A Safety Assessment Method for Delamination Composite Materials Based on the Damage Non-Propagation Principle

By updating the ultimate load model of composite material structures using latent variable models and Bayesian frameworks, the problem of fusion of binary observation data is solved, enabling efficient and accurate damage propagation assessment and supporting safety assessment and maintenance decisions for composite material structures.

CN122091032APending Publication Date: 2026-05-26BEIHANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610060331.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-16
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively integrate binary observation data on delamination damage in composite material structures, resulting in insufficient accuracy in ultimate load prediction and reliability in safety assessment. Traditional methods are computationally inefficient and difficult to adapt to the real-time assessment needs of complex structures.

Method used

A latent variable model is used to associate binary observation data with continuous model parameters. The load difference is mapped to the damage propagation probability through the Probit link function. The model parameters are updated by combining the Bayesian framework and the Markov chain Monte Carlo method, so as to realize the dynamic correction of the ultimate load model.

Benefits of technology

It can reduce parameter uncertainty without real-time damage monitoring, improve the accuracy of ultimate load prediction, enhance the accuracy and engineering applicability of composite structure safety assessment, and support maintenance decisions for risky bases.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122091032A_ABST
    Figure CN122091032A_ABST
Patent Text Reader

Abstract

This invention proposes a safety assessment method for delaminated composite materials based on the principle of damage non-propagation, belonging to the fields of composite material delaminated damage modeling and composite structure safety assessment. The method includes: constructing ultimate load envelopes that meet the damage non-propagation requirement under different damage sizes; for readily available binary observation data in practical engineering, i.e., whether damage propagates under specific damage sizes and load conditions, connecting discrete observations with a continuous parameter space through a latent variable model; using the Probit link function to map the predicted difference between the load and the ultimate load into a damage propagation probability; and constructing a likelihood model using a Bernoulli likelihood function; and optimizing the ultimate load envelope within a Bayesian framework. This invention can effectively support safety assessment and maintenance decisions for composite material structures in aerospace and other fields.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of composite material delamination damage modeling and composite material structural safety assessment, specifically involving a safety assessment method for delamination-containing composite materials based on the principle of damage non-propagation. Background Technology

[0002] Since the 1990s, composite materials have been widely used in the aerospace field due to their excellent specific strength and specific stiffness characteristics. However, delamination damage, as one of the most common damage forms in composite structures, seriously threatens structural safety. In current engineering practice, "damage non-propagation" is the core criterion for the safety assessment of composite structures, requiring that delamination damage not propagate under service loads. Traditional delamination damage assessment methods mainly rely on refined modeling based on finite element methods, which are computationally inefficient and difficult to adapt to the real-time assessment needs of complex structures. In addition, the dispersion of composite material properties, manufacturing defects, and the uncertainty of the load environment make it difficult for traditional deterministic assessment methods to accurately reflect the true safety state of the structure. In actual engineering, binary observation data is more readily available, i.e., the result of whether damage propagates under specific damage size and load conditions. Existing assessment methods cannot effectively integrate such data, resulting in significant uncertainties in model parameters, affecting the accuracy of ultimate load prediction and the reliability of safety assessment. There is an urgent need for an efficient assessment method that integrates theoretical models and measured data. Summary of the Invention

[0003] To address the challenge of obtaining continuous observation data due to the high cost of real-time monitoring in safety assessments of composite structures with delamination damage, this invention provides a safety assessment method for delamination-containing composite materials based on the principle of damage non-propagation. This invention is primarily applied to the safety assessment and maintenance decisions of composite material structures in aerospace vehicles, specifically including: rapid assessment of the service safety of structures with delamination damage to determine whether current loads will lead to damage propagation; damage tolerance analysis during the new aircraft design phase to optimize structural design to meet the "damage non-propagation" requirement; and structural damage diagnosis during the operation and maintenance phase, updating the safety assessment model based on periodic inspection data.

