Composite material structure mechanical property digital twin virtual evaluation method
By establishing a digital twin framework for damage monitoring and mechanical property prediction of CFRP composite structures, the problem of inaccurate mapping between damage state and twin model was solved, enabling rapid, accurate assessment and reliable prediction of composite structure.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2026-02-09
- Publication Date
- 2026-06-12
AI Technical Summary
Existing digital twin technology for composite material structures lacks a fast and accurate automated mapping method between the damage state and the twin model, making it difficult for model updates to reflect the physical entity state in real time. Furthermore, its prediction function and verification process are weak, limiting its engineering application capabilities in scenarios with high real-time requirements.
A digital twin framework for damage monitoring and mechanical property prediction of CFRP composite structures was established. Through an automated data processing flow of Lamb wave signal and damage information, the digital twin model was rapidly updated by combining damage inversion information. The model accuracy was improved by correcting the initial model parameters, thus building a high-confidence prediction capability.
It enables rapid and accurate assessment of the damage state of composite material structures, improves the real-time performance and accuracy of digital twin systems, and ensures reliable prediction and assessment of mechanical properties.
Smart Images
Figure CN122193390A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a digital twin virtual evaluation method for the mechanical properties of composite material structures, belonging to the field of digital twin technology for composite material structures. Background Technology
[0002] In existing technologies, carbon fiber reinforced resin (CFRP) structures are prone to various typical damage forms in complex and variable service environments, such as dispersed crack damage and delamination damage. Among these, dispersed cracks and delamination are significantly insidious, difficult to detect in the early stages, but will gradually propagate and evolve under cyclic loading or sustained stress, eventually leading to critical structural failure and seriously threatening their service safety and service life in high-safety-requirement fields such as aerospace. Therefore, achieving the assessment and prediction of the mechanical properties of CFRP structures under damage states has become a key issue in ensuring the structural integrity and functionality of composite materials.
[0003] Currently, research on digital twins in the field of composite materials has moved from conceptual frameworks to concrete applications, covering multiple aspects such as manufacturing process optimization, microstructure characterization, service condition monitoring, and life prediction. However, some problems still exist. For example, there is a lack of fast and accurate mapping methods between damage states and twin models: one of the core challenges of current digital twins is how to quickly and accurately convert the quantitative damage information identified by online detection into changes in the corresponding geometry, materials, or boundary conditions in the virtual model, thereby updating the model's state. There is a lack of effective automated mapping methods for different damage types, making it difficult for the state updates of digital twin models to remain consistent with the physical entity. Prediction capabilities and experimental verification are also weak: most existing research on digital twins for composite material structures focuses on the "monitoring-diagnosis-state update" process, that is, using data to update the model to reflect the current state. However, the higher-level value of digital twins lies in actively predicting the mechanical response, damage evolution trend, and even the final failure mode of the structure under external loads based on the updated model. Related prediction methods are lacking, and experimental verification of the prediction results is also somewhat insufficient.
[0004] Existing technologies, particularly digital twin methods for composite material structures similar to this invention, still suffer from the following major shortcomings in the assessment and prediction of mechanical properties under structural damage conditions:
[0005] A rapid and accurate mapping method between damage state and digital twin model is scarce: One of the core challenges of current digital twin technology lies in how to quickly and accurately transform the quantitative damage information identified through online detection into corresponding changes in geometry, materials, or boundary conditions in the virtual model, thereby achieving dynamic updates to the model's state. Currently, there is a lack of systematic and efficient automated mapping methods for different damage types (such as delamination, fiber fracture, and matrix cracking), making it difficult to ensure that the state updates of the digital twin model are consistent with the physical entity in real time and with high accuracy. This bottleneck not only affects the reliability of the model during service but also restricts its engineering application capabilities in scenarios with high real-time requirements.
