A method for identifying quality problems of panel structures based on digital twins

Abstract

The present invention discloses a method for identifying quality problems of panel structures based on digital twins, which relates to the field of intelligent detection of aerospace equipment; the method includes obtaining the structural and material parameters of the target panel and the load conditions and boundary conditions applied to the target panel; establishing a simulation analysis model and performing mechanical calculations to obtain the mechanical response of the simulation analysis; obtaining the mechanical response obtained by experimental monitoring of the target panel; establishing a digital twin model, and obtaining the mechanical response of the digital twin according to the mechanical response of the simulation analysis and the mechanical response obtained by experimental monitoring; the digital twin model is constructed by using a data fusion method; according to the mechanical response of the simulation analysis and the mechanical response of the digital twin, calculate the response difference; if the response difference is within the set difference range, it is determined that there are quality problems in the structure of the target panel; the present invention can efficiently and accurately identify the quality problems of panel structures.

Description

Technical Field

[0001] The present invention relates to the field of intelligent detection of aerospace equipment, and particularly to a method for identifying quality problems of panel structures based on digital twins. Background Art

[0002] Aerospace equipment is prone to various and complex quality problems during design, production, testing, and operation. Analyzing and judging the causes and mechanisms of quality problems is an important measure to avoid subsequent recurrence of problems. For aerospace equipment, the panel structure accounts for more than 60% of the dry weight of the equipment, mainly providing load-bearing and sealing functions. It is usually directly facing the harsh load conditions of complex service environments and is prone to various and complex quality problems.

[0003] The current judgment of quality problems in panel structures mainly relies on expert experience, often requiring a large number of investigations and tests, resulting in low efficiency in identifying quality problems and inability to effectively utilize historical data and knowledge. At the same time, the existing quality problem identification methods are mostly applied from the management level, and more focus on simply classifying the causes in terms of structure, unable to accurately characterize quality problems. Therefore, there is a lack of methods for classifying, characterizing, quantifying, and visualizing quality problems in panel structures, and it is impossible to systematically manage the process of identifying panel quality problems, resulting in inaccurate positioning of quality problems. Moreover, for the detection or reproduction of quality problems in panel structures, the test means generally rely on isolated discrete test points (such as strain sensors), but the large size and refinement of panel structures limit the spatial coverage of traditional test means, and it is difficult to achieve real-time monitoring of the full-field response; although simulation analysis can predict the full-field structural response, due to the difficulty of introducing test system deviations in real time, its prediction accuracy for the test process is low and it is impossible to accurately predict quality problems. Therefore, it is difficult to visually evaluate and judge quality problems with high precision by simply relying on traditional test monitoring methods or simulation analysis. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for identifying quality problems of panel structures based on digital twins, which can efficiently and accurately identify quality problems of panel structures.

[0005] To achieve the above purpose, the present invention provides the following solution:

[0006] A method for identifying quality problems of panel structures based on digital twins, the method comprising:

[0007] Obtaining the structural and material parameters of the target panel and the load conditions and boundary conditions applied to the target panel; the structural and material parameters include: material and dimensions;

[0008] Establish a simulation analysis model, perform mechanical calculations according to the dimensions, the load conditions, and the boundary conditions, and obtain the mechanical response of the simulation analysis; the simulation analysis model is constructed using the finite element method; the mechanical response of the simulation analysis includes: multiple coordinate positions and the mechanical responses corresponding to the multiple coordinate positions;

[0009] Obtain the mechanical response obtained by the test monitoring of the target panel; the mechanical response obtained by the test monitoring includes: the coordinates of discrete points and the responses corresponding to the discrete points; the discrete points are the measurement points of the sensors arranged on the target panel;

[0010] Establish a digital twin model, and obtain the mechanical response of the digital twin according to the mechanical response of the simulation analysis and the mechanical response obtained by the test monitoring; the digital twin model is constructed using a data fusion method; the mechanical response of the digital twin includes: multiple position points and the response values corresponding to the multiple position points; each position point includes one of the coordinate positions and one or more of the corresponding discrete points;

[0011] Calculate the response difference according to the mechanical response of the simulation analysis and the mechanical response of the digital twin; the response difference is the maximum deviation percentage or the correlation coefficient;

[0012] If the response difference is within the set difference range, it is determined that there is a structural quality problem with the target panel.

[0013] Optionally, the data fusion method includes: the scaling function method, the Kriging-like method, and the machine learning method.