[0004] This invention associates discrete binary observation data with a continuous model parameter space through a latent variable model. The latent variable is defined as the difference between the actual load and the predicted ultimate load. The Probit link function maps this difference to the damage propagation probability, and a Bernoulli likelihood function is constructed to describe the probability distribution of the binary observations. Within a Bayesian framework, combining the prior distribution of parameters with the binary observation data, the posterior distribution of model parameters is updated using Markov Chain Monte Carlo (MCMC) methods (such as NUTS (Non-Steering Hamiltonian Monte Carlo Sampling) sampling), achieving dynamic correction of the ultimate load model. This invention overcomes the dependence on continuous observation data, reducing parameter uncertainty and improving the accuracy of ultimate load prediction under different damage sizes using only readily available binary inspection results in engineering. This provides more practical technical support for structural safety assessment and maintenance decisions. It solves the problem of adapting the data types of inspections in engineering to traditional assessment models, broadens data sources, and improves the engineering applicability of the assessment method. A complete uncertainty quantification process is established from setting the prior distribution of model parameters to posterior inference, and the mean and variance of the posterior distribution characterize the credibility of the assessment results. This invention provides a quantitative uncertainty index for structural safety assessment, supporting risk-based maintenance decision-making.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A safety assessment method for delaminated composite materials based on the principle of damage non-propagation includes the following steps:

[0007] Step 1: Based on the fracture toughness model obtained from plate and shell simulation, and according to the principle of damage non-propagation, construct the layered damage non-propagation load envelope;

[0008] Step 2: Collect the inspection results of structural operation and maintenance in actual engineering projects to form a binary observation dataset; introduce latent variables to characterize the difference between the observed load values ​​and the actual ultimate load, define the observation data, and map the load difference to the damage propagation probability through the Probit function; construct the Bernoulli likelihood function, combine it with the prior distribution of parameters, and extract samples from the posterior distribution through the Markov chain Monte Carlo method to update the mean and standard deviation of parameters and reduce uncertainty; achieve iterative optimization of the ultimate load envelope through Bayesian parameter update, and provide a model parameter correction method that can fuse the ultimate load observation values ​​or binary observation data on whether damage propagates under specific load conditions.

[0009] Step 3: Calculate the structural ultimate load under a given damage size based on the modified model, predict the ultimate load envelope under different damage sizes, and compare it with the actual service load to determine the structural safety status.

[0010] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described safety assessment method for layered composite materials based on the principle of non-damage propagation.

[0011] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described safety assessment method for layered composite materials based on the principle of non-damage propagation.

[0012] Beneficial effects:

[0013] This invention constructs a hierarchical damage non-proliferation assessment framework based on fracture mechanics and Bayesian updating. By establishing a damage size-ultimate load mapping model and a model parameter correction method based on binary observation data, it associates binary data with the ultimate load model through a Bayesian framework, dynamically updating model parameters and reducing uncertainty. Its application benefits include eliminating the need for real-time damage and load monitoring equipment, resolving the contradiction between computational efficiency and accuracy in traditional methods, and achieving rapid assessment through a reduced-order model. It effectively handles uncertainties in composite material structure assessment, improving the reliability of safety assessments. Utilizing only readily available binary detection data from engineering projects, it can optimize ultimate load prediction under different damage sizes, improving the accuracy and engineering practicality of composite material structure safety assessments, providing a reliable basis for maintenance decisions for equipment such as aircraft, and enhancing the engineering applicability of assessment methods. Attached Figure Description

[0014] Figure 1 Schematic diagram of braided laminate layup dimensions;

[0015] Figure 2 A schematic diagram of the layered test and equivalent plate-shell simulation model of the double cantilever beam;

[0016] Figure 3 This is a schematic diagram of the ultimate load model parameters based on binary observation data; where (a) is... The updated result diagram is shown in (b). Update result illustration;

[0017] Figure 4 This is a schematic diagram of the prediction results of the ultimate load-damage size curve updated based on binary observation data.

[0018] Figure 5 This is a flowchart of a safety assessment method for delaminated composite materials based on the principle of non-damage propagation, according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0020] like Figure 5 As shown, the safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to the present invention includes the following steps:

[0021] Step 1: Based on the fracture toughness model obtained from plate and shell simulation, and according to the principle of damage non-propagation, construct the layered damage non-propagation load envelope.

