Fatigue stiffness degradation prediction method for composite material with wrinkle defect

Abstract

The invention discloses a fatigue rigidity degradation prediction method for a composite material with wrinkle defects, relates to the technical field of composite material fatigue rigidity prediction, and aims to solve the technical problems that an existing rigidity degradation model is prone to distortion, the prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and requirements cannot be met. The fatigue rigidity degradation prediction method comprises the following steps: acquiring a laminated plate of a composite material, and determining a characteristic value representing the wrinkle defect degree of the laminated plate; obtaining the residual stiffness of the laminated plates with different wrinkle defect degrees under different stress levels, and taking the residual stiffness as a data set; reading a data set by using a BP neural network model; optimizing the BP neural network model by using a particle swarm algorithm; training the optimized BP neural network model, and obtaining a residual fatigue stiffness prediction model after a training termination condition is satisfied; and obtaining a residual fatigue stiffness degradation curve by using the residual fatigue stiffness prediction model, and completing fatigue stiffness degradation prediction.

Description

Technical Field

[0001] The invention relates to the technical field of fatigue stiffness prediction of composite materials, and more specifically to a method for predicting fatigue stiffness degradation of composite materials containing wrinkle defects. Background Art

[0002] Fiber wrinkles are common defects in fiber-reinforced composite laminates during the manufacturing process, especially in composite rotor structures of UAVs and composite wind turbine rotor blades. Wrinkle defects can significantly reduce the structural integrity and load-bearing capacity of the composite materials. When composite laminates containing wrinkle defects are located in the main load-bearing components of the structure, cracks are very likely to occur around the wrinkle defect area due to stress concentration in the wrinkle defect area.

[0003] Wrinkle defects in the composite rotor structure of unmanned aerial vehicles will significantly reduce the load-bearing stiffness. Wrinkle defects in the composite blades of wind turbines will cause the blades to break directly during service. UAV rotors and wind turbine composite blades are always subjected to fatigue loads of load cycles during service. Therefore, it is of great significance to study the fatigue performance of composite structures containing fiber wrinkle defects under fatigue loads, especially the degradation of stiffness performance. In the prior art, the stiffness degradation model is prone to distortion in the process of fitting the stiffness degradation law of composite materials under different stress levels. The prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and it cannot meet the stiffness degradation prediction requirements for wrinkle defect materials. Summary of the invention

[0004] The purpose of the present invention is to provide a method for predicting fatigue stiffness degradation of composite materials containing wrinkle defects, which is used to solve the technical problem that the existing stiffness degradation model is prone to distortion in the process of fitting the stiffness degradation law of composite materials under different stress levels, and the prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and it cannot meet the requirements for predicting stiffness degradation of wrinkle defect materials. In view of this, the present invention is implemented through the following scheme.

[0005] A fatigue stiffness degradation prediction method for a composite material containing wrinkle defects, comprising: obtaining a laminate of the composite material; Determining a characteristic value representing the degree of wrinkle defects of the laminate; Obtain the residual stiffness of laminates with different wrinkle defect degrees at different stress levels and use them as data sets; Using a BP neural network model to read the data set; Utilizing particle swarm algorithm to optimize the BP neural network model; The optimized BP neural network model is trained, and the residual fatigue stiffness prediction model is obtained after the training termination conditions are met; The residual fatigue stiffness prediction model is used to obtain a residual fatigue stiffness degradation curve according to characteristic parameters of the laminate containing wrinkle defects, thereby completing fatigue stiffness degradation prediction; the characteristic parameters include the characteristic values.

[0006] Compared with the prior art, in the fatigue stiffness degradation prediction method of composite materials containing wrinkle defects of the present invention, by determining the characteristic value characterizing the wrinkle defect degree of the laminate, and obtaining the residual stiffness of laminates with different wrinkle defect degrees at different stress levels, as a data set, the data set is read through the BP neural network model, and the BP neural network model is optimized by using the particle swarm algorithm, and then the optimized BP neural network model is trained to obtain a residual fatigue stiffness prediction model suitable for the composite materials with wrinkle defects of the present invention; and then the residual fatigue stiffness prediction model is used to obtain the residual fatigue stiffness degradation curve according to the characteristic parameters to complete the fatigue stiffness degradation prediction of the composite materials with wrinkle defects. Through the above technical solution of the present invention, the technical problems that the existing stiffness degradation model is prone to distortion in the process of fitting the stiffness degradation law of composite materials under different stress levels, the prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and the stiffness degradation prediction requirements of wrinkle defect materials cannot be met are solved.