[0006] Weak Prediction and Validation Capabilities: Most existing research on digital twins for composite material structures focuses on a "monitoring-diagnosis-state update" process, primarily updating models to reflect the current structural state through data collection. However, the higher-order value of digital twin systems should lie in their predictive capabilities—based on the updated model, actively predicting the mechanical response, damage evolution trends, and even final failure modes of the structure under future external loads, providing a basis for preventative maintenance and life management decisions. Currently, relevant prediction models and methods are still imperfect, mostly remaining at the theoretical or simulation level, lacking a reliable prediction system for complex working conditions; at the same time, research on systematic experimental verification of prediction results is also relatively insufficient, hindering the functional leap of digital twins from "state reflection" to "behavioral prediction."
[0007] Therefore, there is an urgent need to develop an intelligent method that can quickly update digital twin models and predict structural residual performance in real time, thereby enabling rapid and accurate assessment of the strength and functional reliability of composite material structures and providing strong support for the evaluation of the strength and functional reliability of composite material structures with long service life. Summary of the Invention
[0008] The core technical problem this invention aims to solve is addressing two major bottlenecks in the practical application of digital twins for composite material structures. First, the lack of a fast and accurate automated mapping method between damage states and the twin model makes it difficult for model updates to reflect the real-time state of the physical entity. Second, existing systems have weak predictive capabilities and insufficient experimental verification, limiting their advancement from state reflection to behavioral prediction. This invention provides an intelligent digital twin method capable of intelligently mapping damage information and rapidly updating the model, building high-confidence prediction capabilities, and having undergone thorough experimental verification. This method supports accurate and rapid assessment of the strength and functional reliability of composite material structures under damage states. The method of this invention establishes a digital twin framework for damage monitoring and mechanical property prediction of CFRP composite material structures. Damage monitoring module: Based on the established quantitative inversion model of typical damage (dispersed cracks, delamination) of CFRP laminated structures, an automated data processing flow of Lamb wave signal-damage information was developed; Model update module: Combining damage inversion information, corresponding rapid update methods for digital twin models were proposed for different damage forms; Mechanical performance prediction module: An initial digital twin model parameter correction scheme was proposed, and the mechanical properties of damaged composite material structures were predicted and experimentally verified based on the updated digital twin model of the damage state.
[0009] A digital twin virtual evaluation method for the mechanical properties of composite material structures, comprising the following steps: Step 1: Fabricate the composite material cylinder and composite material beams; Step 2: Establish initial digital twin model A of composite material cylinder and initial digital twin model B of composite material beam, and perform parameter correction based on experimental results; The requirement is that, in the initial state, the deformation calculation results of the initial digital twin model A of the composite material cylinder are the same as the deformation experimental test results of the composite material cylinder made in step one; The requirement is that, in the initial state, the deformation calculation results of the initial digital twin model B of the composite beam are the same as the deformation test results of the composite beam made in step one. Step 3: Perform radial compression fatigue loading on the composite material cylinder prepared in Step 1, and perform bending fatigue loading on the composite material beam prepared in Step 1. Step 4: Perform Lamb wave detection on the composite material cylinder and composite material beam after fatigue loading in Step 3, and obtain Lamb wave detection data A for the composite material cylinder and Lamb wave detection data B for the composite material beam. Step 5: Based on the Lamb wave detection data A obtained in Step 4, invert the damage status A of the composite material cylinder; based on the Lamb wave detection data B obtained in Step 4, invert the damage status B of the composite material beam. Step 6: Update the initial digital twin model A established in Step 2 based on the damage condition A obtained in Step 5; update the initial digital twin model B established in Step 2 based on the damage condition B obtained in Step 5; then virtually evaluate the mechanical behavior and properties of the composite material cylinder based on the updated digital twin model A; and virtually evaluate the mechanical behavior and properties of the composite material beam based on the updated digital twin model B.
