Multi-scale based auxiliary prediction method for the properties of carbon fiber composites

Abstract

The present invention relates to the technical field of assisting in predicting material properties, and discloses a method for assisting in predicting the properties of carbon fiber composite materials based on multiple scales. The method includes: obtaining the multi-scale features of the carbon fiber composite material; performing a correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features, wherein the performance indicators include: target performance indicators and non-target performance indicators; generating constraint conditions for the performance indicators according to the application scenario requirements of the carbon fiber composite material; generating an objective function for the performance of the carbon fiber composite material according to the target performance indicators, the non-target performance indicators and a preset adaptive penalty function. The present invention enhances the practicability and reliability of predicting the properties of carbon fiber composite materials.

Description

Technical Field

[0001] The present invention relates to the technical field of material performance auxiliary prediction, and in particular to a multi-scale carbon fiber composite material performance auxiliary prediction method. Background Art

[0002] As a multiphase composite material, carbon fiber composite materials are composed of carbon fiber and matrix materials (such as resin matrix). Their internal structure is highly complex at different scales. Ordinary carbon fiber composite material performance auxiliary prediction methods often only focus on the macroscopic characteristics of the material and approximate the composite material as a uniform continuum. Multi-scale performance auxiliary prediction methods can help engineers accurately predict the performance of materials in the material design stage, thereby optimizing the composition and structure of materials according to specific engineering requirements. By adjusting the fiber layup, selecting appropriate fiber and matrix materials, and optimizing the manufacturing process, carbon fiber composite materials with better performance and lower cost can be designed to improve the comprehensive performance and cost performance of the material.

[0003] Before predicting the performance of carbon fiber composite materials, it is necessary to optimize the proportion of key components of carbon fiber composite materials. In the process of optimizing the proportion of key components of carbon fiber composite materials, a universal optimization target is often constructed; and in the optimization algorithm of carbon fiber composite material performance, the conventional practice is to establish an objective function. In this process, various performance indicators are comprehensively optimized with fixed weights, which leads to the following defects in the auxiliary prediction of the performance of carbon fiber composite materials:

[0004] 1. The existence of universal optimization goals makes the prediction results disconnected from the actual application scenarios. The universal optimization goals lack customized considerations for specific industry application scenarios and cannot accurately reflect the performance of carbon fiber composite materials in specific industries under actual working conditions. In addition, universal goals may cause the model to rely too much on general data sets, while ignoring industry-specific experimental data or working conditions, further weakening the practicality and reliability of the prediction method for the performance of carbon fiber composite materials.

[0005] 2. Fixed weights are used to coordinately optimize various performance indicators. This strategy lacks in-depth consideration of the characteristics of different application scenarios. The fixed weight optimization method cannot dynamically adjust the optimization direction according to the needs of specific scenarios. This static optimization strategy is difficult to adapt to multi-objective conflicts under complex working conditions. For example, the pursuit of high strength may sacrifice the toughness or processing performance of the material. This optimization method restricts the application value of the prediction method of carbon fiber composite material performance in actual industry. Summary of the invention

[0006] The present invention provides a method for assisting in predicting the properties of carbon fiber composite materials based on multiple scales, and its main purpose is to solve the problem in the prior art that in the optimization process of the properties of carbon fiber composite materials, a general optimization goal is usually constructed, and the general optimization goal lacks customized consideration for specific industry application scenarios, thereby weakening the practicability and reliability of the prediction of the properties of carbon fiber composite materials.

[0007] To achieve the above object, a method for assisting in predicting the properties of carbon fiber composite materials based on multiple scales provided by the present invention includes:

[0008] S1: Obtain the multi-scale features of the carbon fiber composite material;

[0009] S2: Conduct a correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features, where the performance indicators include: target performance indicators and non-target performance indicators;

[0010] S3: Generate constraint conditions for the performance indicators according to the application scenario requirements of the carbon fiber composite material;

[0011] S4: Generate an objective function for the properties of the carbon fiber composite material according to the target performance indicators, the non-target performance indicators, and a preset adaptive penalty function;

[0012] S5: Under the constraint conditions, optimize the parameter set of the objective function based on the gradient descent algorithm to obtain an approximate minimum value of the objective function, and determine the optimal parameter set according to the approximate minimum value of the objective function;

[0013] S6: Determine the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set, and synthesize the carbon fiber composite material according to the optimal parameter set.