[0007] Furthermore, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the characteristic value is the ratio of the height to the width of the wrinkle defect.

[0008] Furthermore, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the step of obtaining the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels comprises: Obtain the static strength of laminates with different wrinkle defect levels; determining the stress level according to the static strength; The laminates are subjected to tension-tension fatigue tests to determine the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels.

[0009] Furthermore, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the optimization of the BP neural network model by using a particle swarm algorithm comprises: Determine the characteristic parameters, and use the characteristic parameters as input variables of the BP neural network model; Using the particle swarm algorithm, the random position and velocity of the particle swarm are initialized, and the number of particles, spatial dimension, number of iterations, inertia factor, and learning factor are determined, and iteration is performed; In each iteration, the individual extreme value and the global extreme value are updated. After the iteration is completed, the random position and speed of the particle swarm are updated, and the optimal neural network parameters are decoded and output; The optimal neural network parameters are input into the BP neural network model to optimize the BP neural network model.

[0010] Furthermore, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the optimized BP neural network model is trained to obtain the residual fatigue stiffness prediction model after the training termination condition is met, including: The optimized BP neural network model is trained through the MATLAB neural network toolbox; During the training process, when the mean square error and the number of iterations meet the termination conditions, the residual fatigue stiffness prediction model is output.

[0011] Furthermore, in the fatigue stiffness degradation prediction method for composite materials containing wrinkle defects of the present invention, the termination condition is that the mean square error is less than or equal to 0.002, or the number of iterations is less than or equal to 1000 times.

[0012] Further, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the characteristic parameters are characteristic parameters of the laminate, and the characteristic parameters include characteristic values, width, thickness, stress level, and normalized fatigue loading times; The characteristic value is the ratio of the height to the width of the wrinkle defect, and the normalized fatigue loading number is determined by a tension-tension fatigue test.

[0013] Furthermore, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, the residual fatigue stiffness prediction model is used to obtain the residual fatigue stiffness degradation curve according to the characteristic parameters of the laminate containing wrinkle defects, including: The residual fatigue stiffness prediction model is used, the characteristic parameters are taken as input, and the output is a residual fatigue stiffness degradation curve.

[0014] Furthermore, in the fatigue stiffness degradation prediction method for composite materials containing wrinkle defects of the present invention, the characteristic value ranges from 0 to 5, and the residual fatigue stiffness degradation curve output by the residual fatigue stiffness prediction model forms a stiffness degradation envelope surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 Schematic diagram of the process of predicting fatigue stiffness degradation of composite materials containing wrinkle defects in Example 3 of the present invention; Figure 2It is the stiffness degradation curve of the intact specimen and the specimens with different fold aspect ratios at 80% stress level in Example 3 of the present invention; Figure 3 It is the stiffness degradation curve of the intact specimen and the specimens with different fold aspect ratios at 70% stress level in Example 3 of the present invention; Figure 4 It is the stiffness degradation curve of the intact specimen and the specimens with different fold aspect ratios at 60% stress level in Example 3 of the present invention; Figure 5 Schematic diagram of the stiffness degradation envelope surface in Example 3 of the present invention; Figure 6 This is a schematic diagram of the structure of a composite material specimen containing wrinkle defects in Example 3 of the present invention. DETAILED DESCRIPTION

[0016] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0018] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. The meaning of "several" is one or more, unless otherwise clearly and specifically defined.

[0019] Wrinkle defects in the composite rotor structure of unmanned aerial vehicles will significantly reduce the load-bearing stiffness. Wrinkle defects in the composite blades of wind turbines will cause the blades to break directly during service. UAV rotors and wind turbine composite blades are always subjected to fatigue loads of load cycles during service. Therefore, it is of great significance to study the fatigue performance of composite structures containing fiber wrinkle defects under fatigue loads, especially the degradation of stiffness performance. In the prior art, the stiffness degradation model is prone to distortion in the process of fitting the stiffness degradation law of composite materials under different stress levels. The prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and it cannot meet the stiffness degradation prediction requirements for wrinkle defect materials.

