A method and system for predicting stress damage evolution in forced clamping of thin-walled aviation composite structures

Abstract

The present invention provides a stress damage evolution prediction method and system in the forced clamping of aviation composite thin-walled structures, which belongs to the field of mechanical analysis and damage prediction of aviation composite structures, including: S1: establishing an analytical solution model of stress distribution and specific damage morphology of composite thin-walled structures; S2: arranging data acquisition devices to collect stress damage data in the forced clamping process; S3: processing and fusion analysis of test data to extract assembly stress damage data features; S4: establishing and optimizing an assembly positioning stress damage finite element simulation model based on mechanism and measured data; S5: training assembly prediction data to evaluate stress distribution and damage possibility under different clamping conditions; S6: obtaining stress distribution and damage area prediction results based on process monitoring data, and alarming when exceeding the dangerous threshold. The present invention can improve the consistency between numbers and reality, identify potential assembly stress concentration and damage risk areas before assembly, and guide the setting of clamping processes.

Description

Technical Field

[0001] The present invention relates to the field of mechanical analysis and damage prediction of aviation composite structures, and in particular to a method and system for predicting stress damage evolution in forced clamping of aviation composite thin-wall structures. Background Art

[0002] Aerospace composite structures are widely used in the aerospace field due to their high strength characteristics. However, in the forced clamping of weak rigid thin-walled workpieces, the applied external force easily leads to uneven distribution of internal stress field, which not only induces structural deformation at the microscopic level, but also promotes the initiation and accelerated expansion of microcracks inside the composite thin-walled structure once the bearing limit of the material itself is exceeded as the assembly process proceeds. In view of this, the stress damage evolution prediction technology in the assembly process is particularly critical. Ignoring this link, especially failing to accurately locate the stress concentration area, will lead to improper adjustment of the size of the positioning clamping force and unreasonable selection of the fixture position, which will directly lead to increased stress accumulation and accelerated damage spread, and finally cause the composite thin-walled structure to break, seriously threatening safety performance. In fact, the assembly of weak rigid composite structures is a dynamic process in which the internal stress distribution is constantly changing, the damage forms are associated and accumulated, the material properties are constantly degraded, and the local structure gradually fails. The causes of internal stress and structural damage in the assembly of composite thin-walled structures are complex. Excessive or uneven distribution of internal stress and structural damage not only affects the geometric accuracy of the assembly process and the strength of the overall assembly structure, but also affects the trend of stress / damage changes, and may even cause damage such as crushing and fracture. The internal stress state and damage extension path of the structure during assembly not only fundamentally determine the bearing capacity and assembly accuracy of aviation composite thin-walled structures, but are also the core reference indicators for guiding future design iteration optimization and refined adjustment of assembly processes.

[0003] Traditional assembly stress and damage evolution prediction methods have the following disadvantages: ① They mostly rely on simple finite element simulation analysis or theoretical modeling calculations, which are poorly integrated with the actual assembly process, and cannot effectively combine mechanism modeling data, finite element simulation data, and field test data. The prediction effect of assembly stress and damage evolution in the forced clamping process is also poor; ② It is impossible to effectively realize the prior evaluation of assembly stress damage data corresponding to the forced positioning clamping process. The improvement analysis of the assembly process mostly relies on a large number of tests, which is costly and inefficient; ③ The processing of test data is relatively rudimentary, lacks effective data filtering and denoising technology and signal analysis tools, and it is difficult to accurately extract significant damage features from massive data. The accuracy and timeliness of diagnosis need to be improved; ④ Traditional methods can only take countermeasures after assembly damage occurs, which has a lag and lacks a real-time monitoring and early warning system for assembly quality.

[0004] Therefore, in order to predict and control the formation of stress / damage of composite thin-walled workpieces under forced clamping operations, as well as the evolution process generated by the assembly operation, how to fully combine the mechanism modeling data, finite element simulation data, and on-site measured data to establish an assembly finite element simulation model and parameter correction optimization mechanism, and thereby establish a data-driven assembly stress / damage distribution prediction model, is the key to solving the problem of rapid and accurate generation of assembly quality data, and prior prediction, evaluation, and control of stress and damage.

[0005] Therefore, it is necessary to study a stress damage evolution prediction method and system in forced clamping of aviation composite thin-walled structures to address the shortcomings of the existing technology and to solve or alleviate one or more of the above problems. Summary of the invention

[0006] In view of this, the present invention provides a method and system for predicting the evolution of stress damage in the forced clamping of aviation composite thin-walled structures. Through the organic integration of mechanism modeling data, measured data, and finite element simulation data and data-driven empowerment means, the stress damage evolution is predicted and analyzed. Potential assembly stress concentration and damage risk areas can be identified before assembly, guiding the setting of a reasonable clamping process plan to improve production efficiency and product quality.

[0007] In one aspect, the present invention provides a method for predicting the evolution of stress damage in forced clamping of an aerospace composite thin-walled structure, the method comprising:

[0008] S1: The theoretical analytical solution of stress distribution and specific damage morphology of composite thin-walled structure assembly is obtained by combining multi-assembly physical field collaborative modeling analysis with macro-micro multi-scale modeling analysis, which is used as mechanism data;

[0009] S2: Arrange data acquisition devices in the key characteristic areas of the composite thin-walled structure to collect measured data on stress and damage during forced clamping;

[0010] S3: Process and integrate the measured data of test stress and damage to extract the characteristics of assembly stress and damage data;

[0011] S4: Establish an assembly positioning stress and damage finite element simulation model based on mechanism data and measured data, obtain finite element simulation data, compare finite element simulation data, measured data and mechanism data, analyze the source of deviation according to the comparison results, and modify and optimize the finite element simulation model and parameters, obtain an optimized assembly positioning stress and damage finite element simulation model, and generate large sample data of assembly quality;

[0012] S5: Use support vector regression to train large sample data of assembly quality to evaluate stress distribution and damage under different clamping conditions;

[0013] S6: Monitor the assembly process of composite weak-rigid thin-walled structures in real time, input process data into the optimized assembly positioning stress and damage finite element simulation model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

[0014] According to the above aspects and any possible implementation, an implementation is further provided, wherein S1 specifically includes:

[0015] S11: Coupling macro and micro scales, a multi-scale stress bridging matrix is ​​established based on the continuum theory. When the material component in a single cell is damaged, its material reduced stiffness is calculated, and the reduced equivalent stiffness of the corresponding micro unit is obtained through the macro-micro stiffness equivalent matrix;

[0016] S12: Considering the influence of factors such as the composite material layup form, structural geometry, clamping constraints and applied boundary conditions, forced clamping load and assembly deformation, a forced assembly macro-micro stress quantification calculation model is established to obtain the assembly internal stress field size and deformation state at the fiber, matrix and interface;

[0017] S13: Integrating the internal stress factors of the clamping, a three-dimensional progressive damage initiation model of composite laminates under forced loading is established to form the damage initiation criterion and obtain the analytical solution of the specific damage morphology and distribution.

[0018] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S2 specifically includes:

[0019] S21: Arrange strain sensors in the expected stress concentration area of ​​the composite panel;

[0020] S22: By combining terahertz, DIC and ultrasonic measurement technologies, the three work together to dynamically monitor and accurately evaluate the stress and damage state of composite thin-walled structures during forced assembly in a non-destructive manner;

[0021] S23: The sensor network is optimally arranged according to the assembly characteristics of the composite structure. Through the comprehensive layout of sensors, a three-dimensional monitoring network is formed to ensure that the key characteristic areas of the composite thin-walled structure are under monitoring. During the forced clamping process of the composite thin-walled structure, the arranged sensor system will collect assembly stress and damage data in real time and synchronously.

[0022] According to the above aspects and any possible implementation, an implementation is further provided, wherein S3 specifically includes:

[0023] S31: pre-processing the internal stress, strain and damage data of the composite assembly, using low-pass filtering, median filtering or Kalman filtering digital signal processing technology to remove irrelevant signals obtained by the sensor measurement, and interpolating missing or inconsistent data points to improve data fusion quality;

[0024] S32: extracting key feature vectors reflecting the evolution of assembly stress and damage from the pre-processed assembly stress and damage data through signal analysis;

[0025] S33: Standardize or normalize the extracted features to ensure that the fused data can be compared at the same level and improve the data processing efficiency of subsequent machine learning models.