[0022] Step 2: Collect inspection results of structural operation and maintenance in actual engineering projects to form a binary observation dataset; introduce latent variables to characterize the difference between observed load values ​​and the true ultimate load, define the observation data, and map the load difference to the damage propagation probability through the Probit function. Construct a Bernoulli likelihood function, combine it with the prior distribution of parameters, and extract samples from the posterior distribution using the MCMC method to update the parameter mean and standard deviation, thereby reducing uncertainty; achieve iterative optimization of the ultimate load envelope through Bayesian parameter updates, and provide a model parameter correction method that can integrate binary observation data on whether damage propagates under specific load conditions or the ultimate load observation values ​​when damage propagates.

[0023] Step 3: Calculate the structural ultimate load under a given damage size based on the modified model, predict the ultimate load envelope under different damage sizes, and compare it with the actual service load to determine the structural safety status.

[0024] Specifically, step 1 includes:

[0025] To meet the requirements of rapid structural strength verification based directly on the local loading state of the structure in practical engineering applications, this section considers providing a lightweight model for evaluating whether structures under different damage sizes meet the requirement of preventing further damage propagation by giving the load envelope.

[0026] Generally speaking, in fracture mechanics, whether delamination damage propagates depends on whether the energy release rate under the current damage and load conditions reaches the damage propagation threshold, i.e., it requires:

[0027] (1)

[0028] in, This is the energy release rate threshold for damage propagation, i.e., fracture toughness; This is a reduced-order model for calculating energy release rate, representing a given load. and delamination damage size Below, the energy release rate of damage propagation. Based on this formula, the ultimate load that meets the damage non-propagation criterion under different damage sizes should satisfy:

[0029] (2)

[0030] in, The damage propagation energy release rate under unit external load is obtained by fitting a shell element-based reduced-order model, reflecting the magnitude of the energy release rate caused by unit load under different damage sizes.

[0031] The relationship between the energy release rate threshold and the damage size during the delamination propagation process of the composite material specimen satisfies:

[0032] (3)

[0033] This formula considers the instability at the beginning of damage propagation. Wherein, This represents the initial fracture toughness corresponding to the initial damage state; while This indicates the fracture toughness during the stable propagation stage; Current damage size Compared with the initial damage size The difference, i.e., the amount of damage expansion; To achieve the damage propagation amount at the stable stage of damage propagation, This is an empirical coefficient.

[0034] Based on the above formula, the following ultimate load prediction model can be given:

[0035] (4)

[0036] in, The ultimate load refers to the load at the current damage size. The maximum load that the structure can withstand when it meets the requirement of "damage not propagating".

[0037] Specifically, step 2 includes:

[0038] In actual aircraft structures, real-time monitoring of the size of delamination damage and the magnitude of the loads it experiences is typically extremely difficult. Even with sophisticated monitoring equipment, it is difficult to directly obtain the specific load values ​​that would lead to delamination expansion at a given damage size.

[0039] Considering the available measured data in practical engineering applications, which are generally obtained through ground-based structural damage detection methods, this data assesses whether damage has propagated under a specific load at a given damage size. Therefore, under these conditions, the observational data applicable to model parameter correction is essentially binary data indicating whether damage has propagated or not. To meet this requirement, it is necessary to explore model parameter updating methods based on this type of binary observational data, building upon the aforementioned research.

[0040] In this problem, to obtain the ultimate load that satisfies the condition of no further damage propagation under different damage sizes, the observed data are as follows:

[0041] (5)

[0042] in, The dataset representing the observed samples, To observe the number of samples, For the first The secondary observation input is the load magnitude borne by the structure at a specific damage size. The output is the result of an observation, which shows whether the damage has progressed under these conditions. .

[0043] The core of solving this problem of updating model parameters based on binary comparative observation data using Bayesian theory lies in mapping the difference between continuous predicted values ​​and true values ​​to the probability of binary observations. The overall idea is to construct a probabilistic model that correlates the binary observation results with a nonlinear function of the parameters. In this problem, a Bernoulli distribution is constructed, and the difference between model predictions and true values ​​is converted into probabilities using the Probit link function.

[0044] Specifically, we first assume that the observed bivariate results are determined by latent continuous variables, and then use a latent variable model to correlate the continuous parameter inferences with the discrete observations. Let the true limit load, which considers random bias, be... satisfy:

[0045] (6)

[0046] in, The ultimate load value is the model prediction obtained based on the current parameter distribution. It is Gaussian noise, with a mean of 0 and a variance of 1. normal distribution This is used to simulate the uncertainty of the damage propagation threshold value.