[0020] In order to solve the above technical problems, the present invention provides a method for predicting fatigue stiffness degradation of a composite material containing wrinkle defects, comprising: obtaining a laminate of the composite material; Determining a characteristic value representing the degree of wrinkle defects of the laminate; Obtain the residual stiffness of laminates with different wrinkle defect degrees at different stress levels and use them as data sets; Using a BP neural network model to read the data set; Utilizing particle swarm algorithm to optimize the BP neural network model; The optimized BP neural network model is trained, and the residual fatigue stiffness prediction model is obtained after the training termination conditions are met; The residual fatigue stiffness prediction model is used to obtain a residual fatigue stiffness degradation curve according to characteristic parameters of the laminate containing wrinkle defects, thereby completing fatigue stiffness degradation prediction; the characteristic parameters include the characteristic values.

[0021] In the case of adopting the above technical scheme, in the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention, by determining the characteristic value characterizing the wrinkle defect degree of the laminate, and obtaining the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels, as a data set, the data set is read through the BP neural network model, and the BP neural network model is optimized by using the particle swarm algorithm, and then the optimized BP neural network model is trained to obtain the residual fatigue stiffness prediction model suitable for the composite material with wrinkle defects of the present invention; and then the residual fatigue stiffness prediction model is used to obtain the residual fatigue stiffness degradation curve according to the characteristic parameters to complete the fatigue stiffness degradation prediction of the composite material with wrinkle defects. Through the above technical scheme of the present invention, the existing stiffness degradation model is easily distorted in the process of fitting the stiffness degradation law of composite materials under different stress levels, the prediction accuracy of fatigue performance degradation under different wrinkle defects is low, and the technical problem of being unable to meet the stiffness degradation prediction requirements of wrinkle defect materials is solved; it should be noted that the laminate obtained above is a laminate containing wrinkle defects, and the laminate can also be prepared.

[0022] In order to better understand the present invention, the content of the present invention is further explained below in conjunction with specific embodiments, but the content of the present invention is not limited to the following embodiments.

[0023] Example 1

[0024] This embodiment provides a method for predicting fatigue stiffness degradation of a composite material containing wrinkle defects, comprising: S100, obtaining a laminate of a composite material; S200, determining a characteristic value characterizing the degree of wrinkle defects of the laminate, wherein the characteristic value is a ratio of a height to a width of the wrinkle defect; S300, to obtain the residual stiffness of laminates with different wrinkle defect degrees at different stress levels and use them as data sets; S400, reading the data set using a BP neural network model; S500, optimizing the BP neural network model using a particle swarm algorithm; S600, training the optimized BP neural network model, and obtaining a residual fatigue stiffness prediction model after satisfying a training termination condition; S700, using the residual fatigue stiffness prediction model, obtaining a residual fatigue stiffness degradation curve according to characteristic parameters of the laminate containing wrinkle defects, and completing fatigue stiffness degradation prediction; the characteristic parameters include the characteristic values.

[0025] Example 2

[0026] This embodiment provides a method for predicting fatigue stiffness degradation of a composite material containing wrinkle defects, comprising: S100, obtaining a laminate of a composite material; S200, determining a characteristic value characterizing the degree of wrinkle defects of the laminate, wherein the characteristic value is a ratio of a height to a width of the wrinkle defect; S300, obtaining the static strength of the laminates with different wrinkle defect degrees, determining the stress level according to the static strength, performing a tension-tension fatigue test on the laminates, determining the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels, and using the residual stiffness as a data set; S400, reading the data set using a BP neural network model; S500, determining the characteristic parameters, which include characteristic values, width, thickness, stress level, and normalized fatigue loading times; using the characteristic parameters as input variables of the BP neural network model; using the particle swarm algorithm to initialize the random position and speed of the particle swarm, and determine the number of particles, space dimension, number of iterations, inertia factor, and learning factor, and iterate; updating the individual extreme value and the global extreme value in each iteration process, updating the random position and speed of the particle swarm after the iteration is completed, decoding and outputting the optimal neural network parameters; inputting the optimal neural network parameters into the BP neural network model to optimize the BP neural network model; S600, trains the optimized BP neural network model through the matlab neural network toolbox; During the training process, when the mean square error and the number of iterations meet the termination condition, the residual fatigue stiffness prediction model is output; wherein the termination condition is that the mean square error is less than or equal to 0.002, or the number of iterations is less than or equal to 1000 times; S700, using the residual fatigue stiffness prediction model, taking the characteristic parameters as input, and outputting a residual fatigue stiffness degradation curve to complete fatigue stiffness degradation prediction; the characteristic parameters include the characteristic values, and the characteristic values ​​are the ratio of the height to the width of the wrinkle defect.