[0014] Optionally, the obtaining of the multi-scale features of the carbon fiber composite material includes:

[0015] SS1: Prepare the carbon fiber composite material samples using different carbon fibers, different matrices, different fiber volume fractions, and different ply layups;

[0016] SS2: Conduct microscopic, mesoscopic, and macroscopic structure analyses on the carbon fiber composite material samples to obtain the multi-scale features of the carbon fiber composite material.

[0017] Optionally, the conducting of the correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features includes:

[0018] S21. Substitute the multi-scale features and the performance metrics into the Pearson correlation coefficient formula to obtain the correlation coefficient of the multi-scale features. The Pearson correlation coefficient formula is as follows:

[0019]

[0020] where, represents the th multi-scale feature of the th carbon fiber composite sample, represents the performance value of the th performance metric of the th carbon fiber composite sample, represents the total number of carbon fiber composite samples, and respectively represent the average values of the performance values of the multi-scale features and the performance metrics among all carbon fiber composite samples, represents the correlation coefficient of the th multi-scale feature of the th carbon fiber composite sample;

[0021] S22. Based on the significance test formula, determine whether the correlation coefficient of the th multi-scale feature of the th carbon fiber composite sample and the performance metric is significant. The significance test formula is as follows:

[0022]

[0023] where, represents the total number of carbon fiber composite samples, represents the correlation coefficient of the th multi-scale feature of the th carbon fiber composite sample, represents the statistic of the th multi-scale feature of the th carbon fiber composite sample;

[0024] S23. In the distribution table, retrieve the corresponding value according to the degrees of freedom and the statistic of the th multi-scale feature of the th carbon fiber composite sample. When the value is less than the preset significance level, it is considered that the th multi-scale feature of the th carbon fiber composite sample and the The performance indicators of a carbon fiber composite material sample are significantly correlated, and the multi-scale features of a carbon fiber composite material sample are marked as the relevant features. Among them,

[0025] Optionally, the constraint conditions are as follows:

[0026]

[0027] Among them, represents the index of the target performance indicator, represents the index of the non-target performance indicator, represents the performance value of the th target performance indicator, represents the superior value of the th target performance indicator, represents the performance value of the th non-target performance indicator, represents the lower threshold of the th non-target performance indicator, represents the set of relevant features corresponding to the target performance indicator, represents the set of relevant features corresponding to the non-target performance indicator, represents the set of the relevant features, represents the total number of the performance indicators.

[0028] Optionally, the objective function is as follows:

[0029]

[0030] Among them, represents the index of the target performance indicator, represents the index of the non-target performance indicator, represents the performance value of the th target performance indicator, represents the superior value of the th target performance indicator, represents the performance value of the th non-target performance indicator, represents the lower threshold of the th non-target performance indicator, represents the set of relevant features corresponding to the target performance indicator, Represents the set of relevant features corresponding to the non-target performance metrics, Represents the set of the relevant features, Represents the total number of the performance metrics, Represents the objective function, Represents the adaptive penalty function, Represents taking and the maximum value of two numbers in

[0031] Optionally, the preset adaptive penalty function is as follows:

[0032]

[0033] Wherein, Represents the adaptive penalty function, Represents the basic penalty factor, Represents the maximum default amount among all non-target performance metrics.