[0026] As described above and any possible implementation method, an implementation method is further provided, in which the process of refining damage characteristics through signal analysis in S3 is specifically: calculating the rate of change of strain values ​​at adjacent time points to identify areas of stress concentration and damage initiation; identifying abnormally high-value areas in stress distribution through statistical methods; converting time domain signals to frequency domain through Fourier transform and other means to analyze the relationship between specific frequency components and stress damage; applying signal processing and pattern recognition algorithms to extract key feature vectors reflecting damage evolution from the data.

[0027] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S4 specifically includes:

[0028] S41: According to the actual geometric dimensions, material properties and boundary conditions of the composite thin-walled structural parts, as well as real constraints such as forced clamping parameters, the contact conditions, friction coefficient and prestress state are set, and the finite element analysis software ABAQUS is used to build a three-dimensional simulation model;

[0029] S42: Using the ABAQUS cohesive bilinear constitutive model and the cohesive bonding unit, a damage simulation model of the composite assembly structure with initial damage defects is established;

[0030] S43: When there are initial delamination defects in the parts, if some weak damage characteristics cannot be monitored, in order to analyze the expansion of defects and the occurrence of new damage during the forced clamping process, the mechanism model in S1 is used. On the basis of the established simulation model with initial defects, the three-dimensional Hashin damage criterion and the constitutive model of composite material performance damage degradation under the current forced clamping process are introduced;

[0031] S44: Analyze the stress response, damage evolution and expansion, and dimensional changes of composite assembly structures under different clamping parameters to obtain the influence of forced clamping operation on assembly quality;

[0032] S45: Compare the simulation data such as stress-strain response, displacement field distribution, and damage sign distribution obtained by finite element operation with the measured data and the data calculated by the mechanism model, identify the source of the deviation between the model and the actual situation based on the comparison results, and continuously adjust the simulation model parameters until the prediction results are consistent with the test results;

[0033] S46: Analyze the assembly stress and damage analysis results under different assembly conditions to accurately generate assembly quality big data samples in forced clamping.

[0034] According to the aspects and any possible implementation methods described above, an implementation method is further provided, in which the process of constructing the constitutive model of damage degradation of composite material performance in S4 is specifically as follows: based on the established multi-scale mechanism model, a damage criterion of the composite structure and a continuous degradation constitutive relationship of stiffness are introduced, and the traction-separation law is used to design a finite element software UMAT subroutine for secondary development, and a bilinear degradation constitutive model of the matrix and fiber interface and a material damage constitutive model are established.

[0035] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S5 specifically includes:

[0036] S51: Standardize and normalize the large sample data obtained through finite element simulation analysis in stage S4. At the same time, extract the key characteristic data of assembly stress and damage evolution based on the analysis results of stage S3;

[0037] S52: Using 70% of the assembly internal stress and damage distribution simulation prediction data as a training set and the remaining 30% as a test set, a support vector regression model is constructed to obtain the optimal hyperplane and the error between the predicted assembly stress damage value and the actual value;

[0038] S53: Based on cross-validation, the penalty coefficient and kernel function parameters are adjusted, and multiple iterations are performed until a set of parameters is found that minimizes the prediction error of the support vector regression model on the validation set;

[0039] S54: The SVR model is trained using the optimized parameters. The performance of the model is evaluated by calculating the mean square error, root mean square error, and determination coefficient between the predicted value and the true value. After the model training is completed, the assembly stress distribution and the possibility of damage under different forced clamping conditions can be quickly and accurately evaluated.

[0040] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S6 specifically includes:

[0041] S61: Real-time monitoring of the assembly process of composite weak-rigid thin-walled structures, collecting stress and strain data during the assembly process from a sensor network placed at key locations;

[0042] S62: Use the trained support vector regression model to process the received process data, and predict the assembly stress distribution inside the structure and the possible damage area in the short term in the future based on the damage evolution law learned through training;

[0043] S63: Compare the stress distribution and damage area predicted by the model with the pre-set safety threshold to distinguish between normal working conditions and potential assembly damage risks;

[0044] S64: When the model prediction results show that the stress level or damage possibility of a clamping area exceeds the preset threshold, an early warning is automatically triggered, and a new round of iterative feedback analysis of the forced clamping process is required.

[0045] According to the above aspects and any possible implementation, a stress damage evolution prediction system in forced clamping of an aviation composite thin-walled structure is further provided, and the stress damage evolution prediction system in forced clamping of an aviation composite thin-walled structure comprises:

[0046] The mechanism modeling and calculation module is used to obtain the theoretical analytical solution of the stress distribution and specific damage morphology of the composite thin-walled structure assembly by combining the collaborative modeling and analysis of multiple assembly physical fields with the macro-micro multi-scale modeling and analysis, and use it as the mechanism data;

[0047] The data acquisition module is used to arrange data acquisition devices in the key characteristic areas of the composite thin-wall structure to collect the measured data of stress and damage during the forced clamping process;

[0048] Data feature extraction module, used to process and integrate the measured data of test stress and damage, and extract assembly stress and damage data features;

[0049] The simulation model optimization module is used to establish a finite element simulation model of assembly positioning stress and damage based on mechanism data and measured data, obtain finite element simulation data, compare finite element simulation data, measured data and mechanism data, analyze the source of deviation according to the comparison results, and correct and optimize the finite element simulation model and parameters, obtain an optimized finite element simulation model of assembly positioning stress and damage, and generate large sample data of assembly quality;

[0050] Prediction model building module, which is used to train large sample data of assembly quality using support vector regression to evaluate stress distribution and damage under different clamping conditions;

[0051] The threshold alarm module is used to monitor the assembly process of composite weak-rigidity thin-walled structures in real time, input the process data into the optimized assembly positioning stress and damage finite element simulation model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

[0052] Compared with the prior art, the present invention can achieve the following technical effects:

[0053] 1) Establish a mechanism model for the internal stress and damage of the composite thin-walled structure during the forced clamping process to provide more explainability for the improvement analysis of the clamping process.

[0054] 2) Precision sensors are placed at key locations of composite materials to collect stress, strain and damage data in real time and comprehensively during forced positioning and clamping, forming a high-density monitoring network.

[0055] 3) Use signal processing technology to filter out noise and extract key features to provide a clear and reliable data basis for subsequent analysis.

[0056] 4) Organically integrate finite element simulation data with measured data and data calculated by mechanism model. By adjusting the simulation model parameters until the predicted results are highly consistent with the experimental results, the problem of high-quality and large sample data required for data-driven can be accurately solved.

[0057] 5) By applying data training-driven technology and adjusting the consistency between numbers and reality, the stress distribution and potential damage areas in the assembly can be quickly predicted in advance, reducing the number of actual assembly tests and measurements, and significantly improving assembly production efficiency.

[0058] 6) Real-time monitoring and input of assembly process data into the prediction model. Once any damage risk exceeding the safety threshold is predicted, the system immediately issues an early warning and quickly conducts a new round of iterative feedback analysis on the forced clamping process.

[0059] Of course, any product implementing the present invention does not necessarily need to achieve all of the above-mentioned technical effects at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 This is a process diagram of a stress damage evolution prediction technology in forced clamping of an aviation composite thin-wall structure of the present invention;

[0062] Figure 2 is a flowchart of an assembly stress and damage prediction technology provided by an embodiment of the present invention;

[0063] Figure 3 It is an exploded view of the overall structure of a certain type of composite wing box component provided by an embodiment of the present invention;

[0064] Figure 4 This is a description of the structure and adjustable functions of a typical box segment assembly tool of the present invention;

[0065] Figure 5 It is a stress and damage visualization cloud map obtained by using ABAQUS finite element simulation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0067] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0069] The present invention provides a method for predicting the evolution of stress damage in forced clamping of an aviation composite thin-walled structure. The method comprises:

[0070] S1: The theoretical analytical solution of stress distribution and specific damage morphology of composite thin-walled structure assembly is obtained by combining multi-assembly physical field collaborative modeling analysis with macro-micro multi-scale modeling analysis, which is used as mechanism data;

[0071] S2: Arrange data acquisition devices in the key characteristic areas of the composite thin-walled structure to collect measured data on stress and damage during forced clamping;

[0072] S3: Process and integrate the measured data of test stress and damage to extract the characteristics of assembly stress and damage data;

[0073] S4: Establish an assembly positioning stress and damage finite element simulation model based on mechanism data and measured data, obtain finite element simulation data, compare finite element simulation data, measured data and mechanism data, analyze the source of deviation according to the comparison results, and modify and optimize the finite element simulation model and parameters, obtain an optimized assembly positioning stress and damage finite element simulation model, and generate large sample data of assembly quality;

[0074] S5: Use support vector regression to train large sample data of assembly quality to evaluate stress distribution and damage under different clamping conditions;

[0075] S6: Monitor the assembly process of composite weak-rigid thin-walled structures in real time, input process data into the optimized assembly positioning stress and damage finite element simulation model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

[0076] The S1 specifically includes:

[0077] S11: Coupling macro and micro scales, a multi-scale stress bridging matrix is ​​established based on the continuum theory. When the material component in a single cell is damaged, its material reduced stiffness is calculated, and the reduced equivalent stiffness of the corresponding micro unit is obtained through the macro-micro stiffness equivalent matrix;

[0078] S12: Considering the influence of factors such as the composite material layup form, structural geometry, clamping constraints and applied boundary conditions, forced clamping load and assembly deformation, a forced assembly macro-micro stress quantification calculation model is established to obtain the assembly internal stress field size and deformation state at the fiber, matrix and interface;

[0079] S13: Integrating the internal stress factors of the clamping, a three-dimensional progressive damage initiation model of composite laminates under forced loading is established to form the damage initiation criterion and obtain the analytical solution of the specific damage morphology and distribution.