[0010] In step one, when fabricating the composite beam, a polytetrafluoroethylene film is implanted between the layers; In step five, the method for inverting the damage condition A of the composite material cylinder based on the Lamb wave detection data A is as follows: Step 511: Filter the Lamb wave detection data A to remove minor fluctuation interference; Step 512: Align the filtered Lamb wave detection data A by aligning the received wave with the excitation wave start time; Step 513: Calculate the Lamb wave velocity using the aligned Lamb wave detection data A; Step 514: Based on the Lamb wave velocity inversion, damage condition A is obtained, i.e., the stiffness of the composite material cylinder after degradation. The stiffness of the composite material cylinder after degradation is:
[0011] in, C P For Lamb wave velocity, f For the excitation frequency, E’ Stiffness after degradation; In step five, the method for inverting the damage condition B of the composite beam based on the Lamb wave detection data B is as follows: Step 521: Filter the Lamb wave detection data B to remove minor fluctuation interference; Step 522: Align the filtered Lamb wave detection data B by aligning the received wave with the excitation wave start time; Step 523: Perform wavelet packet decomposition and reconstruction on the aligned Lamb wave detection data B to obtain the signal features in the time domain, frequency domain, and time-frequency domain. Then, calculate the signal features to obtain the feature values. Step 524: Input the obtained feature values into the XGBoost machine learning model to invert and obtain damage condition B, that is, the delamination damage profile of the composite beam.
[0012] In step six, the method for updating the initial digital twin model A established in step two based on the damage condition A obtained in step five is as follows: Step 611: Divide the digital twin model into partitions based on the sensor network layout; Step 612: Create a damage factor matrix based on the partitioning results; Step 613: Identify the correspondence between each partition, layer, and damage factor matrix through the program; Step 614: Update the parameters of the digital twin model using the inp file to achieve the update of the diffuse crack damage mapping.
[0013] In step six, the method for updating the initial digital twin model B established in step two based on the damage condition B obtained in step five is as follows: Step 621: Convert the layered damage contour into coordinate form and store it; Step 622: The program determines and filters the units in the digital twin model that are within the range of the layered damage contour coordinates; Step 623: Based on the coordinates of the remaining nodes of each unit, distinguish the units that share nodes in the upper and lower layers; Step 625: Copy the nodes within the coordinate range of the layered damage contour, and replace the common nodes in the upper unit with the copied nodes to realize the layered damage mapping and update.
[0014] Beneficial effects 1. Automated Data Processing Method for Dispersed Crack Damage in Composite Materials Based on Lamb Waves: The key to this method lies in establishing a standardized Lamb wave data processing workflow to achieve automated quantitative assessment of dispersed crack damage in composite materials. The core technologies include signal filtering to eliminate environmental noise interference, precise waveform alignment based on the excitation wave initiation moment, calculation of the Lamb wave propagation velocity using the aligned waveform, and establishment of a quantitative relationship model between wave velocity change and material stiffness degradation. Finally, this model is used to predict the degree of stiffness degradation in the detection area. The core of this technical solution lies in the combined application of this fully automated processing method and its specific wave velocity-stiffness degradation model.
[0015] 2. Automated Data Processing Method for Delamination Damage in Composite Materials Based on Lamb Waves: The key to this method lies in combining signal processing and machine learning techniques to achieve automated identification and inversion of delamination damage. The core technology includes preprocessing with signal filtering and alignment, extracting effective feature values using wavelet packet decomposition and reconstruction techniques, constructing a feature value matrix, and inputting it into a pre-trained XGBoost machine learning model to automatically invert the contour and specific location of the delamination damage. The core of this technical solution lies in the automated damage inversion technology path that combines wavelet packet feature extraction with the XGBoost model.
[0016] 3. A Rapid Mapping Method for Digital Twin Models Based on Dispersed Crack Damage Inversion Results: The key to this method lies in achieving a programmed and automated mapping from damage data to changes in virtual model parameters. The core technology includes dividing the digital twin model into grids based on sensor layout, creating a damage factor matrix based on the inversion results, automatically identifying the mapping relationship between each partition and its corresponding layer and the damage factor matrix, and finally automatically updating the model stiffness parameters by modifying the model input file (e.g., the .inp file) to complete the state mapping of dispersed crack damage. The core protection of this technical solution lies in the specific mapping mechanism for batch and automatic updating of model parameters based on partitioning and the damage factor matrix.
