Method and device for calculating b-basis value of small sample-based aviation composite material, computer device and medium

Abstract

This invention provides a method, apparatus, computer equipment, and medium for calculating the B-benchmark value of aerospace composite materials based on a small sample, relating to the field of material performance testing technology. The method includes the following steps: obtaining a small sample test dataset of the multi-directional laminate to be analyzed, and obtaining a large sample historical test dataset of the corresponding unidirectional laminate based on a correspondence; screening the large sample historical test dataset for validity, removing outliers, and verifying batch-to-batch consistency; calculating the coefficient of variation of the screened large sample historical test dataset, and performing dynamic threshold correction on the coefficient of variation; calculating the statistical parameters of the small sample test dataset and the screened large sample historical test dataset respectively, calculating the weighting coefficient based on the layup ratio of the multi-directional laminate, and calculating the B-benchmark value of the aerospace composite material. This method for calculating the B-benchmark value based on a small sample improves the accuracy and calculation stability of the small sample B-benchmark value.

Description

Technical Field

[0001] This invention relates to the field of material performance testing technology, and in particular to a method, apparatus, computer equipment, and medium for calculating the B-benchmark value of aerospace composite materials based on a small sample. Background Technology

[0002] Accurate calculation of the B-benchmark value is a prerequisite for the safe application of composite materials. Currently, the internationally accepted calculation methods mainly follow the CMH-17G standard in the *Composite Materials Handbook*, which includes two main categories for calculating simple random samples: the single-point method and the multi-environment sample merging method. However, both methods face a common bottleneck in practical applications: the rigid requirement for sample size. In scenarios such as the early stages of new composite material development, the verification of thick-section composite materials, and the testing of high-cost material systems (such as ceramic matrix composites), obtaining sufficient standard samples faces problems such as high costs and long preparation cycles.

[0003] Statistical methods for B benchmark values ​​based on small samples include the one-way plate variation method and the minimum batch method. The minimum batch method calculates B benchmark values ​​that are too conservative. The one-way plate variation method does not take into account the conservative problem of the dispersion coefficient in the production and manufacturing of aerospace composite materials, and it does not make full use of material-level historical data. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method for calculating the B-benchmark value of aerospace composite materials based on a small sample size, to solve the technical problems in the prior art where the calculation of the B-benchmark value of aerospace composite materials requires a large sample size, and the small sample method has low accuracy or is overly conservative. The method includes:

[0005] Obtain a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and obtain the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. The large sample historical test dataset is subjected to validity screening, outliers are removed and batch consistency is checked to generate a screened large sample historical test dataset. Calculate the coefficient of variation of the large sample historical test dataset after screening, and perform dynamic threshold correction on the coefficient of variation to generate the corrected coefficient of variation; Statistical parameters are calculated for the small sample test dataset and the filtered large sample historical test dataset, respectively. Weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. Based on the statistical parameters, the corrected coefficient of variation, and the weighting coefficients, the B-baseline value of the aerospace composite material is calculated.

[0006] This invention also provides a device for calculating the B-benchmark value of aerospace composite materials based on a small sample size, to solve the technical problems in the prior art where the calculation of the B-benchmark value of aerospace composite materials requires a large sample size, and the small sample method has low accuracy or is too conservative. The device includes: The dataset acquisition module is used to acquire a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and acquire the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. The dataset filtering module is used to filter the large sample historical test dataset for validity, remove outliers and check batch consistency, and generate a filtered large sample historical test dataset. The dispersion coefficient correction module is used to calculate the dispersion coefficients of the large sample historical test dataset after screening, and to perform dynamic threshold correction on the dispersion coefficients to generate the corrected dispersion coefficients. The benchmark value calculation module is used to calculate the statistical parameters of the small sample test dataset and the screened large sample historical test dataset respectively, calculate the weighting coefficient based on the layup ratio of the multidirectional laminate, and calculate the B benchmark value of the aerospace composite material based on the statistical parameters, the corrected coefficient of dispersion and the weighting coefficient.

[0007] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned method for calculating the B-benchmark value of aerospace composite materials based on a small sample, thereby solving the technical problems in the prior art where the calculation of the B-benchmark value of aerospace composite materials requires a large sample size and the small sample method has low accuracy or is too conservative.

[0008] This invention also provides a computer-readable storage medium storing a computer program that executes any of the above-described methods for calculating the B-benchmark value of aerospace composite materials based on small samples, in order to solve the technical problems in the prior art where the calculation of the B-benchmark value of aerospace composite materials requires a large sample size and the small sample method has low accuracy or is too conservative.