[0080] The S2 specifically includes:

[0081] S21: Arrange strain sensors in the expected stress concentration areas of the composite panel, such as joints, turning points or around openings where high stress or large strain may occur;

[0082] S22: By combining terahertz, DIC and ultrasonic measurement technologies, the three work together to dynamically monitor and accurately evaluate the stress and damage state of composite thin-walled structures during forced assembly in a non-destructive manner;

[0083] S23: The sensor network is optimally arranged according to the assembly characteristics of the composite structure. Through the comprehensive layout of the above sensors, a three-dimensional monitoring network is formed to ensure that the key characteristic areas of the composite thin-walled structure are under monitoring. During the forced clamping process of the composite thin-walled structure, the arranged sensor system will collect assembly stress and damage data in real time and synchronously.

[0084] The S3 specifically includes:

[0085] S31: pre-processing the obtained data of internal stress, strain, damage, etc. of the composite assembly, using digital signal processing technology such as low-pass filtering, median filtering or Kalman filtering to remove irrelevant signals obtained by the sensor measurement, and interpolating missing or inconsistent data points to improve data fusion quality;

[0086] S32: extracting key feature vectors reflecting the evolution of assembly stress and damage from the pre-processed assembly stress and damage data through signal analysis;

[0087] S33: Standardize or normalize the extracted features to ensure that the fused data are comparable at the same level and improve the data processing efficiency of subsequent machine learning models.

[0088] The process of extracting damage characteristics through signal analysis in S32 is specifically as follows: calculating the rate of change of strain values ​​at adjacent time points to identify areas of stress concentration and damage initiation; identifying abnormally high-value areas in stress distribution through statistical methods; converting time domain signals to frequency domain through Fourier transform and other means to analyze the relationship between specific frequency components and stress damage; applying signal processing and pattern recognition algorithms to extract key feature vectors reflecting damage evolution from the data.

[0089] The S4 specifically includes:

[0090] S41: According to the actual geometric dimensions, material properties and boundary conditions of the composite thin-walled structural parts, as well as real constraints such as forced clamping parameters, the contact conditions, friction coefficient and prestress state are set, and the finite element analysis software ABAQUS is used to build a three-dimensional simulation model;

[0091] S42: Using the ABAQUS cohesive bilinear constitutive model and the cohesive bonding unit, a damage simulation model of the composite assembly structure with initial damage defects is established;

[0092] S43: When there are initial delamination defects in the parts, if some weak damage characteristics cannot be monitored, in order to analyze the expansion of defects and the occurrence of new damage during the forced clamping process, the mechanism model in S1 is used. On the basis of the established simulation model with initial defects, the three-dimensional Hashin damage criterion and the constitutive model of composite material performance damage degradation under the current forced clamping process are introduced;

[0093] S44: Analyze the stress response, damage evolution and expansion, and dimensional changes of composite assembly structures under different clamping parameters to obtain the influence of forced clamping operation on assembly quality;

[0094] S45: Compare the simulation data such as stress-strain response, displacement field distribution, and damage sign distribution obtained by finite element operation with the measured data and the data calculated by the mechanism model. Based on the comparison results, identify the source of the deviation between the model and the actual situation, and continuously adjust the simulation model parameters until the prediction results are highly consistent with the test results;

[0095] S46: Analyze the assembly stress and damage analysis results under different assembly conditions to accurately generate assembly quality big data samples in forced clamping.

[0096] The process of constructing the constitutive model of damage degradation of composite material performance in S43 is specifically as follows: based on the established multi-scale mechanism model, the composite structure damage criterion and the stiffness continuous degradation constitutive relationship are introduced, and the traction-separation law is used to design the finite element software UMAT subroutine for secondary development, and establish a bilinear degradation constitutive model of the matrix and fiber interface and a material damage constitutive model.

[0097] The S5 specifically includes:

[0098] S51: Standardize and normalize the large sample data obtained through finite element simulation analysis in stage S4. At the same time, extract the key characteristic data of assembly stress and damage evolution based on the analysis results of stage S3;

[0099] S52: Using 70% of the assembly internal stress and damage distribution simulation prediction data as a training set and the remaining 30% as a test set, a support vector regression model is constructed to obtain the optimal hyperplane and the error between the predicted assembly stress damage value and the actual value;

[0100] S53: Based on cross-validation, the penalty coefficient and kernel function parameters are adjusted, and multiple iterations are performed until a set of parameters is found that minimizes the prediction error of the support vector regression model on the validation set;

[0101] S54: The SVR model is trained using the optimized parameters. The performance of the model is evaluated by calculating the mean square error, root mean square error, and determination coefficient between the predicted value and the true value. After the model training is completed, the assembly stress distribution and the possibility of damage under different forced clamping conditions can be quickly and accurately evaluated.

[0102] The S6 specifically includes:

[0103] S61: Real-time monitoring of the assembly process of composite weak-rigid thin-walled structures, collecting stress and strain data during the assembly process from a sensor network placed at key locations;

[0104] S62: Use the trained support vector regression model to process the received process data, and predict the assembly stress distribution inside the structure and the possible damage area in the short term in the future based on the damage evolution law learned through training;

[0105] S63: Compare the stress distribution and damage area predicted by the model with the pre-set safety threshold to distinguish between normal working conditions and potential assembly damage risks;

[0106] S64: When the model prediction results show that the stress level or damage possibility of a clamping area exceeds the preset threshold, an early warning is automatically triggered, and a new round of iterative feedback analysis of the forced clamping process is required.

[0107] The present invention also provides a stress damage evolution prediction system in forced clamping of an aviation composite thin-walled structure, the stress damage evolution prediction system in forced clamping of an aviation composite thin-walled structure comprising:

[0108] The mechanism modeling and calculation module is used to obtain the theoretical analytical solution of the stress distribution and specific damage morphology of the composite thin-walled structure assembly by combining the collaborative modeling and analysis of multiple assembly physical fields with the macro-micro multi-scale modeling and analysis, and use it as the mechanism data;

[0109] The data acquisition module is used to arrange data acquisition devices in the key characteristic areas of the composite thin-wall structure to collect the measured data of stress and damage during the forced clamping process;

[0110] Data feature extraction module, used to process and integrate the measured data of test stress and damage, and extract assembly stress and damage data features;

[0111] The simulation model optimization module is used to establish a finite element simulation model of assembly positioning stress and damage based on mechanism data and measured data, obtain finite element simulation data, compare finite element simulation data, measured data and mechanism data, analyze the source of deviation according to the comparison results, and correct and optimize the finite element simulation model and parameters, obtain an optimized finite element simulation model of assembly positioning stress and damage, and generate large sample data of assembly quality;

[0112] Prediction model building module, which is used to train large sample data of assembly quality using support vector regression to evaluate stress distribution and damage under different clamping conditions;

[0113] The threshold alarm module is used to monitor the assembly process of composite weak-rigidity thin-walled structures in real time, input the process data into the optimized assembly positioning stress and damage finite element simulation model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