[0034] Optionally, the steps for constructing the adaptive penalty function are as follows:

[0035] Step 1: Calculate the difference between the performance value of each non-target performance metric and the lower threshold of the corresponding non-target performance metric. If there is , it means that the performance value of the -th non-target performance metric has a default amount. Determine the maximum default amount among all non-target performance metrics according to the minimum value of all the differences. The calculation formula for the maximum default amount is as follows:

[0036]

[0037] Wherein, Represents the index of the non-target performance metric, Represents the performance value of the -th non-target performance metric, Represents the lower threshold of the -th non-target performance metric, Represents the adjustment parameter, Represents the maximum default amount among all non-target performance metrics;

[0038] Step 2: Construct an adaptive penalty function based on the maximum default amount and the basic penalty factor.

[0039] Optionally, optimizing the parameter set of the objective function based on the gradient descent algorithm to obtain an approximate minimum value of the objective function includes:

[0040] Optimize the parameter set of the objective function through the update formula of the gradient descent algorithm, where the update formula is as follows:

[0041]

[0042] Among them, represents the set of the relevant features, represents the parameter set at the th iteration, represents the parameter set at the th iteration, represents the learning rate, represents the gradient of the objective function at ;

[0043] When the change in the objective function value is less than a preset threshold, obtain the approximate minimum value of the objective function.

[0044] Optionally, the optimal parameter set is:

[0045]

[0046] Among them, represents the optimal parameter set, represents the updated parameter set at the th iteration, represents the set of the relevant features.

[0047] Optionally, determining the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set, and synthesizing the carbon fiber composite material according to the optimal parameter set includes:

[0048] Calculating the optimal performance prediction value of the carbon fiber composite material based on the material performance prediction algorithm and the optimal parameter set, where the material performance prediction algorithm is as follows:

[0049]

[0050] Among them, represents the optimal parameter set, represents the optimal performance prediction value, represents the material performance prediction model;

[0051] Determine the best material ratio of the carbon fiber composite material according to the optimal parameter set, and synthesize the carbon fiber composite material according to the best material ratio and the manufacturing process.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. Generate constraint conditions for performance indicators according to the application scenario requirements of carbon fiber composites, distinguish target performance indicators from non-target performance indicators, set superior values for the target performance indicators to ensure that they reach or exceed the expected level, and set lower limit thresholds for the non-target performance indicators to meet the requirements for the basic performance of the material under specific scenarios, improving the pertinence of the performance prediction results of carbon fiber composites, enhancing the adaptability and practicability of the prediction model, and providing strong support for the research and development and application of carbon fiber composites;

[0054] 2. By constructing an adaptive penalty function, the flexibility and efficiency of the multi-objective optimization process in the performance assisted prediction method of carbon fiber composites can be improved, the adaptability of the performance assisted prediction method of carbon fiber composites to complex working conditions can be enhanced, enabling it to better meet the diverse requirements of different application scenarios, thereby improving the performance and reliability of carbon fiber composites in practical applications. The adaptive penalty function can dynamically adjust the penalty intensity according to the default situation of non-target performance indicators. When the non-target performance indicator does not reach the lower limit threshold, the value of the adaptive penalty function will increase significantly, increasing the constraint intensity on the optimization process, prompting the optimization algorithm to preferentially adjust the parameters of the objective function to reduce the degree of default; this dynamic adjustment ability not only improves the optimization efficiency but also better balances the conflict problems in multi-objective optimization;

[0055] 3. Under the constraint conditions, optimize the parameter set of the objective function based on the gradient descent algorithm, which can improve the efficiency and accuracy of the multi-objective optimization process, adjust the parameters according to the change of the objective function value, and obtain an approximate minimum value when the change of the objective function value is less than the preset threshold to determine the optimal parameter set. During the optimization process, the adaptive penalty function makes the gradient of the objective function change dynamically with the degree of default of non-target performance indicators, and then makes the gradient descent algorithm more inclined to adjust the parameters of the objective function to improve the degree of default, realizing the dynamic adjustment of the optimization direction. This optimization strategy provides more scientific and reliable guidance for the design and process improvement of carbon fiber composites, and improves the performance and applicability of the material in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic flow chart of a performance assisted prediction method for carbon fiber composites based on multi-scale provided by an embodiment of the present invention;