[0027] Example 3 This embodiment provides a method for predicting fatigue stiffness degradation of a composite material containing wrinkle defects, comprising: S100, obtaining a laminate of a composite material; S200, determining a characteristic value characterizing the degree of wrinkle defects of the laminate, wherein the characteristic value is a ratio of a height to a width of the wrinkle defect; S300, obtaining the static strength of the laminates with different wrinkle defect degrees, determining the stress level according to the static strength, performing a tension-tension fatigue test on the laminates, determining the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels, and using the residual stiffness as a data set; specifically: The tensile-tensile fatigue test of composite materials with fiber wrinkle defects was carried out in room temperature dry environment (RTD). The tensile-tensile fatigue test method and specimen size of composite materials refer to GJB2637-1996 "Fatigue Test Method for Carbon Fiber Resin Matrix Composite Laminates", please refer to Figure 6 For composite laminates with wrinkle defects, the characteristic value, that is, the ratio of the height h and width b of the wrinkle is selected as λ = 0. λ = 0.1 and λ = 0.2; the stress ratio is R=0.1; the fatigue stress levels are 80%, 70% and 60% of the maximum load level respectively, and three tensile-tensile fatigue tests of wrinkle-defect composite materials are completed at each stress level. The loading frequency during the test is 5Hz; Furthermore, the residual stiffness of laminates with different wrinkle defect levels at different stress levels is represented by stiffness degradation curves, see Figures 2 to 4 ,in, Figure 2 is the stiffness degradation curve of the intact specimen and the specimens with different fold aspect ratios under 80% stress level, Figure 3 is the stiffness degradation curve of the intact specimen and the specimens with different fold aspect ratios under 70% stress level, Figure 4 The stiffness degradation curves of intact specimens and specimens with different fold aspect ratios under 60% stress level. The intact specimens mentioned above refer to the fold aspect ratios. λ = 0 laminate; Figures 2 to 4 middle, λ = 0. λ = 0.1 and λ = 0.2 respectively indicates that the ratio of the height h to the width b of the fold is 0, 0.1 and 0.2, where λ = 0 represents an intact specimen; further, in the above process, the static strength of various specimens can be determined by conducting static tests on the specimens, and then the size of the fatigue test stress level can be determined. In addition, the tensile stiffness can be determined by static tests. The stiffness measured by the static test is the residual fatigue stiffness when the fatigue life is 0, which is used to verify the accuracy of the test; the above stress levels are selected as 80%, 70% and 60% of the static strength respectively; S400, using the BP neural network model to read the data set, that is, using the data set as the input of the BP neural network model, the data types of the data set include the specimen width, the specimen thickness, the stress level, the aspect ratio of the fold, and the normalized fatigue loading times, and the characteristic parameters in the following steps are determined by these data types; S500, determining the characteristic parameters, which include characteristic values, width, thickness, stress level, and normalized fatigue loading times; in this step, the BP neural network model is optimized by using a particle swarm algorithm with the training error minimization as the optimization goal, specifically: using the characteristic parameters as input variables of the BP neural network model; using the particle swarm algorithm, initializing the random position and speed of the particle swarm, and determining the number of particles, spatial dimension, number of iterations, inertia factor, and learning factor, and iterating; updating individual extreme values ​​and global extreme values ​​in each iteration process, updating the random position and speed of the particle swarm after the iteration is completed, decoding and outputting the optimal neural network parameters; inputting the optimal neural network parameters into the BP neural network model to optimize the BP neural network model; In the above process, the number of particles is 200, the spatial dimension is 5, the number of iterations is 200, the inertia factor ∈ [0.4, 0.9], and the learning factor ∈ [0, 3.5]; In the above process, the particle speed and position are updated after iteration, and the individual extreme value P and the global extreme value G are updated in each iteration. When the individual extreme value is updated, the first i The current fitness value of each particle is compared with the fitness value of the individual extreme value experienced by the particle. If the current fitness value is better, then update the current fitness value, otherwise keep the current fitness value unchanged and update the global extreme value in the same way; In the above process, PSO decodes and outputs the optimal result and the number of iterations, and passes the optimization result input to the established BP neural network model. The input and output of the neural network and the number of neurons are determined through the optimization algorithm; the activation function is selected, and the initial weight and threshold of the network are determined; S600, trains the optimized BP neural network model through the matlab neural network toolbox; During the training process, when the mean square error and the number of iterations meet the termination condition, the residual fatigue stiffness prediction model is output; wherein the termination condition is that the mean square error is less than or equal to 0.002, or the number of iterations is less than or equal to 1000 times; during the training process, the training parameters of the neural network are automatically adjusted to achieve the optimal training effect; Furthermore, in the process of training the BP neural network model, its training error is also calculated and its training convergence is determined; this embodiment uses the mean square error (MSE) to calculate the training error: ; Among them, MSE represents mean square error, m is the total number of data in the database, Y i is the true value of the test, y i is the predicted value; Furthermore, in this embodiment, the termination condition of the training is set to: The mean square error NSE is less than or equal to 0.002; The number of iterations is less than or equal to 1000; If the training termination condition is not met, return to step S400 and repeat the above steps; If the termination condition of the training is met, the iteration is completed and the residual fatigue stiffness prediction model is output; S700, using the residual fatigue stiffness prediction model, taking the characteristic parameter as input, outputting a residual fatigue stiffness degradation curve, and completing fatigue stiffness degradation prediction; the characteristic parameter includes the characteristic value, and the characteristic value is a ratio of the height to the width of the wrinkle defect; In this step, the characteristic parameters of the specimen to be pre-tested are confirmed, and the characteristic parameters are used as the input of the residual fatigue stiffness prediction model. The residual fatigue stiffness prediction model outputs the residual fatigue stiffness degradation curve, thereby completing the fatigue stiffness degradation prediction of the composite material containing wrinkle defects; further, the characteristic parameters include specimen width, specimen thickness, stress level, aspect ratio of wrinkles, and normalized fatigue loading times. The aspect ratio of wrinkles is the characteristic value of the wrinkle defect degree of the laminate. λ ; Further, when the characteristic value range is 0-5, the residual fatigue stiffness degradation curve output by the residual fatigue stiffness prediction model forms a stiffness degradation envelope surface. For example, a stiffness degradation envelope surface formed by the residual fatigue stiffness degradation curve is as follows Figure 5 As shown; the prediction results obtained in this embodiment are compared with the experimental results, and the results are consistent with each other. 2It is 0.96, indicating that the residual fatigue stiffness prediction model (PSO-BP model) of the present invention can accurately predict the residual fatigue stiffness of composite test pieces with different fold aspect ratios.