[0114] like Figure 2The figure shows a specific implementation flow chart of the present scheme. The general idea adopted by the present invention to solve its technical problems is as follows: First, a combination of multi-assembly physical field collaborative modeling analysis and macro-micro multi-scale modeling analysis is used to obtain theoretical analytical solutions for the assembly stress distribution and specific damage morphology of composite thin-walled structures, providing more interpretability for the improved analysis of the clamping process. Secondly, precision strain sensors are arranged at key positions of composite materials, and a combination of terahertz, DIC and ultrasonic measurement technologies is used to collect stress, strain and damage data in real time and comprehensively in a non-destructive manner during the forced positioning and clamping process, forming a high-density monitoring network. Thirdly, the test stress and damage data are processed and fused for analysis, signal processing technology is used to filter out noise and extract assembly stress and damage data features, low-pass, median and Kalman filtering technologies are used to improve data quality, and wavelet analysis and Fourier transform are used to deeply mine damage signals, providing a clear and reliable data basis for subsequent analysis. Fourth, organically integrate the finite element simulation data with the measured data and the data calculated by the mechanism model, establish the assembly positioning stress and damage finite element simulation model based on the mechanism data and the measured data, and accurately solve the problem of high-quality and large sample data required for data-driven by adjusting the simulation model parameters until the predicted results are highly consistent with the test results. Fifth, use the data-driven method of support vector regression to train the assembly internal stress and damage distribution prediction data, and through the numerical consistency adjustment, quickly and accurately evaluate the stress distribution and damage occurrence possibility under different clamping conditions in advance. Finally, monitor the assembly process of the weak rigid thin-walled structure of the composite material in real time, input the process data into the assembly stress and damage prediction model, and obtain the prediction results of the stress distribution and damage area. Once any damage risk exceeding the safety threshold is predicted, the system immediately issues an early warning and quickly conducts a new round of iterative feedback analysis on the forced clamping process, effectively preventing structural failure and ensuring production safety and the service life of the assembly structure.

[0115] The present invention provides a stress damage evolution prediction technology in forced clamping of aviation composite thin-walled structures. The stress damage evolution prediction technology is aimed at the damage that occurs to aviation composite thin-walled structures under forced positioning clamping during the assembly stage of aviation composite thin-walled structures. In order to suppress the occurrence of damage, it is first necessary to analyze the evolution process of assembly stress damage, and obtain theoretical analytical solutions for assembly stress distribution and specific damage morphology of composite thin-walled structures by combining multi-assembly physical field collaborative modeling analysis with macro-micro multi-scale modeling analysis; by arranging data acquisition devices in key feature areas of composite thin-walled structures, after collecting stress and damage data during forced clamping, processing and fusion analysis are performed to extract assembly stress and damage data. Features: By establishing a finite element simulation model of assembly positioning stress and damage based on mechanism data and measured data, the finite element model and parameters are corrected and optimized in advance, so that it has the ability to accurately generate assembly stress damage data status; support vector regression is used to train assembly internal stress and damage distribution prediction data, and the stress distribution and damage possibility under different clamping conditions are quickly and accurately evaluated; finally, by real-time monitoring of the assembly process of weak-rigidity thin-walled structures of composite materials, short-term stress distribution and damage areas are predicted, and early warnings are immediately triggered in dangerous situations, which can not only ensure the safety and efficiency of the assembly process, but also identify potential assembly stress concentration and damage risk areas before assembly, and guide the setting of reasonable clamping process plans. The stress damage evolution prediction technology in forced clamping of aviation composite thin-walled structures includes the following specific technical implementation processes:

[0116] S1: Theoretical analytical solutions for stress distribution and specific damage morphology of composite thin-walled structure assemblies are obtained by combining collaborative modeling and analysis of multi-assembly physical fields with macro-micro multi-scale modeling and analysis.

[0117] Specifically, according to the characteristics of heterogeneous distribution of composite laminate structures, the macro and micro scales are coupled, and a multi-scale stress bridging matrix is ​​established based on the continuous medium theory. The stress of the unit in the laminate mesoscopic model can be converted and calculated into the stress of the fiber, matrix and interface in the microscopic unit cell model. When the material components in the unit cell are damaged, the reduced stiffness of the material is calculated, and the reduced equivalent stiffness of the corresponding meso unit is obtained through the macro-micro stiffness equivalent matrix.

[0118] Furthermore, considering the influence of factors such as the layup form of composite materials, structural geometry, clamping constraints and imposed boundary conditions, forced clamping loads and assembly deformation, the macroscopic anisotropic elastoplastic parameters of the structure are calculated, the clamping contact force distribution and stress field response of the composite laminate structure are analyzed, and a forced assembly macro- and micro-stress quantification calculation model is established to obtain the size of the assembly internal stress field and the deformation state at the fiber, matrix and interface.

[0119] Furthermore, by integrating the internal stress factors of the clamping, a three-dimensional progressive damage initiation model of composite laminates under forced loading is established, the damage initiation criterion is formed, and the analytical solutions of various assembly damage morphologies and distributions such as crushing, delamination, matrix failure and even fiber fracture are obtained.

[0120] S2: Arrange data acquisition devices in the key feature areas of the composite thin-walled structure to collect stress and damage data during the forced clamping process.

[0121] Specifically, due to the anisotropic characteristics of the composite panel structure, stress concentration areas include locations such as joints, turning points, or around openings where high stress or large strain may occur. Therefore, strain sensors should be arranged more densely in these areas.

[0122] Furthermore, by combining terahertz, DIC and ultrasonic measurement technologies, the three work together to dynamically monitor and accurately evaluate the stress and damage state of composite thin-walled structures during forced assembly in a non-destructive manner;

[0123] Furthermore, when constructing a three-dimensional monitoring network, the two-dimensional large-area characteristics of the composite wall panels and the complexity of their thickness direction are taken into consideration. Therefore, the layout of the sensor network needs to be flexibly adapted to the structural characteristics of the composite wall panels of different shapes and sizes. By optimizing the wiring path and the combination of sensor types, it is ensured that the key characteristic areas of the composite thin-walled structure are under monitoring. The arranged sensor system will collect assembly stress and damage data in real time and synchronously.

[0124] Furthermore, during the assembly process of composite wall panels, sensors will collect stress, strain, damage and other data of key parts of composite wall panels in real time and synchronously, and obtain data reflecting the internal stress distribution of composite wall panels under forced loading, characterizing the shape changes of materials caused by external forces, and the damage and expansion of composite wall panels during loading.

[0125] S3: Process and integrate the test stress and damage data to extract the characteristics of assembly stress and damage data.

[0126] Specifically, the obtained data on internal stress, strain, damage, etc. of the above-mentioned composite assemblies are preprocessed, and low-pass filtering is used to remove high-frequency noise and retain slowly changing real physical signals; median filtering is used to effectively eliminate pulse noise and protect the data from instantaneous interference; Kalman filtering technology is used to provide optimal estimation in a dynamic environment and track the time-varying stress distribution and damage evolution; after removing the irrelevant signals obtained by the above-mentioned sensor measurements, suitable interpolation algorithms such as linear interpolation, spline interpolation, etc. are used to complete the missing or inconsistent data points to ensure the continuity and integrity of the data sequence and improve the quality of data fusion.

[0127] Furthermore, from the pre-processed assembly stress and damage data, by calculating the rate of change of strain values ​​at continuous time points, the sudden changes in stress state are identified, and statistical analysis is applied to calibrate the abnormally high-value areas in the stress distribution on the composite thin-walled structural parts, further confirming the stress concentration points and the areas where damage begins; through the wavelet analysis technology, with its time-frequency localization characteristics, the signal characteristics at different time scales and frequencies are distinguished, and the damage-related signals of the composite thin-walled structural parts are separated from the complex dynamic responses; using Fourier transform, the time domain signal is converted to the frequency domain, revealing the intrinsic connection between specific frequency components and composite material stress and damage; through the above-mentioned application of signal processing and pattern recognition algorithms, the key feature vectors reflecting the evolution of damage are extracted from the data.

[0128] Furthermore, for the extracted data features, Z-score standardization is used to convert the data into a form with a mean of 0 and a standard deviation of 1; normalization usually maps the data to the interval [0, 1] to ensure that the fused data are comparable at the same level, thereby improving the data processing efficiency of subsequent machine learning models.

[0129] S4: Establish a finite element simulation model of assembly positioning stress and damage based on mechanism data and measured data, analyze the source of deviation according to the comparison results, and modify and optimize the finite element model and parameters.

[0130] Specifically, according to the actual geometric dimensions, elastic modulus, Poisson's ratio, damage threshold and other material properties of the composite wall panel, as well as the specific method of forced clamping and boundary conditions such as applied loads, the finite element analysis software ABAQUS was used to construct a three-dimensional model. The grid was rationally planned in the expected stress concentration areas and damage-prone areas to improve the resolution and accuracy of the simulation. At the same time, the contact conditions, friction coefficient and possible prestressed state were reasonably set to ensure the physical authenticity of the model.