[0047] For any set of observation inputs ,in Given the current damage size, This represents the actual load magnitude. The latent variable is defined as... To satisfy them:

[0048] (7)

[0049] Based on the above equation, the relationship between latent variables and observations can be intuitively given, and the prediction output given by the prediction input model satisfies:

[0050] (8)

[0051] in, Given the current damage size, This refers to the actual load it bears. The model's prediction results, Predict damage propagation in real time. The damage is predicted not to spread. As a latent variable, that is, the difference between the observed load and the actual ultimate load, when When the observed load exceeds the actual ultimate load, damage propagation occurs, therefore the model predicts... Conversely, the model predicts... The damage does not spread.

[0052] Probability distribution of predicted output satisfy:

[0053] (9)

[0054] in, This is the cumulative distribution function (CDF) of the standard normal distribution obtained after standardizing the noise term, also known as the Probit link function. It can be seen that the predicted output... The generation process is essentially a comparison and A deterministic event. Based on the definition of the latent variable model above, this is equivalent to a deterministic comparison. and Introduce a small amount of random noise. This makes the observation results become ( Indicates an indicator function. Parameters Physically equivalent to observation error, controlling the "degree of ambiguity" in the comparison: when When the model degenerates into a deterministic comparison; when At that time, the model allows in near The existence of uncertainty at the time avoids the gradient vanishing problem in numerical calculations.

[0055] In fact, besides the Probit function, Logit is also a commonly used cumulative distribution function. The former assumes that the noise follows a normal distribution and uses the standard normal cumulative distribution function (CDF). This is suitable for noise models with clearly defined physical processes. The latter uses the Logistic function. Assuming the noise follows a Logistic distribution, the computation is more efficient but the interpretability is slightly weaker.

[0056] In defining the likelihood function, unlike problems directly based on deterministic continuous observation data, this problem requires the application of the Bernoulli likelihood function. That is, it assumes the binary observation results... It follows a Bernoulli distribution, and its probability is given by the above derivation:

[0057] (10)

[0058] in This represents the Bernoulli distribution, a discrete probability distribution used to describe trials with only two mutually exclusive outcomes, and corresponds to the Bernoulli distribution in formula (10). Time damage propagation and The damage did not extend to either of the two mutually exclusive outcomes.

[0059] Based on this likelihood function, according to Bayes' theorem, the posterior distribution of the parameters is:

[0060] (11)

[0061] In the formula This is the posterior probability distribution, i.e., when obtaining binary observation results. Then, model parameters The probability distribution. It is the likelihood function, that is, given model parameters Under these conditions, a binary result was observed. The probability of. This is a prior probability distribution, meaning that when acquiring observation data... Previously, regarding model parameters The initial probability distribution.

[0062] The likelihood function satisfies:

[0063] (12)

[0064] Since there are multiple observation data, the formula uses... This represents the data index, which identifies the number of observations, where N is the number of observation samples.

[0065] The parameters are sampled from the posterior distribution using the NUTS (No-U-Turn Sampler) method, and the step size and path length are adaptively adjusted. The parameter space is explored step by step by calculating the acceptance probability, and the joint posterior distribution of the parameters is calculated based on the posterior samples.

[0066] In summary, the core of this method lies in connecting discrete observations with a continuous parameter space through latent variables, mapping continuous prediction discrepancies to binary observation probabilities, and using the Probit link function to map linear predictions to probabilities. This enables the model to perform parameter inference based on binary observation data within a Bayesian framework. The key steps in the specific implementation process include: calculating predicted values ​​based on the observed data input. According to its relationship with The difference is used to calculate the Probit probability, evaluate the Bernoulli likelihood function, and calculate the posterior distribution based on the Markov Chain Monte Carlo (MCMC) method.