[0017] 4. A Fast Mapping Method for Digital Twin Models Based on Layered Damage Inversion Results: The key to this method lies in accurately converting the identified layered geometric contours into geometric changes in the digital twin model. The core technology includes converting the damage contours into coordinate data, filtering units within the coordinate region of the model through a program, identifying and copying the common nodes of these units, and separating the upper and lower layers in the damage region through node replacement, thereby accurately reconstructing the geometric morphology of layered damage in the model. The core of this technical solution is a modeling method that automatically updates the geometric morphology of layered damage through coordinate mapping and node operations.
[0018] 5. Initial Digital Twin Model Parameter Correction Method: The key to this method lies in calibrating the initial material parameters of the digital twin model using experimental data to improve its fundamental accuracy. The core technology includes establishing an initial simulation model based on standard material parameters, conducting actual mechanical property tests under healthy conditions, comparing simulation results with experimental data, and making targeted fine-tuning of key material parameters in the model based on the comparison results to offset inherent deviations introduced by manufacturing processes and other factors. The core of this technical solution lies in the specific process and method of iteratively correcting model parameters through "simulation-experiment" comparison.
[0019] 6. Digital Twin Virtual Evaluation Technology Framework for Composite Material Structures: The key to this framework lies in constructing an integrated and automated digital twin system, realizing a complete closed loop from damage monitoring to performance prediction. The core technology is the systematic integration of the aforementioned automated damage data processing methods, damage inversion and rapid model mapping methods, and parameter correction methods into a unified framework, forming a virtual evaluation system capable of autonomously completing the entire process of "data acquisition - damage identification - model update - mechanical response and performance prediction." The core protection of this technical solution lies in: this integrated technical framework and its implemented integrated "monitoring-updating-prediction" operation mode.
[0020] For two typical damage forms, dispersed crack damage and delamination damage, an automated data processing flow from detection data to damage inversion has been achieved, providing fundamental support for the virtual evaluation technology of digital methods for composite material structures.
[0021] Current technology analysis: In current research on composite material damage identification using active monitoring methods such as Lamb waves, data processing often relies on manual intervention and experience-based judgment. This is particularly true for typical damage types with different mechanisms, such as dispersed cracks and delamination, where a standardized, fully automated data processing path is lacking. Existing methods mostly focus on offline analysis of specific damage types, making it difficult to automate the filtering, alignment, feature extraction, and damage parameter quantification of detection signals. This results in low data processing efficiency and poor consistency, failing to support the real-time and accuracy requirements of digital twin systems.
[0022] Advantages of this invention: This invention constructs standardized automated processing workflows for Lamb wave data, targeting both diffuse cracks and delamination damage. For diffuse crack damage, the damage severity is automatically quantified through filtering, waveform alignment, wave velocity calculation, and a stiffness degradation model. For delamination damage, wavelet packet feature extraction and an XGBoost machine learning model are combined to achieve intelligent inversion of damage contours and locations. Both workflows achieve end-to-end automated processing from raw signals to damage parameters, significantly improving processing efficiency and result consistency, and laying a reliable data foundation for subsequent rapid model updates and performance prediction.
[0023] For two typical damage inversion results, namely dispersed crack damage and delamination damage, a fast mapping method for digital twin models was established, which provides efficiency and accuracy assurance for the realization of digital method virtual evaluation technology for composite material structures.
[0024] Analysis of existing technologies: Current digital twin model updates largely rely on manual interpretation of damage data and manual adjustment of model parameters. This is particularly problematic for gradual changes in material properties caused by dispersed cracks and abrupt geometric changes caused by delamination, lacking efficient and accurate automated mapping mechanisms. Stiffness degradation from dispersed cracks often requires manual correction element-by-element, while geometric reconstruction of delamination damage depends on cumbersome modeling operations. This results in delayed and error-prone model updates, making it difficult to synchronize with the physical damage state, severely limiting the dynamic simulation and real-time evaluation capabilities of digital twins.