[0131] Furthermore, the ABAQUS finite element software was started, the cohesive bilinear constitutive model and the Cohesive bonding unit were set, and a damage simulation model of the composite assembly structure with initial damage defects was established.

[0132] Furthermore, when there are initial delamination defects in the parts, if some weak damage characteristics cannot be monitored, in order to analyze the expansion of defects and the occurrence of new damage during the forced clamping process, the mechanism model in S1 is used. On the basis of the established simulation model containing initial defects, the three-dimensional Hashin damage criterion and the constitutive model of damage degradation of composite material performance under the current forced clamping process are introduced.

[0133] Furthermore, based on the established multi-scale mechanism model, damage criterion of composite structure and continuous stiffness degradation constitutive relation were introduced, and the traction-separation law was used to design the finite element software UMAT subroutine for secondary development, and the bilinear degradation constitutive model of matrix and fiber interface and the material damage constitutive model were established.

[0134] Furthermore, the entire forced clamping process, including the loading step, the load holding stage and the unloading stage, was simulated in ABAQUS software. During the simulation, the stress-strain response, displacement field distribution, and possible damage signs of the composite assembly structure under different clamping parameters were obtained. The stress distribution, deformation mode, and prediction of the damage area were intuitively displayed through color coding or contour maps, and the influence of the forced clamping operation on the assembly quality was summarized.

[0135] Furthermore, the test data collected through the sensor network in the previous stage were compared with the simulation data results such as stress-strain response, displacement field distribution, and damage sign distribution obtained through finite element operation. The stress, strain, shape change and damage characteristic distribution state recorded in the test were matched one by one with the simulation data. The difference between the two was quantified through the root mean square error and relative error, and the simulation results were analyzed, such as whether the stress concentration area is consistent, the damage morphology and distribution are consistent with the actual observations, etc.

[0136] Furthermore, based on the results of comparative analysis, the sources of deviation between the simulation model and the actual situation are identified, and the finite element model is corrected and optimized by adjusting the simulation model, material parameters, refining the mesh, improving boundary conditions or adopting a more advanced damage model. Then, the simulation process is repeated until the simulation results achieve a satisfactory match with the test data.

[0137] Furthermore, the corresponding assembly stress and damage analysis results can be simulated under different forced clamping conditions and parameters, so as to accurately generate large data samples of assembly quality in forced clamping.

[0138] S5: Use support vector regression to train assembly internal stress and damage distribution prediction data to quickly and accurately evaluate stress distribution and damage probability under different clamping conditions.

[0139] Specifically, the large sample data generated by the finite element simulation in the S4 stage, including stress distribution, damage distribution, strain state, etc., are summarized. Then, the internal assembly stress, strain, damage distribution and other data of the composite thin-wall assembly structure obtained by the finite element simulation analysis in the S4 stage are standardized and normalized:

[0140] Assume that the original data matrix is ,in Indicates samples, namely Clamping layout, Indicates The standardized operation can be expressed as follows:

[0141]

[0142] in, and They are The mean and standard deviation of the features. Normalization is usually expressed as:

[0143]

[0144] Furthermore, 70% of the simulation prediction data of internal stress and damage distribution of the assembly is used as the training set, and the remaining 30% is used as the test set. Based on the radial basis kernel function in the support vector regression model, the input data is mapped from the low-dimensional space to the high-dimensional space, so that the originally nonlinearly separable data becomes linearly separable in the high-dimensional space, thereby finding an optimal hyperplane to minimize the error between the predicted value and the actual stress damage value.

[0145] Let the original input be , the function mapped to high-dimensional space is , the radial basis kernel function (such as the Gaussian kernel) is defined as:

[0146]

[0147] in, is the width parameter of the kernel function.

[0148] Consider the objective function of SVR, which includes the penalty term and insensitivity factor , which can be formalized as:

[0149]

[0150] The main constraints are:

[0151]

[0152] in, It is The true stress and damage value of the samples, is the slack variable, Represents the weight vector With the mapped input The inner product of .

[0153] Furthermore, for a given and Parameter combination, through Calculate the average prediction error using cross validation , find a set of optimal parameters that minimizes the prediction error of the model on the validation data that did not participate in the training:

[0154]

[0155] Furthermore, the optimized parameters are used to train the SVR model, and the prediction accuracy and fitting quality of the model are evaluated by calculating the mean square error (MSE), root mean square error (RMSE), and determination coefficient (R²) between the predicted value and the true value, so that the assembly stress distribution and the possibility of damage under different forced clamping conditions can be quickly and accurately evaluated. MSE and RMSE are used to measure the size of the prediction error. The smaller the value, the more accurate the prediction; the R² value close to 1 indicates that the model can better explain the variation in the data and the model performance is excellent.

[0156]

[0157]

[0158]

[0159] in, is the model prediction value, is the average of the true values.

[0160] S6: Monitor the assembly process of composite weak-rigid thin-walled structures in real time, input process data into the assembly stress and damage prediction model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the danger threshold is exceeded.

[0161] Specifically, during the dynamic process of composite panel assembly under different clamping layouts, the sensor network arranged at key nodes continuously collects real-time data such as stress, strain, and damage inside the composite panel, directly reflects the immediate response of the composite thin-wall structure under the current clamping state, and transmits it to the central processing unit in real time, providing the latest and most comprehensive information input for the prediction model.

[0162] Furthermore, after receiving the real-time data, the pre-trained support vector regression model is immediately put into operation. Based on the damage evolution law learned through training and data processing, it can quickly predict the stress change trend and potential high-risk areas of damage inside the composite wall panel in the short term in the future.

[0163] Furthermore, the stress distribution and damage area predicted by the support regression model are compared with the pre-set safety thresholds to accurately distinguish between normal operating conditions and potential structural safety hazards.

[0164] Furthermore, when the model prediction results show that the stress level or damage possibility of a clamping area exceeds the preset threshold, an early warning is automatically triggered, and a new round of iterative feedback analysis of the forced clamping process is required to guide the setting of a reasonable forced clamping process plan.

[0165] The present invention provides a stress damage evolution prediction technology in the forced clamping of aviation composite thin-walled structures, and deeply explores and quantifies the influence of the force during the assembly process on the internal microstructural damage and overall stress distribution of aviation composite thin-walled structures. In order to solve the problems of internal stress accumulation and structural damage caused by forced assembly during the assembly of aviation composite thin-walled structures, the stress damage evolution prediction and analysis is carried out through the organic integration of mechanism modeling data, measured data, and finite element simulation data and data-driven enabling means. It can identify stress concentration areas and potential damage risk areas that may be caused by forced assembly before or in the early stage of assembly, and set a reasonable clamping process plan in advance, thereby avoiding structural failure in the later stage.

[0166] The stress damage evolution prediction technology in the forced clamping of aviation composite thin-walled structures proposed in this invention mainly includes the following six parts, namely: ① assembly stress and damage mechanism modeling and calculation; ② on-site assembly quality data collection; ③ data processing and feature extraction; ④ finite element simulation model optimization; ⑤ assembly stress and damage prediction model construction and training; ⑥ real-time monitoring and early warning. After example optimization verification, the present invention can achieve:

[0167] 1) Establish a mechanism model for the internal stress and damage of the composite thin-walled structure during the forced clamping process to provide more explainability for the improvement analysis of the clamping process.

[0168] 2) Precision sensors are placed at key locations of composite materials to collect stress, strain and damage data in real time and comprehensively during forced positioning and clamping, forming a high-density monitoring network.

[0169] 3) Use signal processing technology to filter out noise and extract key features to provide a clear and reliable data basis for subsequent analysis.

[0170] 4) Organically integrate finite element simulation data with measured data and data calculated by mechanism model. By adjusting the simulation model parameters until the predicted results are highly consistent with the experimental results, the problem of high-quality and large sample data required for data-driven can be accurately solved.

[0171] 5) By applying data training-driven technology and adjusting the consistency between numbers and reality, the stress distribution and potential damage areas in the assembly can be quickly predicted in advance, reducing the number of actual assembly tests and measurements, and significantly improving assembly production efficiency.

[0172] 6) Real-time monitoring and input of assembly process data into the prediction model. Once any damage risk exceeding the safety threshold is predicted, the system immediately issues an early warning and quickly conducts a new round of iterative feedback analysis on the forced clamping process.