[0014] Optionally, the calculation formula for the maximum deviation percentage is:

[0015]

[0016] Or, the calculation formula for the maximum deviation percentage is:

[0017]

[0018] The calculation formula for the correlation coefficient is:

[0019]

[0020] where, e max is the maximum deviation percentage; abs is the absolute value function; r 2 is the correlation coefficient; min(f fem ) is the minimum mechanical response; min(f dt ) is the minimum response value; max(f fem ) is the maximum mechanical response; max(f dt) is the maximum response value; is the mechanical response at the i-th coordinate position; is the average value of the mechanical responses at the i-th coordinate position; is the response value at the i-th position point in the digital twin model; is the average value of the response values at the i-th position point in the digital twin model; n is the total number of coordinate positions and position points, and i is the serial number.

[0021] Optionally, the method further includes:

[0022] Output and display the response difference and the corresponding coordinate position or position point of the response difference.

[0023] According to the specific embodiments provided by the present invention, the following technical effects are disclosed:

[0024] The present invention provides a method for identifying quality problems of panel structures based on digital twins. By using the digital twin model and the simulation analysis model constructed by the data fusion method, the mechanical responses of the simulation analysis and the mechanical responses of the digital twins are obtained respectively, and then the response difference is calculated. According to the response difference, it is judged whether there are structural quality problems in the current structure of the target panel; since the digital twin model is constructed by the data fusion method, the calculation accuracy of the mechanical response can be improved. And because the simulation analysis model calculates the full-field mechanical response for the target panel, and the mechanical response obtained by experimental monitoring is an accurate calculation of the mechanical response of the target panel, and then by combining the digital twin model and the simulation analysis model, it can quickly determine whether there are structural quality problems in the target panel; therefore, the present invention can efficiently and accurately identify the quality problems of the panel structure. Description of the Drawings

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0026] Figure 1 is the flowchart of the method for identifying quality problems of panel structures based on digital twins provided by the embodiments of the present invention;

[0027] Figure 2 is the specific flowchart of the method for identifying quality problems of panel structures based on digital twins provided by the embodiments of the present invention in practical applications;

[0028] Figure 3Schematic diagram of the load conditions at the boundary of the panel structure in the simulation analysis model provided by the embodiments of the present invention;

[0029] Figure 4 Schematic diagram of the calculation results of the finite element transverse strain in the simulation analysis model provided by the embodiments of the present invention;

[0030] Figure 5 Schematic diagram of the layout of the strain sensors provided by the embodiments of the present invention;

[0031] Figure 6 Schematic diagram of the calculation results of the digital twin model provided by the embodiments of the present invention. Detailed implementation manners

[0032] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0033] The purpose of the present invention is to provide a method for identifying quality problems of panel structures based on digital twins, which can efficiently and accurately identify quality problems of panel structures.

[0034] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0035] Embodiment 1

[0036] As Figure 1 shown, the embodiments of the present invention provide a method for identifying quality problems of panel structures based on digital twins, and the method includes:

[0037] Step 100: Obtain the structural and material parameters of the target panel and the load conditions and boundary conditions applied to the target panel; the structural and material parameters include: materials and dimensions. The load conditions are the restrictions on the external loads applied to the target panel, such as axial compression and axial tension applied to the target panel.

[0038] The boundary conditions are the variation laws between the mechanical responses and the coordinate variables obtained when performing mechanical calculations on the target panel.

[0039] Step 200: Establish a simulation analysis model, perform mechanical calculations according to the dimensions, load conditions, and boundary conditions, and obtain the mechanical responses of the simulation analysis; the simulation analysis model is constructed using the finite element method; the mechanical responses of the simulation analysis include: multiple coordinate positions and the mechanical responses corresponding to the multiple coordinate positions.

[0040] Step 300: Obtain the mechanical responses monitored in the target panel test; the mechanical responses monitored in the test include: the coordinates of discrete points and the responses corresponding to the discrete points; the discrete points are the measurement points of the sensors arranged on the target panel.

[0041] Step 400: Establish a digital twin model, and obtain the mechanical response of the digital twin based on the mechanical responses from simulation analysis and those monitored in the test; the digital twin model is constructed using a data fusion method; the mechanical response of the digital twin includes: multiple position points and the response values corresponding to the multiple position points; each position point includes a coordinate position and one or more corresponding discrete points.

[0042] Step 500: Calculate the response difference based on the mechanical responses from simulation analysis and the mechanical response of the digital twin; the response difference is the maximum deviation percentage or the correlation coefficient.