[0025] Advantages of this invention: This invention proposes programmed and automated rapid model mapping methods for the physical characteristics of two types of damage. For diffuse crack damage, sensor partitioning and damage factor matrices enable batch and accurate updates of stiffness parameters; for delamination damage, coordinate mapping and automatic node copying and replacement enable geometrically accurate reconstruction of delamination regions. Both methods automatically modify the model through algorithms, significantly improving update speed and accuracy, ensuring that the digital twin model can accurately reflect the structural damage state in real time, and providing a high-quality model foundation for efficient prediction of subsequent mechanical behavior.
[0026] The idea and method of initial digital twin model correction were proposed, and a framework for a digital twin virtual evaluation method of composite material structures was built, integrating automated processing of detection data, damage inversion and mapping, structural mechanical behavior and performance.
[0027] Analysis of existing technologies: Current digital twin research often directly uses standard material parameters to establish initial models, neglecting the performance dispersion caused by manufacturing processes, resulting in insufficient basic model accuracy. Furthermore, existing frameworks tend to focus on monitoring and diagnostic aspects, or only achieve partial functional integration, failing to form a complete technical closed loop covering "data acquisition - damage identification - model update - performance prediction - experimental verification." This limits the overall prediction confidence and engineering practical value of digital twin systems.
[0028] Advantages of this invention: This invention innovatively introduces an initial model parameter correction step, iteratively fine-tuning material parameters through "simulation-experiment" comparison, improving model accuracy from the source. Furthermore, it systematically integrates automated data processing, intelligent damage inversion, rapid model mapping, mechanical property prediction, and experimental verification into a unified framework, forming a fully automated, highly integrated digital twin virtual evaluation system. This framework not only achieves seamless integration from damage perception to performance prediction, but also significantly improves the credibility and practicality of prediction results through built-in correction and verification mechanisms, fully supporting the rapid and accurate assessment of the strength and functional reliability of composite material structures under damage conditions. Attached Figure Description
[0029] Figure 1 Initial digital twin model and boundary conditions for the composite material cylinder; Figure 2 Sensor arrangement and radial compressive fatigue loading of composite material cylinder; Figure 3 Composite material cylinder sensor array and damage factor matrix; Figure 4 Composite material cylinder region division and model update Figure 5 Deformation behavior and residual strength prediction of composite material cylinders; Figure 6 Initial digital twin model and boundary conditions of composite beam; Figure 7 Three-point bending test of composite beam; Figure 8 Layered damage inversion profile; Figure 9 The initial twin model of the composite beam is updated by introducing delamination damage; Figure 10 Flexural deformation behavior and flexural strength prediction of composite beams. Detailed Implementation
[0030] The present invention will be further described below with reference to the embodiments.
[0031] Example 1 A digital twin virtual evaluation method for the mechanical properties of composite material structures, comprising the following steps: Step 1: A composite cylinder was prepared using T300 / 9A16 carbon fiber reinforced resin-based unidirectional prepreg, with a layup sequence of [0 / 90 / 0 / 90]2s. Vacuum-assisted hot pressing was used to ensure the cylinder was dense and free of obvious manufacturing defects. Cylinder dimensions: wall thickness 1.0 mm, diameter 300 mm, height 300 mm. After molding, the cylinder was water-jet cut and trimmed to obtain a smooth surface and geometrically accurate sample. Material parameters, layup information, and process conditions were recorded during the preparation process to provide basic data for subsequent digital twin modeling.
[0032] Step 2: Based on the geometric dimensions, layup structure, and material parameters of the composite material cylinder, establish its initial digital twin model, such as... Figure 1 As shown in Table 1, the model is discretized using shell elements, and the material parameters are initially assigned based on supplier data. Radial compression experiments were conducted on the cylinder in its initial state to obtain the experimental load-displacement curves. Simultaneously, simulation calculations were performed using a digital twin model under the same boundary conditions to obtain the simulated load-displacement curves. By comparing the experimental and simulation results, material parameters such as the elastic modulus and shear modulus in the model were adjusted using parameter inversion methods (such as least squares fitting or optimization algorithms) until the curves showed good agreement. The material parameters are shown in Table 1. Finally, the corrected initial digital twin model A was obtained, ensuring that it can accurately predict the deformation behavior of the cylinder in an undamaged state.