[0057] The realization, functional characteristics and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0059] An embodiment of the present application provides a method for assisting in predicting the performance of carbon fiber composite materials based on multiple scales. The execution subject of the method for assisting in predicting the performance of carbon fiber composite materials based on multiple scales includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for assisting in predicting the performance of carbon fiber composite materials based on multiple scales can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0060] Referring to Figure 1 As shown, it is a schematic flowchart of a method for assisting in predicting the performance of carbon fiber composite materials based on multiple scales provided by an embodiment of the present invention. In this embodiment, the method for assisting in predicting the performance of carbon fiber composite materials based on multiple scales includes:

[0061] S1: Obtain the multi-scale features of the carbon fiber composite material.

[0062] In the embodiment of the present invention, the obtaining of the multi-scale features of the carbon fiber composite material includes:

[0063] SS1: Prepare the carbon fiber composite material sample using different carbon fibers, different matrices, different fiber volume fractions, and different layup methods;

[0064] SS2: Perform microscopic, mesoscopic, and macroscopic structure analyses on the carbon fiber composite material sample to obtain the multi-scale features of the carbon fiber composite material.

[0065] Specifically, observe the carbon fiber and its bonding interface with the matrix through a scanning electron microscope to obtain the microscopic morphology of the carbon fiber composite material; obtain the 3D structure inside the carbon fiber composite material through a multi-photon microscope, and use a differential scanning calorimeter (DSC) and a thermogravimetric analyzer (TGA) to study the thermal stability and thermal behavior of the carbon fiber composite material.

[0066] Among them, the multi-scale feature representation describes the information of the internal structure of the carbon fiber composite material sample at different scales. The multi-scale features include micro-scale data (single fiber diameter, fiber surface roughness), meso-scale data (fiber volume fraction, fiber arrangement and directionality), and macro-scale data (overall laminated structure, overall geometric dimensions and shape). Performance indicators such as rigidity, elasticity, toughness, and fatigue life are obtained through methods such as tensile tests, impact tests, and fatigue tests.

[0067] For example, the parameters of different carbon fiber composite material samples are as follows:

[0068]

[0069] S2: Perform a correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features, where the performance indicators include: target performance indicators and non-target performance indicators.

[0070] In the embodiment of the present invention, the performing a correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features includes:

[0071] S21: Substitute the multi-scale features and the performance indicators into the Pearson correlation coefficient formula to obtain the correlation coefficients of the multi-scale features, where the Pearson correlation coefficient formula is as follows:

[0072]

[0073] Among them, represents the th multi-scale feature of the th carbon fiber composite material sample, represents the performance value of the th performance indicator of the th carbon fiber composite material sample, represents the total number of the carbon fiber composite material samples, and respectively represent the average values of the performance values of the multi-scale features and the performance indicators in all carbon fiber composite material samples, represents the th correlation coefficient of the th multi-scale feature of the

[0074] Among them, the Pearson correlation coefficient formula is used to measure the linear correlation between two variables, and its value range is [-1, 1]. A value of 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no correlation;

[0075] S22. Determine whether the correlation coefficient of the th multi-scale feature of the th carbon fiber composite material sample and the performance index is significant. The significance test formula is as follows:

[0076]

[0077] where represents the total number of the carbon fiber composite material samples, represents the th th correlation coefficient of the multi-scale feature of the th carbon fiber composite material sample, represents the th statistic of the

[0078] S23. In the distribution table, retrieve the corresponding value according to the degrees of freedom and the th statistic of the th multi-scale feature of the th carbon fiber composite material sample. When the value is less than the preset significance level, it is considered that the th multi-scale feature of the th carbon fiber composite material sample and the th performance index of the th carbon fiber composite material sample have a significant correlation, and mark the th multi-scale feature of the th carbon fiber composite material sample as the relevant feature. Here, the

[0079] value represents the probability of obtaining the current data result under the condition that the null hypothesis holds, and the degrees of freedom represent the number of observation data points that can vary independently when performing the correlation analysis using the multi-scale feature and the performance index.