[0043] Step 600: If the response difference is within the set difference range, it is determined that there is a structural quality problem with the target panel.

[0044] The data fusion methods include the scaling function method, Kriging-based methods, and machine learning methods. Data fusion is based on a surrogate model for multi-source data fusion.

[0045] Specifically, the scaling function method can be divided into the additive scaling function method, the multiplicative scaling function method, and the hybrid scaling function method; among them, the hybrid scaling function method is more widely applied and has higher accuracy than the additive scaling function method and the multiplicative scaling function method. The hybrid scaling function method first constructs a low-fidelity surrogate model for the simulation data, i.e., the mechanical response from simulation analysis, based on the RBF model, and then constructs a scaling function surrogate model based on the difference between the test data and the simulation data (i.e., the mechanical response from simulation analysis and the mechanical response monitored in the test). Adding the low-fidelity surrogate model and the scaling function surrogate model can construct a data fusion model, i.e., the digital twin. This method can achieve the establishment of a model with high accuracy based on a small sample size.

[0046] Kriging-based methods mainly include the Co-Kriging method and the hierarchical Kriging method, etc.; machine learning methods include the transfer learning method.

[0047] The Co-Kriging method interpolates the high-precision sample points, i.e., the mechanical responses monitored in the test, while the hierarchical Kriging uses an efficient global optimization method for optimization; for the specific data fusion process, please refer to "Research Progress on Variable-Fidelity Approximation Models and Their Applications in the Optimal Design of Complex Equipment".

[0048] The data fusion method of the transfer learning method constructs a neural network, trains and optimizes the constructed neural network, and finally obtains the digital twin model.

[0049] Among them, the method for establishing a digital twin model using transfer learning is as follows:

[0050] Construct a neural network based on the mechanical responses obtained from simulation analysis; train the weights and thresholds of the neural network using the Adaptive Moment Estimation algorithm (Adam) to obtain a pre-trained model.

[0051] For the pre-trained model, the first several layers of the neural network (DNN) learn general features. As the network depth increases, the subsequent layers focus more on task-specific features, and the task-specific features become more advanced closer to the output layer. The good hierarchical structure enables the neural network (DNN) to be transferable. Therefore, transfer learning is performed on the pre-trained model.

[0052] Performing transfer learning on the pre-trained model specifically includes:

[0053] Fix the first several layers of the pre-trained model and adjust the learning rates of the thresholds and weights of the last layer of the neural network. Secondly, use the mechanical responses obtained from experimental monitoring as the target domain sample set to fine-tune the last layer of the network, that is, re-train the model parameters of the last layer of the neural network of the pre-trained model with the mechanical responses obtained from a small number of experimental monitoring and a small learning rate. After fine-tuning and optimization, the digital twin model can be obtained.

[0054] The construction of the digital twin model is essentially the fusion of two types of data (finite element simulation data and experimental measurement point data, that is, the mechanical responses obtained from simulation analysis and the mechanical responses obtained from the experimental monitoring), integrating the advantages of both (the finite element is for the whole field and the experimental measurement points are accurate), and finally constructing a high-precision, full-field digital twin. Since the digital twin is full-field, response prediction can be performed based on any input.

[0055] After the digital twin model is determined, input the structural and material parameters of the target wall panel and the load conditions and boundary conditions applied to the target wall panel into the digital twin model, and the mechanical response of the digital twin can be obtained.

[0056] Specifically, the calculation formula for the maximum deviation percentage is:

[0057]

[0058] Or, the calculation formula for the maximum deviation percentage is:

[0059]

[0060] The percentage of maximum deviation can be calculated by selecting the maximum or minimum value (max / min) according to different actual problems (such as axial compression, axial tension, etc.) and different calculated mechanical responses (axial strain, transverse strain, Mises stress, etc.).

[0061] The calculation formula for the correlation coefficient is as follows:

[0062]

[0063] Where, e max is the percentage of maximum deviation; abs is the absolute value function; r 2 is the correlation coefficient; min(f fem ) is the minimum mechanical response; min(f dt ) is the minimum response value; max(f fem ) is the maximum mechanical response; max(f dt ) is the maximum response value; is the mechanical response at the i-th coordinate position; is the average value of the mechanical responses at the i-th coordinate position; is the response value at the i-th position point in the digital twin model; is the average value of the response values at the i-th position point in the digital twin model; n is the total number of coordinate positions and position points, and i is the serial number.

[0064] As an alternative implementation, the method further includes:

[0065] Output and display the response difference and the corresponding coordinate position or position point of the response difference.