[0033] Table 1 Material Parameters
[0034] Step 3: Arrange a 3×8 PZT sensor array on the surface of the composite material cylinder to construct a multipath Lamb wave detection network. Mount the composite material cylinder on a fatigue testing machine and apply a cyclic radial compressive load along a specified direction, such as... Figure 2 As shown. The peak load was set to 70% of the failure load of the cylinder (350 N), the loading frequency was 10 Hz, and a tension-unloading-tension fatigue mode was adopted. Multi-stage cyclic loading was applied to the cylinder, and the number of cycles and load history were recorded. The damage initiation and accumulation process was observed by real-time monitoring of cylinder deformation and surface condition.
[0035] Step 4: Using a self-built guided wave detection system, a 5-cycle sinusoidal excitation signal modulated by a Hanning window is applied at an excitation frequency of 150 kHz, and the received waveforms of each sensor path are acquired. By comparing the waveforms of the same path before and after damage, the Lamb wave velocity is calculated to obtain the wave velocity distribution data under damage (i.e., Lamb wave detection data A). This data reflects the degradation of elastic properties inside the material caused by fatigue damage.
[0036] Step 5: Calculate the degradation value of the material's equivalent elastic modulus along each detection path using the Lamb wave detection data A obtained in Step 4. Filter the Lamb wave detection data A to remove minor fluctuation interference. Align the filtered Lamb wave detection data A by aligning the received wave with the excitation wave's initial moment. Calculate the Lamb wave velocity using the aligned Lamb wave detection data A. Based on the Lamb wave velocity, invert the damage condition A, i.e., the stiffness of the composite material cylinder after degradation. The stiffness of the composite material cylinder after degradation is:
[0037] Based on the spatial distribution of the sensor network, the cylinder is divided into 96 minimum update units (8 circumferential regions × 3 axial regions × 2 layup directions), generating a damage factor matrix, such as... Figure 3 As shown in the figure, this matrix quantitatively describes the percentage decrease in modulus caused by fatigue damage in each region of the cylinder, thereby enabling visualization and quantitative characterization of the location, extent, and distribution of damage in the cylinder.
[0038] Step Six: Based on the damage factor matrix obtained in Step Five, map the modulus degradation information of each region to the initial digital twin model A using a self-developed program, and update the material properties of the corresponding elements (updated modulus = initial modulus × (1 - damage factor)), such as... Figure 4 As shown, using the updated digital twin model, the radial compressive deformation and residual strength of the damaged cylinder under different load directions are simulated and predicted, as follows. Figure 5 As shown, by comparing the prediction results of the updated model with the experimental data of the actual damaged cylinder, the effectiveness of the model update is verified, and virtual evaluation of composite material structures is realized within the digital twin framework.
[0039] Example 2 A digital twin virtual evaluation method for the mechanical properties of composite material structures, comprising the following steps: Step 1: Using T300 / 9A16 carbon fiber reinforced resin-based unidirectional prepreg, composite laminates were prepared according to the designed layup sequence (0 / 45 / 45 / 0). During the fabrication process, a polyvinyl fluoride film was implanted to induce delamination damage. Vacuum-assisted hot pressing was used to ensure the material was dense and free of significant porosity defects. After molding, the laminates were processed into standard rectangular cross-section straight-edge beam specimens with dimensions of 70 mm × 35 mm × 0.9 mm using a waterjet cutting system. The specimens had smooth surfaces, smooth edges, and accurate geometric dimensions, providing a consistent physical sample for subsequent bending experiments and digital twin modeling. Material parameters, layup information, and process conditions were recorded during the fabrication process as basic data for model input.