[0173] In this embodiment, the stress damage evolution prediction application of a certain type of aircraft composite wing box assembly process is taken as an example. First, based on the assembly process sequence of the wing box components of this type and the positioning method of each part, combined with the actual manufacturing error of each composite thin-walled composite structure during the assembly process, the clamping load and the positioning position constraint and other factors, the assembly stress distribution between the composite wing box skeleton structure, the skeleton assembly and the upper and lower wall panels, and the theoretical analytical solution of the specific damage morphology are obtained by combining the multi-assembly physical field collaborative modeling analysis with the macro-micro multi-scale modeling analysis of the composite parts. Secondly, a strain gauge measurement system is arranged around the contact parts between the components of the wing box components. By means of the collaborative work of terahertz, DIC and ultrasonic measurement technology, the stress and damage state of each component of the wing box under the clamping force of the tooling locator is dynamically monitored and accurately evaluated in a non-destructive manner. The stress and damage data characteristics of each assembly stage are extracted through the processing and fusion analysis of the measurement data, and the stress concentration points and damage areas are located. Third, the finite element software ABAQUS is used to simulate the wing box assembly process to obtain the stress, strain, damage distribution and state of each component inside the wing box under different clamping layouts. By comparing with the mechanism data and measured data, the model parameters are continuously adjusted until the predicted results are highly consistent with the experimental observations, so that the finite element model has the ability to accurately generate assembly stress damage data states. Fourth, support vector regression is used to train the internal stress and damage distribution prediction data of each assembly stage to quickly and accurately evaluate the stress distribution and damage possibility of each component of the wing box under different clamping conditions. Finally, the assembly process of each component of the wing box is monitored in real time, and the detection data is input into the data-driven model to predict the stress distribution and damage area of ​​the next assembly stage. According to the comparison between the predicted results and the safety threshold, potential risks are identified in time. Once the threshold is exceeded, the early warning system is immediately triggered to ensure the safety of the forced clamping operation of each part, and identify the potential assembly stress concentration and damage risk areas before the subsequent assembly steps are carried out, so as to guide the setting of a reasonable clamping process plan. During the assembly process of a certain type of wing box, the stress and damage cloud map at a certain moment, such as Figure 5 As shown. The stress damage evolution prediction technology of a certain type of wing box includes the following steps:

[0174] Specifically, Figure 3A certain type of composite wing box in the invention consists of thin-walled structures such as the left wall, the right wall, the first frame, the second frame, the third frame, the fourth frame, and the upper and lower wall panels. Its assembly is mainly divided into four steps: ① Fix and install the left and right walls and the first and fourth frame parts on the tooling; ② Use this as the positioning reference to install the second and fourth frames; ③ Install the composite lower wall panel assembly; ④ Install the composite lower wall panel assembly to form a complete wing box component. The above positioning and clamping installation processes all require the assistance of assembly tooling fixtures.

[0175] Furthermore, on the basis of the assembly process sequence of the wing box components and the positioning method of each part, combined with the actual manufacturing errors of each composite thin-walled composite structure in the assembly process, clamping loads and positioning position constraints, the method combines the collaborative modeling analysis of multiple assembly physical fields with the macro-micro multi-scale modeling analysis of composite parts, and obtains the reduced equivalent stiffness of the corresponding micro-unit through the macro-micro stiffness equivalent matrix. The clamping contact force distribution and stress field response of the composite laminate structure are analyzed, and a forced assembly macro-micro stress quantification calculation model is established to obtain the size and deformation state of the assembly internal stress field at the fiber, matrix and interface. A three-dimensional progressive damage initiation model between the composite wing box skeleton structure and between the skeleton components and the upper and lower wall panels under forced loading is established to form a damage initiation criterion, and to obtain analytical solutions for various assembly damage morphologies and distributions such as crushing, delamination, matrix damage and even fiber fracture.

[0176] Furthermore, a strain gauge measurement system is arranged around the contact parts between the components of the wing box assembly, and a three-dimensional monitoring network covering the composite wing box skeleton structure and between the skeleton assembly and the upper and lower wall panels under forced loading is constructed by means of the coordinated work of terahertz, DIC and ultrasonic measurement technologies. The stress and damage state of each component of the wing box under the clamping force of the tooling locator is dynamically monitored and accurately evaluated in a non-destructive manner. The stress and damage data characteristics of the wing box components in four assembly stages are extracted through real-time and synchronous collection of measurement data, as well as preprocessing and fusion analysis. While accurately locating the stress concentration points and damage areas, it ensures that the fused data are comparable at the same level.

[0177] Furthermore, according to the actual geometric dimensions, material properties and boundary conditions of the composite wing box assembly structural parts, as well as real constraints such as forced clamping parameters, the contact conditions, friction coefficient and prestress state are set, and the finite element analysis software ABAQUS is used to construct a three-dimensional simulation model of the wing box assembly process. During the simulation process, it is also necessary to define the cohesive bilinear constitutive model and the Cohesive bonding unit according to the characteristics of the wing box composite thin-walled structural parts, introduce the three-dimensional Hashin damage criterion and the constitutive model of the performance damage degradation of each component of the wing box under the current forced clamping process, and obtain the stress, strain, damage distribution and state of each component inside the wing box under different clamping layouts and load conditions, and intuitively display the stress distribution, deformation mode, and prediction of the damage area through color coding or contour maps.

[0178] Furthermore, by comparing the simulation data with the mechanism data and the measured data, the source of the deviation between the wing box assembly data obtained by finite element analysis and the actual situation is identified, and the wing box finite element model and the simulation process are corrected and optimized by adjusting material parameters, refining the grid, improving boundary conditions or adopting a more advanced damage model. Then, the simulation process is repeated until the simulation results and the test data achieve a satisfactory match, so that the finite element model has the ability to accurately generate the stress damage data state of the wing box at each assembly stage.

[0179] Furthermore, combined with the specific assembly stage of the wing box components, the assembly stress and damage analysis results under different assembly loads and positioning position conditions are analyzed, so as to quickly and accurately generate big data samples of assembly quality of key parts in forced clamping.

[0180] Furthermore, the large sample data of each assembly stage of the wing box components generated by the finite element simulation in the S4 stage are summarized, including stress distribution, damage distribution, strain state, etc. Then, the internal assembly stress, strain, damage distribution and other data of the composite wing box structure obtained by the finite element simulation analysis in the S4 stage are standardized and normalized:

[0181] Assume that the original data matrix is ,in Indicates samples, namely Clamping layout, Indicates The standardized operation can be expressed as follows:

[0182]

[0183] in, and They are The mean and standard deviation of the features. Normalization is usually expressed as:

[0184]

[0185] Furthermore, 70% of the simulation prediction data of assembly stress and damage distribution inside the wing box components are used as training sets, and the remaining 30% are used as test sets. Based on the radial basis kernel function in the support vector regression model, the input data is mapped from low-dimensional space to high-dimensional space, so that the originally nonlinearly separable data becomes linearly separable in the high-dimensional space, thereby finding an optimal hyperplane to minimize the error between the predicted value and the actual stress damage value.

[0186] Let the original input be , the function mapped to high-dimensional space is , the radial basis kernel function (such as the Gaussian kernel) is defined as:

[0187]

[0188] in, is the width parameter of the kernel function.

[0189] Consider the objective function of SVR, which includes the penalty term and insensitivity factor , which can be formalized as:

[0190]

[0191] The main constraints are:

[0192]

[0193] in, It is The true stress and damage value of the samples, is the slack variable, Represents the weight vector With the mapped input The inner product of .

[0194] Furthermore, for a given and Parameter combination, through Calculate the average prediction error using cross validation , find a set of optimal parameters that minimizes the prediction error of the model on the validation data that did not participate in the training:

[0195]

[0196] Furthermore, the optimized parameters are used to train the SVR model, and the prediction accuracy and fitting quality of the model are evaluated by calculating the mean square error (MSE), root mean square error (RMSE), and determination coefficient (R²) between the predicted value and the true value, so that the stress distribution and damage probability of the wing box components at each assembly stage under different forced clamping conditions can be quickly and accurately evaluated. MSE and RMSE are used to measure the size of the prediction error. The smaller the value, the more accurate the prediction; the R² value close to 1 indicates that the model can better explain the variation in the data and the model performance is excellent.

[0197]

[0198]

[0199]

[0200] in, is the model prediction value, is the average of the true values.