[0066] For the method for identifying quality problems of panel structures based on digital twins provided in this embodiment, the steps in actual application can also be as follows:

[0067] When performing visual identification of quality problems of panel structures based on digital twins, first establish a simulation analysis model of the panel structure and perform calculations to obtain the mechanical responses of the panel structure as the simulation analysis data set; then deploy sensors, conduct relevant mechanical tests, and obtain the test measurement point positions and sensor response information of the structure as the test data set; furthermore, construct a high-precision full-field digital twin of the panel structure according to the simulation analysis data set and the test data set; then calculate the evaluation index value of the quality problem through the digital twin results and the calculation results of the simulation analysis model; finally, compare the evaluation index with the reference standard range to determine whether the current structure has quality problems and the specific types of quality problems, and visually output the corresponding information. Figure 2 It is a specific flowchart in actual application.

[0068] First step, calculate the response of the simulation analysis model. According to the actual panel structure, establish its simulation analysis model, conduct mechanical response analysis, and use the obtained structural coordinates and corresponding mechanical response calculation results as the simulation analysis data set. The mechanical response includes stress response, strain response, displacement response, etc. The panel structure refers to, for the actual structural form (specific structural materials, dimensions, etc.), constructing a simulation analysis model by using the finite element method, or using finite element analysis software (such as Ansys, Abaqus, etc.) for structural modeling and finite element mesh generation to obtain a finite element model, that is, the simulation analysis model; then conduct mechanical analysis according to the actual working conditions (axial compression, internal pressure, etc.), and output the mechanical response of the finite element calculation results in the form of coordinate-response.

[0069] When analyzing the response of the simulation analysis model, the mechanical response analysis of the panel structure simulation can be carried out according to the actual situation, including but not limited to stress response, strain response, displacement response, etc.

[0070] Second step, deploy sensors and conduct mechanical tests. According to the simulation analysis results, deploy sensor measurement points on the panel structure and conduct mechanical tests, and use the obtained discrete measurement point coordinates and the mechanical responses measured in the tests as the test data set. The deployment methods of the test measurement points include uniform spatial deployment, adaptive deployment based on the simulation analysis results, etc. The mechanical responses measured in the tests include stress response, strain response, displacement response, etc. The load types of the mechanical tests include axial compression load, axial tension load, concentrated force load, water pressure load, airtight load, etc.

[0071] Specifically, when deploying the sensor positions, the sensor deployment method can be selected according to the actual situation, including but not limited to uniform spatial deployment, adaptive deployment based on the simulation analysis results, etc. When conducting the structural mechanical test, the test measurement of the mechanical response can be carried out according to the actual needs, including but not limited to stress response, strain response, displacement response, etc. When applying the load in the mechanical test, the load application method can be selected according to the characteristics of the structure and the service conditions, including but not limited to axial compression load, axial tension load, concentrated force load, internal pressure load, etc.

[0072] Third step, construct the full-field digital twin, that is, construct the digital twin model. Perform data fusion on the simulation analysis response data set obtained in the first step and the test measurement point data set obtained in the second step to construct a digital twin full-field visualization model. The data fusion methods include scale function method, Co-Kriging method, hierarchical Kriging method, transfer learning method, etc.

[0073] When constructing a full-field digital twin, different data fusion methods can be selected according to the actual problem scale (such as the amount of simulation data of the panel structure and the number of test measurement points), including but not limited to the scaling function method, Co-Kriging method, hierarchical Kriging method, transfer learning method, etc. Among them, when the amount of data is small (less than 1000 simulation data and less than 50 test measurement points), it is recommended to use the Co-Kriging method or the hierarchical Kriging method; when the amount of data is medium (less than 50000 simulation data and less than 100 test measurement points), it is recommended to use the scaling function method based on the RBF surrogate model; when the amount of data is large (more than 50000 simulation data and more than 100 test measurement points), it is recommended to use the transfer learning method based on the deep neural network.

[0074] Step 4: Calculate the quality problem evaluation index. The deviation between the simulation analysis model and the digital twin is used as the evaluation index, such as the maximum deviation percentage e max and the correlation coefficient (Pearson) r 2 etc.

[0075] In calculating the quality problem evaluation index, appropriate evaluation indexes can be selected for the analysis and judgment of quality problems, including but not limited to the maximum deviation percentage e max and the correlation coefficient (Pearson) r 2 etc. In addition, as long as any calculation result is within the set recommended range, it is judged that there is a quality problem.