[0028] See also Figure 1 , based on the above embodiments, for example, combined with Figure 1 The technical solution of the present invention is further described: Figure 1 It can be seen that the fatigue stiffness degradation prediction method of the composite material containing wrinkle defects of the present invention can be divided into four parts, namely, establishing a fatigue test database, establishing a particle swarm optimization model, establishing a neural network prediction model (for outputting the above-mentioned residual fatigue stiffness prediction model) and fatigue stiffness prediction. In the process of establishing the fatigue test database, the wrinkle defect aspect ratio (characteristic value of the wrinkle) and the wrinkle defect fatigue test stress ratio are determined respectively, and the residual fatigue stiffness degradation law curve is obtained by performing a tension-tension fatigue test on the specimen, that is, the residual stiffness of the laminate with different wrinkle defect degrees at different stress levels in step S300 of the above-mentioned embodiment 3, and is used as a data set, and a neural network database is established; in the process of establishing the particle swarm optimization model, the BP neural network model is used to read the residual stiffness database data (the data set in embodiment 3), and by setting The particle swarm algorithm (PSO) optimizes the parameters and initializes them, performs PSO generation, updates the particle speed and position, and PSO decodes and outputs the optimal result and the number of iterations. If the iteration is not completed, it continues to return to the "PSO generation" for cyclic iteration until the iteration is completed. In the process of establishing the neural network prediction model, the BP neural network parameters are initialized and the BP neural network is trained. After adjusting the BP neural network parameters, its convergence and the training error (MSE error) are judged. For details, see step S600 in the above embodiment 3. When the termination condition of the training is met, the residual fatigue stiffness prediction model is output. In the process of fatigue stiffness prediction, the residual fatigue stiffness prediction model outputted above is used, and the characteristic parameters are used as input, and the output is the residual fatigue stiffness degradation curve, so that the residual fatigue stiffness of any wrinkle defect aspect ratio can be obtained, thereby completing the fatigue stiffness degradation prediction.