[0067] During this process, parameters The introduction of parameters further characterizes the uncertainty of the comparison process, enabling the model to flexibly adapt to observational data with different noise levels. If the posterior approaches zero, it indicates a high degree of certainty in the observation; conversely, a larger posterior indicates a lower degree of certainty. This may reveal model mismatch or insufficient modeling of noise. This framework transforms the hard constraints of binary comparisons into soft constraints of probabilistic form, enabling Bayesian inference to be performed in a continuous parameter space while preserving the intuitive interpretation of the physical process.

[0068] Example:

[0069] In this embodiment, Example 1 is a double cantilever beam (DCB) delamination propagation test. The specimens used in this embodiment are made of woven prepreg (G0814 / 913), and the material is T300 carbon fiber and epoxy resin matrix. Figure 1 As shown, 'e' represents the orthogonal coordinate system direction of the composite laminate. The corresponding fiber longitudinal direction (principal direction) of the composite material corresponds to the "listed in Table 1" This refers to the elastic modulus in that direction; Corresponding to the in-plane transverse direction of the composite material (perpendicular to) (in-plane direction), Table 1 "E" 22 “G” 12 (In-plane shear modulus) is related to this direction and reflects the mechanical properties of the material in the plane perpendicular to the fiber direction; Corresponding to the thickness direction of the composite material (the direction perpendicular to the plane of the laminate), Table 1's "E"33 “G” 31 “G” 32 "(Out-of-plane shear modulus) refers to this direction and is directly related to the propagation of delamination damage (interlaminar separation along the thickness direction). 'e' is used to identify the spatial reference axis of the material to describe the weave direction and anisotropic properties of different plies. It represents a multi-directional arrangement of plies, with the weave directions of each layer alternating." / and The total number of ply layers is 15. The delamination is located between the seventh and eighth layers of the laminate, and the upper and lower sides of the delamination interface have different ply directions: the weave direction of the upper layer material is... / The yarn direction of the lower layer material is . Figure 1 Purple and red represent two different weave directions. Purple layers represent layers with a weave direction of 0° / 90°, while red layers represent layers with a weave direction of +45° / -45°.

[0070] The equivalent material properties of the undelaminated region of the sample and the upper and lower delaminated laminates are shown in Table 1:

[0071] Table 1. Equivalent stiffness performance parameters of layered and unlayered regions

[0072]

[0073] Table 1 They are respectively , , Elastic modulus in the direction. for - In-plane shear modulus of a plane for - Out-of-plane shear modulus of a plane for - In-plane shear modulus of a plane. Poisson's ratio in plane refers to the ratio of a material's surface area under a given surface. When subjected to tensile / compressive loads in the direction, Lateral strain generated by the direction and The ratio of axial strain in two directions reflects the coupling relationship between deformation in two directions within the plane and is a key parameter for calculating the stress and deformation within the structure's plane. and The out-of-plane Poisson's ratio refers to the material's... When subjected to directional load, lateral strain and The ratio of axial strain in the direction and in When subjected to directional load, lateral strain and The ratio of axial strain to axial strain.

[0074] The sample's length, width, and thickness are L = 200 mm, b = 25 mm, and 2h = 3.7 mm, respectively. Based on the opening displacement applied by the hinge, the initial delamination length is... This is the distance from the right end of the hinge to the delamination front. In model building, the specimen is also modeled starting from the right side of the hinge, with an initial delamination length. The experimental setup and simulation model are as follows: Figure 2 As shown, In the length direction, In the width direction, The thickness direction is consistent with the interlayer direction. Modeling method (1) uses single-layer shell elements to model the undamaged region, and models the damaged region according to the number of layers. "Hard contact" is applied between each layer to prevent mutual penetration. At the layer boundary, the shell element boundary of each layered region is bound to the element boundary of the unlayered region. The nodes on both sides of the constraint are in close contact and have the same translational and rotational displacements, thus simulating the connection relationship between the layered and unlayered regions and ensuring deformation continuity. Modeling method (2) uses multi-layer shell elements after layering to model the undamaged region, instead of single-layer shell elements with undamaged cross-sectional properties. The properties between the elements of each layer in the layered damaged region are defined based on "hard contact", allowing slippage and separation. The elements of each layer in the undamaged region are bound, but only the translational displacement is restricted to be the same, not the rotational degree of freedom. Compared with method 1, the deformation restriction of the local undamaged region connected at the end of the damaged region is relaxed. Parametric simulation is carried out to solve the relationship between the opening displacement and the layer size under a fixed unit load.