[0080] S3: Generate the constraint conditions for the performance index according to the application scenario requirements of the carbon fiber composite material.

[0081]

[0082] where represents the index of the target performance index, represents the index of the non-target performance index, Denote the performance value of the th target performance indicator, Denote the superiority value of the th target performance indicator, Denote the performance value of the th non - target performance indicator, Denote the lower threshold value of the th non - target performance indicator, Denote the set of relevant features corresponding to the target performance indicator, Denote the set of relevant features corresponding to the non - target performance indicator, Denote the set of the relevant features, Denote the total number of the performance indicators.

[0083] Among them, the superiority value represents the optimal performance value of the target performance indicator, which is used to ensure that the target performance indicator reaches or exceeds the expected target level.

[0084] S4: Generate an objective function for the performance of the carbon fiber composite material according to the target performance indicator, the non - target performance indicator and a preset adaptive penalty function.

[0085] In the embodiment of the present invention, the objective function is as follows:

[0086]

[0087] Among them, Denote the index of the target performance indicator, Denote the index of the non - target performance indicator, Denote the th performance value of the target performance indicator, Denote the th superiority value of the target performance indicator, Denote the th performance value of the non - target performance indicator, Denote the th lower threshold value of the non - target performance indicator, Denote the set of relevant features corresponding to the target performance indicator, Denote the set of relevant features corresponding to the non - target performance indicator, Denote the set of the relevant features, Denote the total number of the performance indicators, Denote the objective function, Denote the adaptive penalty function, Denote taking and the maximum value of the two numbers in.

[0088] In , represents the optimization direction of the performance value representing the target performance index, that is, it is desired that is as close as possible to or exceeds the superior value, while satisfying the constraints of the non-target performance index, is used to measure the degree to which the non-target performance index fails to meet the standard and impose constraints on the optimization process, that is represents the penalty term. When fails to reach , then a penalty is imposed. When all are met, the value of the penalty term is 0, and the objective function is only . When a certain fails to reach , then the penalty term will increase 's value, driving the adjustment of parameters during the optimization process to reduce the degree of default.

[0089] Specifically, the preset adaptive penalty function is as follows:

[0090]

[0091] Among them, represents the adaptive penalty function, represents the basic penalty factor, represents the maximum default amount among all non-target performance indicators.

[0092] Specifically, the construction steps of the adaptive penalty function are as follows:

[0093] Step 1: Calculate the difference between the performance value of all the non-target performance indicators and the lower threshold of the corresponding non-target performance indicator. If there exists , it means that there is a default amount in the performance value of the th non-target performance indicator. According to the minimum value of all the differences, determine the maximum default amount among all non-target performance indicators. Among them, the calculation formula of the maximum default amount is as follows:

[0094]

[0095] Among them, represents the index of the non-target performance indicator, represents the th performance value of the non-target performance indicator, represents the th lower threshold of the non-target performance indicator, represents the adjustment parameter, Represents the maximum default amount among all non-target performance metrics;

[0096] Step 2: Construct an adaptive penalty function based on the maximum default amount and the base penalty factor.

[0097] When all reach the corresponding , that is , then , so , and further makes have a smaller value; when at least one does not reach , that is , then significantly increases, and further makes increase.

[0098] S5: Under the constraint conditions, optimize the parameter set of the objective function based on the gradient descent algorithm to obtain the approximate minimum value of the objective function, and determine the optimal parameter set according to the approximate minimum value of the objective function.

[0099] In the embodiment of the present invention, the optimizing the parameter set of the objective function based on the gradient descent algorithm to obtain the approximate minimum value of the objective function includes:

[0100] Optimize the parameter set of the objective function through the update formula of the gradient descent algorithm, where the update formula is as follows:

[0101]

[0102] Among them, represents the set of the relevant features, represents the parameter set at the th iteration, represents the parameter set at the th iteration, represents the learning rate, represents the gradient of the objective function at .