[0076] The typical quality problems considered in the present invention for the panel structure include but are not limited to geometric defects, eccentric load application, thickness idealization deviation, weld model idealization deviation, and end frame model idealization deviation, and their specific detailed explanations are as follows:

[0077] 1. Geometric defects: The simulation analysis model of the panel structure is often simplified based on the 3D CAD model, and it is difficult to consider the geometric defects generated in the panel specimens during processing, manufacturing, transportation, and storage, which easily leads to differences between the simulation strength analysis results and the real test results.

[0078] 2. Eccentric load application: During the test, due to loading application errors or improper operations, the load is applied eccentrically, which in turn affects the judgment of the structural performance, and the situation where it is considered that the test has passed the check (actually it may not have passed) may occur, resulting in the failure of the model task.

[0079] 3. Thickness idealization deviation: In the panel structure, there are complex variable thickness regions. However, for the sake of simplifying calculations, shell elements are usually used for simulation modeling in engineering, and it is difficult for shell elements to reflect the real structural thickness changes, resulting in deviations in stiffness simulation, and further leading to significant differences between the simulation results and the real test results.

[0080] 4. Idealization deviation of the weld model: The welding quality between components is low, such as typical problems like undercut, overlap, porosity, surface cracks, internal cracks, and non-compliance with the specified weld position or size, which affect the actual strength and stiffness of the structure. However, in the actual analysis model, simulations are still carried out according to the idealized model, resulting in deviations.

[0081] 5. Idealization deviation of the end frame model: For complex panel structures, in order to simplify calculations, it is necessary to simplify the modeling of the end frame model and simulate the solid model using beam and shell elements. However, the simplified model cannot accurately simulate details such as local variable thickness and grooves, leading to differences between the local simulation results and the actual test results.

[0082] For a panel structure, the above five quality problems generally do not occur simultaneously, and the areas of concern are different. Geometric defects are deviation indicators calculated from the responses of the full-field finite element model and the full-field digital twin; load application eccentricity is calculated by dividing the digital twin into left and right parts of the data; thickness idealization deviation, weld model idealization deviation, end frame model idealization deviation, etc. are calculated by comparing the responses of the local variable thickness, weld, and end frame finite element models with those of the digital twin. (Generally, not all structures have variable thickness, welds, and end frame areas. There may be one of them or none, and even if all of the above situations exist simultaneously, the indicators are calculated for their local areas).

[0083] Step 5: Visualize and output quality problem information. Compare the evaluation indicators calculated in the fourth step with the reference standard range. If the calculated results of the evaluation indicators are within the set recommended range, it is considered that the corresponding quality problem has occurred in the structure, and the corresponding quality problem information is visually output.

[0084] In specific applications, establish a simulation analysis model of the panel structure, fix the lower end of the structure, apply an axial compressive load (axial compression of 0.3 mm) to the upper end, conduct a mechanical response analysis, and use the obtained structural coordinates and the corresponding transverse strain response calculation results as the simulation analysis data set. The calculation results are as Figure 3 and Figure 4 shown. Since the simulation analysis calculation results do not consider the influence of loading eccentricity, the strain nephogram is an idealized nephogram that is symmetric left and right.

[0085] Based on the spatially uniform distribution method, sensors are uniformly arranged on the panel structure, as Figure 5 shown. The total number of test sensor measurement points is 24. During the test loading, the deviation from the center point is 10 mm, and the axial compression is 0.3 mm. The obtained 24 measurement point coordinates and the strain responses measured in the test are used as the test data set. Figure 5 The black dots in

[0086] Predict the structural full-field strain response through the digital twin model. The full-field prediction results are as follows Figure 6 As shown. It can be seen from the digital twin nephogram that due to the eccentricity of the structural load, the left and right sides of the digital twin nephogram are not symmetrical, and there are obvious differences in the maximum strains of the structures on the left and right sides along the central axis.

[0087] Divide the full-field prediction results of the digital twin into two parts of data on the left and right along the central axis for calculating e max , and the calculation results are shown in Table 1. The current maximum deviation calculation result is 29%.

[0088] Table 1 Example of the discrimination of the eccentricity mass problem of the load application

[0089] Suggested range Current calculation result Is there any quality problem <![CDATA[e max > >15％ 29％ Yes

[0090] Compare the maximum deviation calculation result with the recommended range, and it is found that the calculated value is within the recommended range of the quality problem. Therefore, it can be judged that there is an eccentricity mass problem in the current load application, and then the quality problem information is output and displayed to achieve the purpose of visualization.