[0075] The above method is applied and verified considering the delamination areas of the composite material. In this embodiment, the initial fracture toughness is assumed to be... Fracture toughness during the stable propagation phase The parameters to be updated are of primary concern, and their initial prior distributions and other deterministic parameters remain unchanged. Only the observed data are modified, and the true ultimate load envelope is calculated using the aforementioned technique. Random sampling is then performed around this envelope for different damage sizes and loads. Here, considering that the actual load values ​​obtained are not limited to the vicinity of the threshold value, the sampling variance needs to be relatively large. Based on the comparison between the sampled loads and the true ultimate loads, the observed data is used for model updates.

[0076] Parameters are corrected using a Bayesian approach and through iterative iteration. Probit probabilities and Bernoulli likelihoods are calculated based on observed data, and the posterior distribution of the parameters is calculated using NUTS sampling. The parameter update results are as follows: Figure 3 As shown, observation group 0 (prior) represents the group of predicted / simulated results obtained based on prior knowledge. Observation group 1 (posterior), observation group 2 (posterior), and observation group 3 (posterior) represent different groups of predicted / simulated results obtained after posterior correction using actual observation data. The true value represents the actual observed results used to verify the accuracy of the model's predictions. N represents the number of observation samples, that is, the total number of all observation data used for model prior analysis and posterior correction, which is the basic data scale for conducting statistical inference and model parameter correction. Figure 3 (a) is a parameter The updated result image. This represents the initial fracture toughness, corresponding to the initial damage state, and mainly controls the ultimate load magnitude before damage propagates to the stable stage. The figure illustrates this through probability density curves. The distribution changes after fusing binary observation data show that, with the introduction of observation data, the peak probability density gradually approaches the true value, and the distribution width gradually decreases, indicating that the uncertainty of the parameters is effectively controlled. Figure 3 (b) is a parameter The updated result image. This represents the fracture toughness during the stable propagation stage, i.e., the fracture toughness when the damage enters the stable propagation stage. The figure also illustrates this using probability density curves. The distribution changes after fusing binary observation data show that the peak probability density gradually approaches the true value and the distribution width narrows, indicating that the uncertainty of the parameters is reduced.

[0077] The ultimate load-damage size results predicted based on the successively updated model parameters are as follows: Figure 4 As shown, where This indicates the state where a binary indicator variable is 0, typically representing a result of a binary event such as damage not propagating. This indicates the state where a binary indicator variable is 1, typically representing another outcome of binary events such as damage expansion, and is the true... The model represents the true probability distribution. The prior mean represents the mean of the probability distribution obtained based on prior knowledge when the model does not incorporate actual observation data. The prior 95% confidence interval represents the interval under the prior distribution with a 95% probability of containing the true value, reflecting the uncertainty range of the prior knowledge. The posterior 1 mean and posterior 1 95% confidence interval represent the probability distribution mean and 95% confidence interval obtained after the first incorporation of observation data and Bayesian correction. The posterior 2 mean and posterior 2 95% confidence interval represent the updated posterior distribution mean and 95% confidence interval obtained after the second incorporation of observation data. The posterior 3 mean and posterior 3 95% confidence interval represent the updated posterior distribution mean and 95% confidence interval obtained after the third incorporation of observation data. Posterior 1, Posterior 2, and Posterior 3 are the results of Bayesian posterior correction of the model parameters. The core principle is to update the prior distribution using Bayes' theorem by incorporating actual observation data, resulting in a more realistic posterior distribution. These are the probability distribution results after multiple updates of the model based on observation data, representing the gradual refinement process from prior to posterior. Figure 4 It can be seen that, compared to directly modifying the model based on damage propagation limit load observation data, updating model parameters based solely on binary comparative observation data indicating whether damage propagation has occurred requires a relatively larger amount of data to achieve the same result as before. However, this type of data is readily available in practical engineering applications, eliminating the need for online real-time damage and load monitoring equipment, and is based on data that is easily obtained from general structural maintenance results. Therefore, this method still has practical application value in terms of modifying model parameters based on measured data and guiding structural safety assessments.