[0201] Furthermore, the assembly process of each component of the wing box is monitored in real time, and the real-time data of stress, strain, damage, etc. generated by each component of the wing box during assembly are collected to reflect the immediate response of the wing box in the current clamping state in real time. After that, the detection data of the wing box assembly process is input into the data-driven model to predict the stress distribution and damage area of ​​the next assembly stage, and the potential risks are identified in time according to the comparison between the prediction results and the safety threshold. Once the threshold is exceeded, the early warning system is immediately triggered to ensure the safety of the forced clamping operation of each part, and the potential assembly stress concentration and damage risk area are identified before the subsequent assembly steps are carried out, so as to guide the setting of a reasonable clamping process plan.

[0202] The stress damage evolution prediction technology in forced clamping of aviation composite thin-walled structures proposed in the present invention mainly includes the following six parts, namely: ① assembly stress and damage mechanism modeling and calculation; ② on-site assembly quality data collection; ③ data processing and feature extraction; ④ finite element simulation model optimization; ⑤ assembly stress and damage prediction model construction and training; ⑥ real-time monitoring and early warning. The following can be achieved: ① Establish a mechanism model for the occurrence of internal stress and damage in the forced clamping process of composite thin-walled structures, providing more explainability for the improvement and analysis of the clamping process; ② Arrange precision sensors at key positions of composite materials to collect stress, strain and damage data in real time and comprehensively during the forced positioning clamping process to form a high-density monitoring network; ③ Use signal processing technology to filter out noise and extract key features to provide a clear and reliable data basis for subsequent analysis; ④ Organically integrate finite element simulation data with measured data and data calculated by the mechanism model, and adjust the simulation model parameters until the predicted results match the test results, which can accurately solve the problem of high-quality and large sample data required for data-driven; ⑤ Apply data training-driven technology, through digital-real consistency adjustment, the internal stress distribution and potential damage area of ​​the assembly can be quickly predicted in advance, reducing the number of actual assembly tests and measurements, and significantly improving assembly production efficiency; ⑥ Real-time monitoring and input of assembly process data into the prediction model. Once any damage risk exceeding the safety threshold is predicted, the system immediately issues an early warning and quickly conducts a new round of iterative feedback analysis on the forced clamping process. The above is a detailed introduction to a stress damage evolution prediction method and system for forced clamping of thin-walled composite aviation structures provided in the embodiments of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application; at the same time, for those skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0203] For example, certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that hardware manufacturers may use different nouns to refer to the same component. This specification and claims do not use differences in names as a way to distinguish components, but use differences in the functions of components as the criteria for distinction. As mentioned throughout the specification and claims, "including" and "comprising" are open-ended terms, so they should be interpreted as "including / including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve the technical problem within a certain error range and basically achieve the technical effect. The subsequent description of the specification is a preferred embodiment of the present application, but the description is for the purpose of illustrating the general principles of the present application, and is not used to limit the scope of the present application. The scope of protection of the present application shall be determined by the definition of the attached claims.

[0204] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such a product or system. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the product or system including the elements.

[0205] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0206] The above description shows and describes several preferred embodiments of the present application, but as mentioned above, it should be understood that the present application is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the application concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present application, and should be within the scope of protection of the claims attached to the present application.

Claims

1. A method for predicting stress damage evolution in forced clamping of thin-walled aviation composite structures, characterized in that: The stress damage evolution prediction method in the forced clamping of the aviation composite thin-walled structure includes: S1: The theoretical analytical solution of stress distribution and specific damage morphology of composite thin-walled structure assembly is obtained by combining multi-assembly physical field collaborative modeling analysis with macro-micro multi-scale modeling analysis, which is used as mechanism data; The S1 specifically includes: S11: Coupling macro and micro scales, a multi-scale stress bridging matrix is ​​established based on the continuum theory. When the material component in a single cell is damaged, its material reduced stiffness is calculated, and the reduced equivalent stiffness of the corresponding micro unit is obtained through the macro-micro stiffness equivalent matrix; S12: Considering the influence of composite material layup form, structural geometry, clamping constraints and applied boundary conditions, forced clamping load and assembly deformation factors, a forced assembly macro-micro stress quantification calculation model is established to obtain the assembly internal stress field size and deformation state at the fiber, matrix and interface; S13: Integrate the internal stress factors of the clamping to establish a three-dimensional progressive damage initiation model of composite laminates under forced loading, form damage initiation criteria, and obtain analytical solutions for specific damage morphology and distribution; S2: Arrange data acquisition devices in the key characteristic areas of the composite thin-walled structure to collect measured data on stress and damage during forced clamping; S3: Process and integrate the measured data of test stress and damage to extract key feature vectors; The S2 specifically includes: S21: Arrange strain sensors in the expected stress concentration area of ​​the composite panel; S22: By combining terahertz, DIC and ultrasonic measurement technologies, the three work together to dynamically monitor and accurately evaluate the stress and damage state of composite thin-walled structures during forced assembly in a non-destructive manner; S23: The sensor network is optimally arranged according to the assembly characteristics of the composite structure. Through the comprehensive layout of sensors, a three-dimensional monitoring network is formed to ensure that the key characteristic areas of the composite thin-walled structure are under monitoring. During the forced clamping process of the composite thin-walled structure, the arranged sensor system will collect assembly stress and damage data in real time and synchronously; The S3 specifically includes: S31: pre-processing the internal stress, strain and damage data of the composite assembly, using low-pass filtering, median filtering or Kalman filtering digital signal processing technology to remove irrelevant signals obtained by the sensor measurement, and interpolating missing or inconsistent data points to improve data fusion quality; S32: extracting key feature vectors reflecting the evolution of assembly stress and damage from the pre-processed assembly stress and damage data through signal analysis; S33: Standardize or normalize the key feature vectors to ensure that the fused data can be compared at the same level and improve the data processing efficiency of subsequent machine learning models; The process of extracting damage characteristics through signal analysis is specifically as follows: calculating the change rate of strain values ​​at adjacent time points to identify the areas where stress concentration and damage begin; identifying abnormally high-value areas in stress distribution by statistical methods; converting time domain signals to frequency domain by Fourier transform to analyze the relationship between specific frequency components and stress damage; applying signal processing and pattern recognition algorithms to extract key feature vectors reflecting damage evolution from the data; S4: Establish an assembly positioning stress and damage finite element simulation model based on mechanism data and key feature vectors, obtain finite element simulation data, compare finite element simulation data, key feature vectors and mechanism data, analyze the source of deviation according to the comparison results, and modify and optimize the finite element simulation model and parameters, obtain an optimized assembly positioning stress and damage finite element simulation model, and generate large sample data of assembly quality; The S4 specifically includes: S41: According to the actual geometric dimensions, material properties and boundary conditions of the composite thin-walled structural parts, as well as the real constraints of the forced clamping parameters, the contact conditions, friction coefficient and prestress state are set, and the finite element analysis software ABAQUS is used to build a three-dimensional simulation model; S42: Using the ABAQUS cohesive bilinear constitutive model and the cohesive bonding unit, a damage simulation model of the composite assembly structure with initial damage defects is established; S43: When there are initial delamination defects in the parts, if some weak damage characteristics cannot be monitored, in order to analyze the expansion of defects and the occurrence of new damage during the forced clamping process, the mechanism model in S1 is used. On the basis of the established simulation model with initial defects, the three-dimensional Hashin damage criterion and the constitutive model of composite material performance damage degradation under the current forced clamping process are introduced; S44: Analyze the stress response, damage evolution and expansion, and dimensional changes of composite assembly structures under different clamping parameters to obtain the influence of forced clamping operation on assembly quality; S45: Compare the stress-strain response, displacement field distribution, and damage sign distribution simulation data obtained through finite element operation with the key eigenvectors and the data calculated by the mechanism model, identify the source of the deviation between the model and the actual situation based on the comparison results, and continuously adjust the simulation model parameters until the prediction results are consistent with the test results; S46: Analyze the assembly stress and damage analysis results under different assembly conditions to accurately generate assembly quality big data samples in forced clamping; S5: Use support vector regression to train large sample data of assembly quality to evaluate stress distribution and damage under different clamping conditions; The S5 specifically includes: S51: Standardize and normalize the large sample data obtained through finite element simulation analysis in stage S4. At the same time, extract the key feature vectors of assembly stress and damage evolution based on the analysis results of stage S3; S52: Using 70% of the assembly internal stress and damage distribution simulation prediction data as a training set and the remaining 30% as a test set, a support vector regression model is constructed to obtain the optimal hyperplane and the error between the predicted assembly stress damage value and the actual value; S53: Based on cross-validation, the penalty coefficient and kernel function parameters are adjusted, and multiple iterations are performed until a set of parameters is found that minimizes the prediction error of the support vector regression model on the validation set, and the optimized parameters are obtained to train the SVR model; S54: Use the optimized parameters to train the SVR model, and evaluate the performance of the model by calculating the mean square error, root mean square error, and determination coefficient between the predicted value and the true value. After the model training is completed, the assembly stress distribution and the possibility of damage under different forced clamping conditions can be quickly and accurately evaluated; S6: Monitor the assembly process of composite weak-rigid thin-walled structures in real time, input process data into the optimized parameter training SVR model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

2. The method for predicting stress damage evolution in forced clamping of thin-walled aviation composite structures according to claim 1 is characterized in that: The process of constructing the constitutive model of damage degradation of composite material performance is specifically as follows: based on the established multi-scale mechanism model, the damage criterion of the composite structure and the continuous degradation constitutive relation of stiffness are introduced, and the traction-separation law is used to design the finite element software UMAT subroutine for secondary development, and establish a bilinear degradation constitutive model of the matrix and fiber interface and a material damage constitutive model.