[0103] Among them, is calculated as follows:

[0104]

[0105] Among them, and respectively represent and with respect to , gradients, denote with respect to the gradient of , denote taking the value 1 when and 0 otherwise denote the adaptive penalty function

[0106] wherein, the formula for calculating the gradient of the penalty term with respect to is as follows

[0107]

[0108] wherein denote the gradient with respect to the gradient of denote with respect to the gradient of , denote the adaptive penalty function

[0109] wherein, the formula for calculating the gradient of the adaptive penalty function with respect to is as follows

[0110]

[0111] wherein denote with respect to the gradient of denote the performance value of the th non - target performance metric denote the th lower threshold of the non - target performance metric denote the base penalty factor denote the adjustment parameter

[0112] When the change in the objective function value is less than the preset threshold, the approximate minimum value of the objective function is obtained

[0113] In the objective function, when is close to , has a small value, indicating that the penalty term in the objective function is small and has a weak impact on optimization; when is much lower than , significantly increases, resulting in a sharp increase in the value of the objective function. During the optimization process, the gradient becomes larger, making the gradient descent algorithm more inclined to adjust the parameters of the objective function to improve the degree of default, and has not reached the penalty intensity of the difference part generated increases dynamically with the degree of default

[0114] Specifically, during the optimization process, if a non-target performance metric fails to reach the corresponding lower threshold , and the difference between the two is large, the adaptive penalty function will increase significantly, so as to impose a greater correction force during the gradient descent process of the objective function, enabling the optimization algorithm to preferentially adjust the parameters of the objective function to reduce the above difference. This adaptive adjustment can avoid the problems of slow convergence or difficulty in meeting all constraints that may be encountered in traditional fixed penalty methods.

[0115] Specifically, the optimal parameter set is:

[0116]

[0117] where represents the optimal parameter set, represents the parameter set updated at the -th iteration, represents the set of the relevant features.

[0118] S6: Determine the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set, and synthesize the carbon fiber composite material according to the optimal parameter set.

[0119] In the embodiment of the present invention, the determining the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set and synthesizing the carbon fiber composite material according to the optimal parameter set includes:

[0120] Calculating the optimal performance prediction value of the carbon fiber composite material based on the material performance prediction algorithm and the optimal parameter set, where the material performance prediction algorithm is as follows:

[0121]

[0122] where represents the optimal parameter set, represents the optimal performance prediction value, represents the material performance prediction model;

[0123] where describes the mapping relationship from the optimal parameter set to the optimal performance prediction value, The model can be constructed based on machine learning algorithms. By leveraging machine learning algorithms such as neural networks and support vector machines, it learns the complex non-linear relationship between parameters and performance from a large amount of experimental data. For example, using a multi-layer perceptron (MLP) neural network, with the parameter set as the input layer and the performance prediction value as the output layer, the network is trained with the training data to obtain a model that can accurately predict performance.

[0124] Determine the optimal material ratio of the carbon fiber composite according to the optimal parameter set, and synthesize the carbon fiber composite according to the optimal material ratio and manufacturing process.

[0125] For example, set the target performance index as rigidity and set its corresponding superior value as 100. The non-target performances are toughness and fatigue life which need to meet the minimum requirements. The lower threshold of toughness is 55, and the lower threshold of fatigue life is 65; Initialize parameters: rigidity

[0126] toughness and fatigue life ;

[0127] Optimize the objective function: basic penalty factor adjustment parameter learning rate = 0.1, maximum number of iterations = 50;

[0128] Calculate the maximum default amount: ;

[0129] Calculate the adaptive penalty function: ;

[0130] Optimize rigidity toughness and fatigue life The specific steps are as follows:

[0131] Update the target performance index:

[0132] Update the non-target performance index:

[0133]

[0134]

[0135] ​​Recalculate the maximum default quantity: ;

[0136] Calculate the new adaptive penalty function: ;

[0137] Repeat the steps of optimizing rigidity , toughness , fatigue life for 50 iterations. After 50 iterations:

[0138] , , , , the target performance index (rigidity) is maximized, increasing from 80 to 100.2. The non-target performance indices (toughness, fatigue life) meet the minimum requirements. The adaptive penalty function is automatically adjusted during the optimization process to improve the optimization accuracy. Finally, the optimal parameter set is determined and the optimization goal is achieved.