[0029] In the description of the above embodiments, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0030] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting fatigue stiffness degradation of a composite material containing wrinkle defects, characterized in that: include: obtaining a laminate of the composite material; Determining a characteristic value representing the degree of wrinkle defects of the laminate; Obtain the residual stiffness of laminates with different wrinkle defect degrees at different stress levels and use them as data sets; Using a BP neural network model to read the data set; Utilizing particle swarm algorithm to optimize the BP neural network model; The optimized BP neural network model is trained, and the residual fatigue stiffness prediction model is obtained after the training termination conditions are met; The residual fatigue stiffness prediction model is used to obtain a residual fatigue stiffness degradation curve according to characteristic parameters of the laminate containing wrinkle defects, thereby completing fatigue stiffness degradation prediction; the characteristic parameters include the characteristic values.

2. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 1, characterized in that: The characteristic value is the ratio of the height to the width of the wrinkle defect.

3. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 2, characterized in that: The method of obtaining the residual stiffness of laminates with different wrinkle defect degrees at different stress levels includes: Obtain the static strength of laminates with different wrinkle defect levels; determining the stress level according to the static strength; The laminates are subjected to tension-tension fatigue tests to determine the residual stiffness of the laminates with different wrinkle defect degrees at different stress levels.

4. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 3, characterized in that: The method of optimizing the BP neural network model by using a particle swarm algorithm includes: Determine the characteristic parameters, and use the characteristic parameters as input variables of the BP neural network model; Using the particle swarm algorithm, the random position and velocity of the particle swarm are initialized, and the number of particles, spatial dimension, number of iterations, inertia factor, and learning factor are determined, and iteration is performed; In each iteration, the individual extreme value and the global extreme value are updated. After the iteration is completed, the random position and speed of the particle swarm are updated, and the optimal neural network parameters are decoded and output; The optimal neural network parameters are input into the BP neural network model to optimize the BP neural network model.

5. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 4, characterized in that: The optimized BP neural network model is trained to obtain a residual fatigue stiffness prediction model after the training termination condition is met, including: The optimized BP neural network model is trained through the MATLAB neural network toolbox; During the training process, when the mean square error and the number of iterations meet the termination conditions, the residual fatigue stiffness prediction model is output.

6. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 5, characterized in that: The termination condition is that the mean square error is less than or equal to 0.002, or the number of iterations is less than or equal to 1000.

7. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 6, characterized in that: The characteristic parameters are characteristic parameters of the laminate, and the characteristic parameters include characteristic value, width, thickness, stress level, and normalized fatigue loading times; The characteristic value is the ratio of the height to the width of the wrinkle defect, and the normalized fatigue loading number is determined by a tension-tension fatigue test.

8. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 7, characterized in that: The residual fatigue stiffness prediction model is used to obtain the residual fatigue stiffness degradation curve according to the characteristic parameters of the laminate with wrinkle defects, including: The residual fatigue stiffness prediction model is used, the characteristic parameters are taken as input, and the output is a residual fatigue stiffness degradation curve.

9. The fatigue stiffness degradation prediction method of a composite material containing wrinkle defects according to claim 8, characterized in that: The characteristic value ranges from 0 to 5, and the residual fatigue stiffness degradation curve output by the residual fatigue stiffness prediction model forms a stiffness degradation envelope surface.

Images