[0009] Compared with the prior art, the beneficial effects that at least one technical solution adopted in the embodiments of this specification can achieve include at least: The B-benchmark value calculation method based on the embodiments of the present invention improves the accuracy and calculation stability of the B-benchmark value for small samples. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of a method for calculating the B-benchmark value of aerospace composite materials based on a small sample, provided by an embodiment of the present invention. Figure 2 This is a flowchart illustrating a method for calculating the B-benchmark value of aerospace composite materials based on a small sample, provided by an embodiment of the present invention. Figure 3 This is a structural block diagram of a computer device provided in an embodiment of the present invention; Figure 4 This is a structural block diagram of a device for calculating the B-benchmark value of aerospace composite materials based on a small sample, provided in an embodiment of the present invention. Detailed Implementation

[0012] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0013] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] In this embodiment of the invention, a method for calculating the B-benchmark value of aerospace composite materials based on a small sample is provided, such as... Figure 1 and Figure 2 As shown, the method includes: Step S101: Obtain a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and obtain the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. Step S102: Perform validity screening on the large sample historical test dataset, remove abnormal data and check batch consistency to generate a screened large sample historical test dataset; Step S103: Calculate the coefficient of variation of the large sample historical test dataset after screening, and perform dynamic threshold correction on the coefficient of variation to generate the corrected coefficient of variation; Step S104: Calculate the statistical parameters of the small sample test dataset and the filtered large sample historical test dataset respectively, calculate the weighting coefficient based on the layup ratio of the multidirectional laminate, and calculate the B benchmark value of the aerospace composite material based on the statistical parameters, the corrected coefficient of dispersion and the weighting coefficient.

[0015] In practice, the correspondence between the small sample test dataset and the large sample test dataset is determined through the following steps: When the data in the small sample test dataset comes from tensile tests of multidirectional laminates, the 0° tensile strength data and ±45° shear strength data in the large sample historical test dataset are obtained; when the data in the small sample test dataset comes from compression or shear tests of multidirectional laminates, the 0° compression strength data and ±45° shear strength data in the large sample historical test dataset are obtained.

[0016] In practice, the following steps are used to screen the large sample historical test dataset for validity, remove outlier data, and verify batch-to-batch consistency, thereby generating a screened large sample historical test dataset: For each type of large-sample historical experimental data, perform the following operations until all data types have been processed, then merge all batch-consistent datasets to obtain a valid large-sample historical experimental dataset: Calculate the maximum normalized residual statistic and the corresponding critical value of the large sample historical test dataset; If the maximum normalized residual statistic of the data point is greater than the critical value, the current data point is removed, and the iteration is repeated until all data points have been judged to generate an anomaly-free dataset. The anomaly-free dataset is grouped by batch, and the test statistic and critical value of each batch of the anomaly-free dataset are calculated using the k-sample Anderson-Darling test method. If the test statistic is greater than or equal to the critical value, the batch of the current data type is determined to be inconsistent and marked as unusable; otherwise, all data of the current data type is retained to generate the batch-consistent dataset.

[0017] In specific implementation, the following steps are used to calculate the coefficient of variation of the large sample historical test dataset after screening, and to dynamically adjust the coefficient of variation with a threshold to generate the adjusted coefficient of variation: Obtain historical data of at least 5 batches of unidirectional plates from the composite material system to be analyzed from the material certification database, and calculate the coefficient of variation of the strength data at 0° tension, 0° compression and ±45° shear for each batch. For the strength data of each data type in the historical data of the one-way plate, an initial correction threshold is obtained through the coefficient of variation. If the initial correction threshold is less than a preset threshold, the final correction threshold of the current data type is set to the preset threshold. If the initial correction threshold is greater than or equal to the preset threshold, the final correction threshold of the current data type is set to the initial correction threshold. The original coefficients of variation of the 0° tensile strength data, the 0° compressive strength data, and the ±45° shear strength data in the large sample historical test dataset after screening are calculated respectively. For the strength data of each data type in the large sample historical test dataset after screening, if the original coefficient of variation is less than the final correction threshold, the corrected coefficient of variation of the current data type is set to the final correction threshold, and the variance of the current data type is updated according to the corrected coefficient of variation. If the original coefficient of variation is greater than or equal to the final correction threshold, the original coefficient of variation is used as the corrected coefficient of variation.