[0139] Specifically, according to the optimal parameter set, first analyze the key component ratios of the corresponding carbon fiber composite material, including carbon fiber types (such as high modulus carbon fiber, high strength carbon fiber), matrix materials (such as epoxy resin, polyamide, phenolic resin), fiber volume fractions (such as 30%, 50%, 70%), and layup methods (uniaxial layup, cross-ply layup, random layup), etc. Considering the physical, chemical, and mechanical properties of the material comprehensively, determine the best material ratio that meets the requirements of the target performance index; after obtaining the best material ratio, combine the manufacturing process parameters (such as prepreg preparation, hot pressing and curing, vacuum assisted resin transfer molding (VARTM), automatic layup process, etc.), optimize the process parameters such as temperature, pressure, curing time, and layup sequence to ensure that the finally synthesized carbon fiber composite material achieves the expected effect in the target application scenario. At the same time, use material simulation and experimental verification to conduct performance tests on the manufactured carbon fiber composite material, such as tensile test, impact test, fatigue test, etc., and further adjust and optimize the ratio and process to ensure the reliability, stability, and manufacturability of the finally synthesized material, so as to meet specific engineering requirements.

[0140] It should be noted that steps S1 - S5 are auxiliary processes for determining the optimal performance prediction value of the carbon fiber composite material.

[0141] In several embodiments provided by the present invention, it should be understood that the disclosed method can be implemented in other ways.

[0142] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0143] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence is a theory, method, and technology that uses a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-scale-based method for auxiliary prediction of the properties of carbon fiber composites, characterized in that, The method includes: S1: Obtain the multi-scale features of the carbon fiber composite material; S2: Conduct a correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features, where the performance indicators include: target performance indicators and non-target performance indicators; S3: Generate constraint conditions for the performance indicators according to the application scenario requirements of the carbon fiber composite material; S4: Generate an objective function for the performance of the carbon fiber composite material according to the target performance indicators, the non-target performance indicators, and a preset adaptive penalty function, where the preset adaptive penalty function is as follows: Among them, represents the adaptive penalty function, represents the basic penalty factor, represents the maximum default amount among all non-target performance indicators; The construction steps of the adaptive penalty function are as follows: Step 1: Calculate the difference between the performance value of all the non-target performance indicators and the lower threshold value of the corresponding non-target performance indicators. If there is , it means that there is a default amount in the performance value of the th non-target performance indicator. Determine the maximum default amount among all non-target performance indicators according to the minimum value of all the differences. The calculation formula of the maximum default amount is as follows: Among them, represents the index of the non-target performance indicator, represents the th performance value of the non-target performance indicator, represents the th lower threshold of the non-target performance indicator, represents the adjustment parameter, represents the maximum default amount among all non-target performance indicators; Step 2: Construct an adaptive penalty function based on the maximum default amount and the basic penalty factor; S5: Under the constraint conditions, optimize the parameter set of the objective function based on the gradient descent algorithm to obtain an approximate minimum value of the objective function, and determine the optimal parameter set according to the approximate minimum value of the objective function; S6: Determine the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set, and synthesize the carbon fiber composite material according to the optimal parameter set.

2. The multi-scale-based performance auxiliary prediction method for carbon fiber composite materials according to claim 1, wherein, The obtaining of the multi-scale features of the carbon fiber composite material includes: SS1: Prepare the carbon fiber composite material samples using different carbon fibers, different matrices, different fiber volume fractions, and different layup methods; SS2: Conduct microscopic, mesoscopic, and macroscopic structure analyses on the carbon fiber composite material samples to obtain the multi-scale features of the carbon fiber composite material.