3. The stress damage evolution prediction method in forced clamping of aviation composite thin-walled structures according to claim 1 is characterized in that: The S6 specifically includes: S61: Real-time monitoring of the assembly process of composite weak-rigid thin-walled structures, collecting stress and strain data during the assembly process from a sensor network placed at key locations; S62: Use the trained support vector regression model to process the received process data, and predict the assembly stress distribution inside the structure and the possible damage area in the short term in the future based on the damage evolution law learned through training; S63: Compare the stress distribution and damage area predicted by the model with the pre-set safety threshold to distinguish between normal working conditions and potential assembly damage risks; S64: When the model prediction results show that the stress level or damage possibility of a clamping area exceeds the preset threshold, an early warning is automatically triggered, and a new round of iterative feedback analysis of the forced clamping process is required.

4. A stress damage evolution prediction system for forced clamping of thin-walled composite aviation structures, characterized in that: The stress damage evolution prediction system in forced clamping of aviation composite thin-walled structures includes: The mechanism modeling and calculation module is used to obtain the theoretical analytical solution of the stress distribution and specific damage morphology of the composite thin-walled structure assembly by combining the collaborative modeling analysis of multiple assembly physical fields with the macro-micro multi-scale modeling analysis, and use it as the mechanism data. Specifically, it includes: coupling the macro- and micro-scales, establishing a multi-scale stress bridging matrix based on the continuous medium theory, calculating the reduced stiffness of the material when the material component in the single cell is damaged, and obtaining the reduced equivalent stiffness of the corresponding micro-unit through the macro-micro stiffness equivalent matrix; considering the ply form, structural geometry, clamping constraints and applied boundary conditions of the composite material, Influenced by the forced clamping load and assembly deformation, a forced assembly macro-micro stress quantification calculation model is established to obtain the size and deformation state of the assembly internal stress field at the fiber, matrix and interface; the clamping internal stress factor is integrated to establish a three-dimensional progressive damage initiation model of composite laminates under forced loading, form a damage initiation criterion, and obtain an analytical solution for the specific damage morphology and distribution; data acquisition devices are arranged in the key feature areas of the composite thin-walled structure to collect the measured data of stress and damage during the forced clamping process; the measured data of experimental stress and damage are processed and fused and analyzed to extract key feature vectors; The data acquisition module is used to arrange data acquisition devices in the key characteristic areas of the composite thin-walled structure to collect measured data on stress and damage during the forced clamping process; specifically, it includes: arranging strain sensors in the expected stress concentration areas of the composite wall panels; through the combination of terahertz, DIC and ultrasonic measurement technologies, the three work together to dynamically monitor and accurately evaluate the stress and damage state of the composite thin-walled structure during the forced assembly process in a non-destructive manner; the sensor network is optimally arranged according to the assembly characteristics of the composite structure, and a three-dimensional monitoring network is formed through the comprehensive layout of sensors to ensure that the key characteristic areas of the composite thin-walled structure are under monitoring. During the forced clamping process of the composite thin-walled structure, the arranged sensor system will collect assembly stress and damage data in real time and synchronously; The data feature extraction module is used to process and fuse the measured data of the test stress and damage, and extract key feature vectors; specifically, it includes: pre-processing the internal stress, strain, and damage data of the above-mentioned composite assembly, using low-pass filtering, median filtering or Kalman filtering digital signal processing technology to remove the irrelevant signals obtained by the above-mentioned sensor measurement, and interpolating the missing or inconsistent data points to improve the data fusion quality; from the pre-processed assembly stress and damage data, extract the key feature vectors reflecting the evolution of assembly stress and damage through signal analysis; standardize or normalize the key feature vectors to ensure that the fused data can be compared at the same level, so as to improve the data processing efficiency of the subsequent machine learning model; the process of refining damage features through signal analysis is specifically as follows: calculating the rate of change of strain values ​​at adjacent time points, identifying the areas where stress concentration and damage begin; identifying abnormally high value areas in stress distribution through statistical methods; converting the time domain signal to the frequency domain through Fourier transform, and analyzing the relationship between specific frequency components and stress damage; applying signal processing and pattern recognition algorithms to extract key feature vectors reflecting damage evolution from the data; The simulation model optimization module is used to establish an assembly positioning stress and damage finite element simulation model based on mechanism data and key feature vectors, obtain finite element simulation data, compare finite element simulation data, key feature vectors and mechanism data, analyze the source of deviation according to the comparison results, and correct and optimize the finite element simulation model and parameters, obtain the optimized assembly positioning stress and damage finite element simulation model, and generate large sample data of assembly quality; specifically including: According to the actual geometric dimensions, material properties and boundary conditions of composite thin-walled structural parts, as well as the real constraints of forced clamping parameters, the contact conditions, friction coefficient and prestress state are set, and a three-dimensional simulation model is constructed using the finite element analysis software ABAQUS. The ABAQUS cohesive bilinear constitutive model and Cohesive bonding unit are used to establish a composite assembly structure damage simulation model with initial damage defects. When there are initial delamination defects in the parts, if some weak damage characteristics cannot be monitored, in order to analyze the expansion of defects and the occurrence of new damage during the forced clamping process, the mechanism model in the mechanism modeling calculation module is used. On the basis of the established simulation model with initial defects, a three-dimensional Hash in damage criteria and the constitutive model of composite material performance damage degradation under the current forced clamping process; analyze the stress response, damage evolution and expansion, and dimensional changes of the composite assembly structure under different clamping parameters, and obtain the influence of forced clamping operation on assembly quality; compare the stress-strain response, displacement field distribution, and damage sign distribution simulation data obtained by finite element operation with the data calculated by key eigenvectors and mechanism models, identify the source of deviation between the model and the actual situation based on the comparison results, and continuously adjust the simulation model parameters until the prediction results are consistent with the test results; analyze the assembly stress and damage analysis results under different assembly conditions, so as to accurately generate assembly quality big data samples in forced clamping; The prediction model building module is used to use support vector regression to train large sample data of assembly quality and evaluate whether stress distribution and damage occur under different clamping conditions. Specifically, it includes: standardizing and normalizing the large sample data obtained by the simulation model optimization module through finite element simulation analysis. At the same time, according to the analysis results of the data feature extraction module, the key feature vectors of assembly stress and damage evolution are extracted; 70% of the simulation prediction data of assembly internal stress and damage distribution are used as the training set, and the remaining 30% are used as the test set to build a support vector regression model, obtain the optimal hyperplane, and obtain the error between the predicted assembly stress damage value and the actual value; based on cross-validation, the penalty coefficient and kernel function parameters are adjusted, and through multiple iterations, a set of parameters is found to minimize the prediction error of the support vector regression model on the validation set, and the optimized parameters are obtained to train the SVR model; the optimized parameters are used to train the SVR model, and the performance of the model is evaluated by calculating the mean square error, root mean square error, and determination coefficient between the predicted value and the true value. After the model training is completed, the assembly stress distribution and the possibility of damage under different forced clamping conditions can be quickly and accurately evaluated; The threshold alarm module is used to monitor the assembly process of composite weak-rigidity thin-walled structures in real time, input process data into the optimized parameter training SVR model, obtain the prediction results of stress distribution and damage area, and issue an alarm when the preset danger threshold is exceeded.

Images