[0091] Advantages of the present invention:

[0092] 1. Aiming at the problem that it is difficult to accurately judge the quality problem by relying solely on the test monitoring method or the simulation method, the present invention establishes a digital twin model, makes full use of the advantages of test data and simulation analysis data, and establishes a high-precision full-field visualization model of mechanical response, which can effectively improve the recognition accuracy of quality problems.

[0093] 2. Aiming at the problem that the traditional method cannot accurately characterize the quality problem of the panel structure, the present invention establishes specific quantitative evaluation indexes (including but not limited to the maximum deviation percentage e max and the Pearson correlation coefficient r 2 etc.), and details the typical quality problems (including but not limited to geometric defects, load application eccentricity, thickness idealization deviation, weld model idealization deviation, end frame model idealization deviation), and determines the reference standard range, which can accurately locate and identify the type and form of quality problems.

[0094] 3. The present invention can visually identify the quality problems of the panel structure for various mechanical responses, and has certain applicability, including but not limited to stress response, strain response or displacement response, etc. When arranging the sensor positions, the sensor arrangement method can be selected according to the actual situation, including but not limited to uniform spatial arrangement, adaptive arrangement based on the simulation analysis results, etc. When performing the mechanical test loading, the load can be applied according to the actual load conditions faced by the structure, and the load application types include but not limited to axial compression load, axial tension load, concentrated force load, water pressure load, airtight load, etc.

[0095] 4. The present invention can select different data fusion methods to construct the digital twin according to the actual problem scale (such as the simulation data volume of the panel structure and the number of test measurement points), including but not limited to the scaling function method, the Co-Kriging method, the hierarchical Kriging method, the transfer learning method, etc. By selecting the appropriate data fusion method for different problems, the construction accuracy of the digital twin can be effectively improved.

[0096] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other.

[0097] Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only for helping to understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for identifying quality problems of panel structures based on digital twins, characterized in that, The method includes: Obtaining the structural and material parameters of the target wall panel and the load conditions and boundary conditions applied to the target wall panel; the structural and material parameters include: material and dimensions; Establishing a simulation analysis model, performing mechanical calculations according to the dimensions, the load conditions and the boundary conditions, and obtaining the mechanical response of the simulation analysis; the simulation analysis model is constructed by the finite element method; the mechanical response of the simulation analysis includes: multiple coordinate positions and the mechanical responses corresponding to the multiple coordinate positions; Obtaining the mechanical response obtained by test monitoring of the target wall panel; the mechanical response obtained by test monitoring includes: the coordinates of discrete points and the responses corresponding to the discrete points; the discrete points are the measuring points of sensors arranged on the target wall panel; Establishing a digital twin model, and obtaining the mechanical response of the digital twin according to the mechanical response of the simulation analysis and the mechanical response obtained by test monitoring; the digital twin model is constructed by the method of data fusion; the mechanical response of the digital twin includes: multiple position points and the response values corresponding to the multiple position points; each of the position points includes one of the coordinate positions and one or more of the corresponding discrete points; Calculating a response difference according to the mechanical response of the simulation analysis and the mechanical response of the digital twin; the response difference is the maximum deviation percentage or the correlation coefficient; If the response difference is within the set difference range, it is determined that there is a structural quality problem with the target wall panel.

2. The method for identifying quality problems of the panel structure based on digital twin according to claim 1, wherein The method of data fusion includes: the scaling function method, the Kriging-like method and the machine learning method.

3. The method for identifying quality problems of the panel structure based on digital twin according to claim 1, wherein The calculation formula for the maximum deviation percentage is: Or, the calculation formula for the maximum deviation percentage is: The calculation formula for the correlation coefficient is: where, e max is the maximum deviation percentage; abs is the absolute value function; r 2 is the correlation coefficient; min(f fem ) is the minimum mechanical response; min(f dt ) is the minimum response value; max(f fem ) is the maximum mechanical response; max(f dt ) is the maximum response value; is the mechanical response at the i-th coordinate position; is the average value of the mechanical responses at the i-th coordinate position; is the response value at the i-th position point in the digital twin model; is the average value of the response values at the i-th position point in the digital twin model; n is the total number of coordinate positions and position points, and i is the serial number.

4. The method for identifying quality problems of the panel structure based on digital twin according to claim 1, wherein The method further includes: Outputting and displaying the response difference and the coordinate position or position point corresponding to the response difference.

Images