAQUS, outputting a complete mesh file containing the element type, panel type, and material properties of the honeycomb core. When generating the solid element mesh for the honeycomb core and panels in ABAQUS, the core can be set as hexahedral elements and the panels as shell elements based on the spatial location data corresponding to the extracted node numbers. Assuming the element type of the honeycomb core is C3D8, the panel type is S4R, and the material properties correspond to aluminum alloy and composite materials respectively, the output is sent to an INP file to obtain the complete mesh file. This method ensures the consistency between the mesh generation and the actual structure.

[0049] Preferably, in step S3, the Newmark-β method (a method that generalizes the linear acceleration method) is used to set the time integration parameters. The load step increment is determined by dividing the material yield strength by a preset ratio, and the time step for dynamic simulation calculation is generated based on the load step increment. When setting the time integration parameters using the Newmark-β method, the β value can be set to 0.25 and the γ value to 0.5 to ensure numerical stability. The load step increment is obtained by dividing the material yield strength, such as 200 MPa, by a scaling factor of 2. This setting is suitable for stress analysis under dynamic loading; in this embodiment, through... Calculate the yield strength of the material, where For the shear strength of the material, The yield strength of the material.

[0050] Optionally, in step S4, a response dataset is obtained, and the time-domain signal in the response dataset is processed by a fast Fourier transform to obtain a frequency-domain feature set. The frequency-domain feature set is then integrated into a multi-dimensional feature vector for characterizing the frequency-domain properties of stress.

[0051] The response dataset contains stress values ​​at multiple time points during the peeling process of the cellular panel; for example, the stress is 50 MPa at 0.1 seconds and increases to 75 MPa at 0.2 seconds. This data forms the basis of the time-domain signal, providing the initial foundation for subsequent analysis. When processing the time-domain signal using Fast Fourier Transform (FFT), the stress-time data can be converted into a frequency-domain signal, extracting features such as the dominant frequency amplitude and phase difference. Assuming the transformation reveals a dominant frequency of 10 Hz, an amplitude of 30 Hz, and a corresponding phase difference of 0.5 radians, this set of frequency-domain features reflects the periodic variation of stress during the peeling process. It should be noted that this transformation helps reveal hidden frequency components in the signal, laying the foundation for subsequent feature construction. When constructing a multi-dimensional feature vector, the frequency-domain feature set can be further organized into a multi-dimensional feature vector, such as combining the dominant frequency amplitude, phase difference, and other secondary frequency components to form a complete feature description. This multi-dimensional feature vector can comprehensively characterize the dynamic response characteristics during the peeling process.

[0052] It is worth noting that in step S4, the multidimensional feature vector is input into a pre-trained random forest regression model, which is trained using historical data.

[0053] The dynamic response feature set is analyzed using the random forest regression model to obtain the predicted value of the stripping intensity.

[0054] When inputting multidimensional feature vectors into a random forest regression model, the model has already been trained on a large amount of historical peeling experimental data, such as stress-time data and corresponding peeling strength values ​​from the past 100 sets of honeycomb panel peeling experiments. The random forest regression model predicts based on the input multidimensional feature vectors and outputs a specific peeling strength value, such as a predicted value of 8.5 kN. It should be noted that the advantage of the random forest model lies in its ability to handle high-dimensional feature data and its good fitting ability for nonlinear relationships.

[0055] By combining frequency domain features and machine learning models, we can gain a deeper understanding of the mechanical behavior during the peeling process and provide data support for optimizing material structure design.

[0056] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and these variations still fall within the protection scope of the present invention.