[0018] In specific implementation, the following steps are used to calculate the weighting coefficient based on the layup ratio of the multidirectional laminate, and to calculate the B-baseline value of the aerospace composite material based on the statistical parameters, the corrected coefficient of variation, and the weighting coefficient: The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. ; Obtain the small sample mean of the small sample test dataset from the statistical parameters. Small sample variance and small sample size From the statistical parameters corresponding to the corrected coefficients of dispersion, obtain the large sample mean of the filtered large sample historical test dataset for each data type. Large sample corrected variance and large sample size ,in, i Index the data type of the large sample historical test dataset after filtering; calculate the mean ratio coefficient. ,in, The small sample size based on the small sample test dataset. n and the total degrees of freedom of the large sample historical test dataset after filtering. The two-dimensional one-sided tolerance coefficient is determined by an approximate formula. ; through the two-dimensional one-sided tolerance coefficient The weighting coefficients The mean ratio coefficient The variance The large sample corrected variance The small sample size and the large sample size The baseline value of B is calculated. ,in, m The total number of data types in the filtered large-sample historical test dataset.

[0019] In practice, the following steps are used to calculate the weighting coefficient based on the layup ratio of the multidirectional laminate. : The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. , This represents the corresponding number of layers in a multidirectional laminate.

[0020] In practice, after outputting the B benchmark value, the statistical stability of the B benchmark value is evaluated with confidence through the following steps: After outputting the benchmark value B, the statistical stability of the benchmark value B is assessed for confidence level: the variance of the benchmark value estimator is obtained through first-order Taylor expansion or Monte Carlo sampling. ,in, For the reference value of aerospace composite material B, The mean of the small sample test dataset. The variance of the small sample test dataset, The standard deviation of the small sample test dataset. For the sample size of the small sample test dataset, For the first i The standard deviation of the large sample historical test dataset after filtering, as described above. For the first i The corrected variance of the large sample historical test dataset after filtering, as described above. For the first i The sample size of large-sample historical experimental data i An index for the data type of the large sample historical test dataset. To obtain the partial derivative; through the variance Var ( B The relative standard deviation of the B benchmark value is calculated. ,in, B The B-baseline value for aerospace composite materials is set; a stability threshold is set. If the relative standard deviation is less than or equal to the stability threshold, the B-baseline value is output and marked as valid; if the relative standard deviation is greater than the stability threshold, the B-baseline value is output and a warning label is added.

[0021] One-way laminate mutation method (LVM) is based on the assumption that the variability of the intrinsic strength of a multi-way laminate (small) dataset is no greater than the variability of the intrinsic strength of a one-way laminate (large) dataset.

[0022] Based on this fundamental assumption, its B benchmark value is: (1) Where X, S, and N correspond to the mean, standard deviation, and number of samples, respectively, while subscripts 1 and 2 refer to small sample datasets (multidirectional laminates) and large sample datasets (unidirectional laminates), respectively. It is a two-dimensional one-sided tolerance coefficient, which is obtained from the two-sided statistical table, or it can be calculated by equations (2) and (3).

[0023] (2) (3) in, , The value is defined by the B benchmark. , .

[0024] However, existing unidirectional plate variation methods only utilize 0° tension and 0° compression from material-level tests, and their calculation results are highly dependent on the variability of 0° tensile and compressive strength. To fully utilize material-level data and consider the influence of different ply directions in the laminate, this embodiment of the invention adds ±45° shear test data to the historical data. Among them, the 90° ply causes matrix cracking and exits the main load-bearing capacity at extremely low strain levels, and its contribution to the ultimate strength is negligible, so the 90° tensile and compressive test data are not considered. At this time, equation (1) becomes: (4) in , , These are the mean, variance, and sample size of a small sample dataset, respectively. , , These are the mean, variance, and sample size of a large dataset, respectively. This corresponds to historical datasets from different experiments at the material level. Mean ratio coefficient .

[0025] Since the ply ratio affects the ultimate strength of the laminate, different weights are assigned to the degrees of freedom of the large sample dataset based on the relative proportions of 0° and ±45° plies. The weighting coefficients are: (5) in, These represent the corresponding number of layers in the laminate. The formula for calculating the B baseline value is: (6) It should be noted that due to the addition of weights, the large sample dataset and the small sample dataset no longer satisfy the assumption of homogeneity of variance. According to the Welch-Satterthwaite theorem, the effective degrees of freedom should be derived to correct the degrees of freedom in formulas (2) and (3). However, after verification in engineering practice, it was found that whether or not the correction is made has almost no impact on the final result. Therefore, the calculation can continue to be performed according to formulas (2) and (3).

[0026] In addition, the excessively low dispersion coefficient obtained in the qualification test was compensated by modifying the dispersion coefficient CV method to improve the ultimate load of the laminate to above 0.06. The modification rules are shown in Formula 7.