[0040] Step 2: Based on the geometry, layup sequence, and material system of the composite beam, establish its initial digital twin model, such as... Figure 6 As shown in Table 2, the model is discretized using shell elements (SC8R), and the material properties are defined based on the Hashin damage criterion. The elastic parameters of the materials are shown in Table 2, the damage initiation criterion parameters in Table 3, and the damage evolution criterion in Table 4. Three-point bending experiments were conducted on the beam specimens to obtain the experimental load-displacement curves. Simultaneously, simulation calculations were performed under the same boundary conditions (simply supported at both ends, centrally loaded) to obtain the simulated load-displacement curves. By comparing the experimental and simulation results, the elastic modulus and strength parameters in the model were adjusted using parametric inversion methods until the two curves matched perfectly. Finally, a corrected initial digital twin model B was obtained, ensuring that it could accurately predict the bending deformation and strength of the beam in the undamaged state.
[0041] Table 2 Material Elastic Parameters
[0042] Table 3 Parameters for the Injury Initiation Criteria
[0043] Table 3 Parameters of Damage Evolution Criteria
[0044] Step 3: Install the composite beam on a fatigue testing machine and perform three-point bending cyclic loading, such as... Figure 7As shown in the figure. The peak load was set to 70% of the beam's failure load, the loading frequency was 10 Hz, and a cyclic bending-unloading mode was adopted. Multi-stage fatigue loading was performed, and the load history and number of cycles were recorded. By monitoring the beam's deflection changes and surface condition in real time, the initiation and propagation behavior of delamination damage was observed, providing an experimental basis for subsequent damage detection and performance degradation analysis.
[0045] Step 4: Arrange a PZT sensor array on the surface of the composite beam to construct a multi-path Lamb wave detection network. Using a self-built guided wave detection system, apply a sinusoidal excitation signal modulated by a 150 kHz Hanning window and acquire the received waveforms of each sensor path. Calculate the Lamb wave eigenvalues using wavelet packet decomposition and reconstruction methods to obtain the effective eigenvalues under the damaged state (i.e., Lamb wave detection data B). This data reflects the material delamination damage caused by bending fatigue within the beam.
[0046] Step 5: Invert the location and size of delamination damage using the Lamb wave detection data B obtained in Step 4. Filter the Lamb wave detection data B to remove minor fluctuation interference. Align the filtered Lamb wave detection data B by aligning the received wave with the excitation wave start time. Perform wavelet packet decomposition and reconstruction on the aligned Lamb wave detection data B to obtain the signal features in the time domain, frequency domain, and time-frequency domain. Calculate the feature values and input them into the XGBoost machine learning model to invert the damage condition B, i.e., the delamination damage profile of the composite beam, as shown below. Figure 8 As shown, the spatial distribution of the sensor network is used to predict the layered damage profile.
[0047] Step Six: Based on the damage contour information obtained in Step Five, the layered damage contour is converted into coordinate form and stored. The program then filters and identifies units within the coordinate range of the layered damage contour in the digital twin model. Based on the coordinates of the remaining nodes in each unit, units sharing nodes between the upper and lower layers are distinguished. Nodes within the coordinate range of the layered damage contour are copied, and the shared nodes in the upper-layer units are replaced with the copied nodes. This achieves layered damage mapping and updating, introducing a layered interface (node separation), such as... Figure 9 As shown, using the updated digital twin model, the bending deformation and residual strength of the damaged beam under different load conditions are simulated and predicted, obtaining the updated load-displacement curves and failure modes, as shown. Figure 10 As shown, the effectiveness of the model update is verified by comparing the prediction results of the updated model with the experimental data of actual damaged beams, thus realizing the virtual evaluation of composite beams within a digital twin framework.