Claims

1. A method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology, characterized in that, Includes the following steps: S1: Obtain parameter data of the three-dimensional geometric model of the honeycomb panel through a pre-established three-dimensional geometric model of the honeycomb panel. The parameter data includes the size of the honeycomb unit and the material distribution characteristics. Then, perform meshing on the three-dimensional geometric model of the honeycomb panel to obtain an initial mesh model. S2: Based on the material distribution characteristics in the initial mesh model, establish a layered index in the preset material property table, obtain the elastic modulus and shear strength parameters reflecting the peel strength in each mesh from the preset material property table, and determine the stress distribution data under the displacement constraint boundary conditions; S3: For stress distribution data, ABAQUS is used to dynamically calculate the stress response of the three-dimensional geometric model of the honeycomb panel during the peeling process, and stress and time-related response datasets are obtained. S4: Transform the response dataset into a multidimensional feature vector representing the frequency domain characteristics of stress. Use a trained random forest regression model, take the multidimensional feature vector as input, and output the predicted value of the peeling intensity.

2. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 1, characterized in that: In step S1, a three-dimensional geometric model of a honeycomb panel is extracted from a preset honeycomb panel model database. After verifying the data integrity of the three-dimensional geometric model of the honeycomb panel, the size of the honeycomb unit and the material distribution characteristics of the three-dimensional geometric model of the honeycomb panel are extracted to obtain a classification parameter dataset. The triangulation algorithm is driven by the classification parameter dataset to divide the geometric surface of the honeycomb panel into multiple triangular units. Each triangular unit is used as a grid unit, and each grid unit is assigned a unique grid number to generate the initial grid structure.

3. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 1, characterized in that: In step S2, the material type identifier for each mesh is obtained from the material distribution characteristics in the initial mesh model, and the identifier is an integer value; Based on the material type identifier, a hierarchical index is created in the material property table material_properties in the PostgreSQL database via the JDBC interface to establish a relationship between the material type identifier and the material ID in the material property table material_properties according to the material type. Based on the material type identifier, query the parameters in the elastic modulus and shear strength fields of the material property table corresponding to the material ID corresponding to the material type identifier, convert the elastic modulus and shear strength parameters to MPa unit system, and obtain the material property dataset.

4. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 3, characterized in that: In step S2, the Scikit-learn SVC classifier is used on the material property dataset, with the kernel function being the radial basis function. ±2σ is set as the outlier threshold, and grid cells exceeding the threshold are marked to generate the classified_parameters.csv file. Based on the classified_parameters.csv file, the displacement constraint boundary of the initial mesh model is set. The Cauchy stress tensor is calculated using the second-order central difference scheme through the ANSYS Mechanical APDL platform, with the time step satisfying the CFL condition, to obtain the initial stress distribution data, which is then output to the stress_initial.rst file.

5. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 1, characterized in that: In step S3, the complete mesh file and step size control file are loaded into the ABAQUS explicit dynamics module, simulation calculations are performed, and the simulated Cauchy stress tensor of the mesh element integration points at each time step is output. Principal stress components are extracted from the simulated Cauchy stress tensor. A sixth-order polynomial model is fitted to the principal stress components using the leastsq function of SciPy. The stress-time curve is obtained based on the fitting results of the sixth-order polynomial model, and the relationship between stress and time is used as the response dataset.

6. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 5, characterized in that: In step S3, the node coordinate matrix and element connection table are obtained from the stress distribution data in the stress_initial.rst file, and the spatial location data corresponding to the node number is extracted from the node coordinate matrix. Based on spatial location data and element connection table, generate solid mesh elements for the honeycomb core layer and panel in ABAQUS, and output a complete mesh file containing the element type, panel type and material properties of the honeycomb core layer.

7. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 6, characterized in that: In step S3, the Newmark-β method is used to set the time integration parameters, the load step increment is determined by dividing the material yield strength by a preset ratio, and the time step for dynamic simulation calculation is generated based on the load step increment.

8. The method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 1, characterized in that: In step S4, a response dataset is obtained, and the time-domain signal in the response dataset is processed by a fast Fourier transform to obtain a frequency-domain feature set. The frequency-domain feature set is then integrated into a multi-dimensional feature vector for characterizing the frequency-domain properties of stress.

9. A method for simulating and analyzing the peel strength of honeycomb panels using digital twin technology according to claim 8, characterized in that: In step S4, the multidimensional feature vector is input into a pre-trained random forest regression model, which is trained using historical data. The response dataset is analyzed using the random forest regression model to obtain a predicted value for the stripping intensity.

Images