[0078] In summary, this invention, based on plate and shell element simulation, provides the ultimate load envelope for different damage sizes. Furthermore, it presents a model parameter correction method that can fuse ultimate load observations during damage propagation or binary observation data on whether damage propagates under specific load conditions, supporting efficient structural safety assessment in practical engineering applications.

[0079] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method for safety assessment of layered composite material structures based on the principle of non-damage propagation.

[0080] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for safety assessment of layered composite material structures based on the principle of non-damage propagation.

[0081] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0082] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0085] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0086] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A safety assessment method for delaminated composite materials based on the principle of non-damage propagation, characterized in that, Includes the following steps: Step 1: Based on the fracture toughness model obtained from plate and shell simulation, and according to the principle of damage non-propagation, construct the layered damage non-propagation load envelope; Step 2: Collect the inspection results of structural operation and maintenance in actual engineering projects to form a binary observation dataset; introduce latent variables to characterize the difference between the observed load values ​​and the actual ultimate load, define the observation data, and map the load difference to the damage propagation probability through the Probit function; construct the Bernoulli likelihood function, combine it with the prior distribution of parameters, and extract samples from the posterior distribution through the Markov chain Monte Carlo method to update the mean and standard deviation of parameters and reduce uncertainty; achieve iterative optimization of the ultimate load envelope through Bayesian parameter update, and provide a model parameter correction method that can fuse the ultimate load observation values ​​or binary observation data on whether damage propagates under specific load conditions. Step 3: Calculate the structural ultimate load under a given damage size based on the modified model, predict the ultimate load envelope under different damage sizes, and compare it with the actual service load to determine the structural safety status.

2. The safety assessment method for delaminated composite materials based on the principle of non-damage propagation as described in claim 1, characterized in that, In step 1, a structural simulation model with delamination damage is first established using plate and shell elements. Then, the energy release rate corresponding to different damage sizes is calculated using a fracture toughness model. Combining the principle of no damage propagation, the energy release rate is compared with the fracture toughness to obtain the ultimate load under each damage size, thereby constructing a complete load envelope for delamination damage without propagation.

3. The safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 1, characterized in that, In step 2, the binary observation data obtained from the actual inspection is first organized into a triple of "damage size - load - whether it expands". Then, the discrete observations and the continuous parameter space are connected by latent variables. The difference between the load and the ultimate load is mapped to the damage expansion probability through the Probit function. Finally, the probability distribution of the binary observations is described by the Bernoulli likelihood function, thereby realizing the correlation between the observation data and the model parameters.

4. The safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 3, characterized in that, Within the Bayesian framework, a prior distribution is first set for the model parameters. Then, the Markov chain Monte Carlo method is used to extract samples from the posterior distribution. The mean and standard deviation of the parameters are updated through the samples, thereby reducing the parameter uncertainty and iteratively correcting the limit load envelope.

5. A safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 4, characterized in that, The Markov chain Monte Carlo method uses a non-steering Hamiltonian Monte Carlo sampler for sampling. In each iteration, the step size and path length are adaptively adjusted. The parameter space is explored step by step according to the acceptance probability to obtain stable posterior samples, which are used to calculate the updated limit load envelope.

6. The safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 1, characterized in that, In step 3, the updated ultimate load envelope is used to calculate the structural ultimate load for a given damage size. The predicted result is then compared with the actual service load. If the service load is less than the ultimate load, the structure is determined to be in a safe state; otherwise, there is a risk of damage propagation.

7. The safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 1, characterized in that, The binary observation data comes from the ground inspection results during the structural operation and maintenance phase. It can be obtained without real-time monitoring equipment and is directly used for updating the model parameters in step 2.

8. A safety assessment method for delaminated composite materials based on the principle of non-damage propagation according to claim 7, characterized in that, It is applicable to safety assessment of composite material structures in service, damage tolerance analysis in the new aircraft design phase, and structural damage diagnosis in the operation and maintenance phase. By iteratively updating the ultimate load envelope, it supports maintenance decisions.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of a safety assessment method for layered composite materials based on the principle of non-damage propagation as described in any one of claims 1 to 8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a safety assessment method for layered composite materials based on the principle of non-damage propagation as described in any one of claims 1 to 8.