3. The multi-scale based performance assisted prediction method for carbon fiber composite materials according to claim 1, wherein, The conducting of the correlation analysis on the multi-scale features and the performance indicators of the carbon fiber composite material to obtain the relevant features in the multi-scale features includes: S21: Substitute the multi-scale features and the performance indicators into the Pearson correlation coefficient formula to obtain the correlation coefficients of the multi-scale features, where the Pearson correlation coefficient formula is as follows: Among them, represents the th multi-scale feature of the th carbon fiber composite material sample, represents the performance value of the th performance index of the th carbon fiber composite material sample, represents the total number of the carbon fiber composite material samples, and respectively represent the average values of the multi-scale features and the performance values of the performance indexes in all the carbon fiber composite material samples, represents the correlation coefficient of the th multi-scale feature of the th carbon fiber composite material sample; S22. Determine whether the correlation coefficient of the th multi-scale feature of the th carbon fiber composite sample and the performance index is significant. The significance test formula is as follows: Among them, represents the total number of the carbon fiber composite material samples, represents the th correlation coefficient of the represents the th statistic of the th multiscale feature of the th carbon fiber composite material sample; S23. At in the distribution table, according to the degree of freedom and the statistics of the th multi-scale feature of the th carbon fiber composite sample, retrieve the corresponding value. When the value is less than the preset significance level, it is considered that the th multi-scale feature of the th carbon fiber composite sample has a significant correlation with the th performance index, and mark the th multi-scale feature of the th carbon fiber composite sample as the relevant feature. Among them, the value represents the probability of obtaining the current data result under the condition that the null hypothesis holds.

4. The multi-scale based performance assisted prediction method for carbon fiber composite materials according to claim 1, characterized in that, The constraint conditions are as follows: Among them, represents the index of the target performance metric, represents the index of the non-target performance metric, represents the performance value of the th target performance metric, represents the superiority value of the th target performance metric, represents the performance value of the th non-target performance metric, represents the lower threshold of the th non-target performance metric, represents the set of relevant features corresponding to the target performance metric, represents the set of relevant features corresponding to the non-target performance metric, represents the total number of the performance metrics.

5. The multi-scale-based performance auxiliary prediction method for carbon fiber composite materials according to claim 1, characterized in that The objective function is as follows: Among them, represents the index of the target performance metric, represents the index of the non-target performance metric, represents the performance value of the th target performance metric, represents the superior value of the th target performance metric, represents the performance value of the th non-target performance metric, represents the lower threshold value of the th non-target performance metric, represents the total number of the performance metrics, represents the objective function, represents the adaptive penalty function, represents taking and the maximum value of the two numbers.

6. The multi-scale based method for auxiliary prediction of carbon fiber composite material properties according to claim 1, characterized in that, The optimizing of the parameter set of the objective function based on the gradient descent algorithm to obtain an approximate minimum value of the objective function includes: Optimize the parameter set of the objective function through the update formula of the gradient descent algorithm, where the update formula is as follows: Among them, represents the set of the said relevant features, represents the parameter set at the th iteration, parameter set at the th iteration, represents the learning rate, represents the gradient of the said objective function at When the change in the objective function value is less than a preset threshold, obtain the approximate minimum value of the objective function.

7. The performance auxiliary prediction method for carbon fiber composite materials based on multi-scale according to claim 1, characterized in that The optimal parameter set is: Among them, represents the set of optimal parameters, represents the parameter set updated at the -th iteration, represents the set of the said relevant features.

8. The method for auxiliary prediction of the performance of carbon fiber composite materials based on multi-scale as described in claim 1, wherein The determining of the optimal performance prediction value of the carbon fiber composite material according to the optimal parameter set and the synthesizing of the carbon fiber composite material according to the optimal parameter set include: Calculate the optimal performance prediction value of the carbon fiber composite material based on the material performance prediction algorithm and the optimal parameter set, where the material performance prediction algorithm is as follows: Among them, represents the set of the optimal parameters, represents the predicted value of the optimal performance, represents the material performance prediction model; Determine the best material ratio of the carbon fiber composite material according to the optimal parameter set, and synthesize the carbon fiber composite material according to the best material ratio and the manufacturing process.

Images