[0048] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A digital twin virtual evaluation method for the mechanical properties of composite material structures, characterized in that... The steps of this method include: Step 1: Fabricate the composite material cylinder and composite material beams; Step 2: Establish initial digital twin model A of composite material cylinder and initial digital twin model B of composite material beam, and perform parameter correction based on experimental results; Step 3: Perform radial compression fatigue loading on the composite material cylinder prepared in Step 1, and perform bending fatigue loading on the composite material beam prepared in Step 1. Step 4: Perform Lamb wave detection on the composite material cylinder and composite material beam after fatigue loading in Step 3, and obtain Lamb wave detection data A for the composite material cylinder and Lamb wave detection data B for the composite material beam. Step 5: Based on the Lamb wave detection data A obtained in Step 4, invert the damage status A of the composite material cylinder; based on the Lamb wave detection data B obtained in Step 4, invert the damage status B of the composite material beam. Step 6: Update the initial digital twin model A established in Step 2 based on the damage condition A obtained in Step 5; update the initial digital twin model B established in Step 2 based on the damage condition B obtained in Step 5; then virtually evaluate the mechanical behavior and properties of the composite material cylinder based on the updated digital twin model A; and virtually evaluate the mechanical behavior and properties of the composite material beam based on the updated digital twin model B.
2. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step one, when fabricating the composite beam, a polytetrafluoroethylene film is implanted between the layers.
3. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step two, the method for parameter correction based on the experimental results is as follows: The requirement is that, in the initial state, the deformation calculation results of the initial digital twin model A of the composite material cylinder are the same as the deformation experimental test results of the composite material cylinder made in step one; The requirement is that, in the initial state, the deformation calculation results of the initial digital twin model B of the composite beam are the same as the deformation test results of the composite beam fabricated in step one.
4. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step five, the method for inverting the damage condition A of the composite material cylinder based on the Lamb wave detection data A is as follows: Step 511: Filter the Lamb wave detection data A to remove minor fluctuation interference; Step 512: Align the filtered Lamb wave detection data A by aligning the received wave with the start time of the excitation wave; Step 513: Calculate the Lamb wave velocity using the aligned Lamb wave detection data A; Step 514: Based on the Lamb wave velocity inversion, damage condition A is obtained, i.e., the stiffness of the composite material cylinder after degradation. The stiffness of the composite material cylinder after degradation is: in, C P For Lamb wave velocity, f For the excitation frequency, E’ This refers to the stiffness after degradation.
5. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step five, the method for inverting the damage condition B of the composite beam based on the Lamb wave detection data B is as follows: Step 521: Filter the Lamb wave detection data B to remove minor fluctuation interference; Step 522: Align the filtered Lamb wave detection data B by aligning the received wave with the excitation wave start time; Step 523: Perform wavelet packet decomposition and reconstruction on the aligned Lamb wave detection data B to obtain the signal features in the time domain, frequency domain, and time-frequency domain. Then, calculate the signal features to obtain the feature values. Step 524: Input the obtained feature values into the XGBoost machine learning model to invert and obtain damage condition B, that is, the delamination damage profile of the composite beam.
6. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step six, the method for updating the initial digital twin model A established in step two based on the damage condition A obtained in step five is as follows: Step 611: Divide the digital twin model into partitions based on the sensor network layout; Step 612: Create a damage factor matrix based on the partitioning results; Step 613: Identify the correspondence between each partition, layer, and damage factor matrix through the program; Step 614: Update the parameters of the digital twin model using the inp file to achieve the update of the diffuse crack damage mapping.
7. The digital twin virtual evaluation method for the mechanical properties of composite material structures according to claim 1, characterized in that: In step six, the method for updating the initial digital twin model B established in step two based on the damage condition B obtained in step five is as follows: Step 621: Convert the layered damage contour into coordinate form and store it; Step 622: The program determines and filters the units in the digital twin model that are within the range of the layered damage contour coordinates; Step 623: Based on the coordinates of the remaining nodes of each unit, distinguish the units that share nodes in the upper and lower layers; Step 625: Copy the nodes within the coordinate range of the layered damage contour, and replace the common nodes in the upper unit with the copied nodes to realize the layered damage mapping update.