[0027] (7) Example: First, statistical calculations were performed on the test data of unidirectional composite plates for aerospace applications, including calculating the coefficients of variation for 0° tensile strength, 0° compressive strength, and ±45° shear strength. Then, based on Table 1, corresponding material-grade test data were selected, and statistical calculations were performed on the test data of multidirectional laminates. Further, depending on the needs, it was determined whether to correct the coefficients of variation, and finally, the B baseline value was calculated.

[0028] Table 1. Correspondence between the dispersion coefficients of one-way laminates and multi-way laminates

[0029] In this embodiment of the invention, a method for calculating the B-benchmark value of aerospace composite materials based on a small sample is provided. The method includes the following steps.

[0030] Step 1: Obtain data and establish corresponding relationships.

[0031] First, a small sample test dataset of the multidirectional laminate to be analyzed is obtained. This small sample test dataset contains at least five ultimate strength test values ​​of the multidirectional laminate, typically derived from a small number of tests conducted during structural design and validation. For example, an unnotched tensile test is performed on a certain type of composite laminate, yielding six ultimate strength values.

[0032] Then, the correspondence between the small-sample experimental dataset and the large-sample experimental dataset is determined. Specifically: When small sample test data comes from tensile tests of multidirectional laminates (including unnotched tensile, filled-hole tensile, and open-hole tensile), 0° tensile strength data and ±45° shear strength data are extracted from large sample historical test data.

[0033] When small sample test data comes from compression tests (including notched compression, filled-hole compression, open-hole compression, and mechanical connection) or shear tests (including in-plane shear) of multi-directional laminates, 0° compression strength data and ±45° shear strength data are extracted from large sample historical test data.

[0034] Large-sample historical test datasets are historical accumulated data from unidirectional laminates of the same material system, including 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. Each data type usually contains test values ​​from multiple batches, with a large sample size (e.g., 20 to 50 data points per type).

[0035] Step 2: Perform validity screening on the large sample historical test dataset, remove abnormal data and check the consistency between batches to generate a screened large sample historical test dataset.

[0036] This step processes each data type (0° stretch, 0° compression, ±45° shear) separately, specifically including two aspects: (1) Abnormal data removal.

[0037] Outlier detection is performed using the Maximum Normalized Residual (MNR) method. For a dataset of size m, the maximum normalized residual statistic is calculated as follows: , among which, among which, For the intensity data of the i-th data point, The sample mean. The standard deviation is denoted as .

[0038] Calculate the critical value: .

[0039] If MNR>CV MNR Then determine that The largest data point is considered an outlier and is removed. Repeat the above process iteratively (recalculating the mean and standard deviation after each removal) until no outliers are detected, obtaining an outlier-free dataset.

[0040] (2)Batch-to-batch consistency test.

[0041] Group the outlier-free dataset by batch and use the k-sample Anderson-Darling test to determine whether each batch comes from the same population. Calculate the Anderson-Darling statistic ADK and the critical value ADC. If ADK < ADC, accept the null hypothesis (each batch comes from the same population) and retain all data of this data type; if ADK ≥ ADC, determine that each batch of this data type does not come from the same population, mark this data type as unavailable, and do not process the data of this data type in subsequent steps.

[0042] After performing the above operations on all data types, merge all data types that pass the test to obtain a filtered large-sample historical test dataset.

[0043] Step 3: Calculate the coefficient of variation of the filtered large-sample historical test dataset and perform dynamic threshold correction on the coefficient of variation to generate a corrected coefficient of variation.

[0044] This step includes two sub-steps: (1)Determine the dynamic correction threshold.

[0045] Obtain the historical data of at least 5 batches of unidirectional plates in the composite material system to be analyzed from the material certification database (independent of the above filtered dataset and only used to determine the threshold). Calculate the coefficient of variation CV of the 0° tensile, 0° compressive, and ±45° shear strength data for each batch respectively batch , for each data type, take the 10th percentile of the coefficient of variation of each batch as the initial correction threshold CV th,init . Set a preset threshold CV min = 0.06 (this value is from the lower bound of the typical coefficient of variation in the composite material handbook). If CV th,init < 0.06, take the final correction threshold CV th = 0.06, otherwise take CV th = CV th,init .

[0046] (2)Correct the coefficient of variation of the current dataset.

[0047] Calculate the original coefficient of variation of the 0° tensile, 0° compressive, and ±45° shear strength data in the filtered large-sample historical test dataset respectively , where is the standard deviation, This is the mean. For each data type... i : If the original discrete coefficients Then the coefficient of variation of this type of data is corrected to And update the variance of this type of data based on the corrected coefficients of dispersion. : If the original discrete coefficients If so, the original discrete coefficients and original variance remain unchanged.

[0048] Output the corrected coefficients of dispersion for each class of data. and the corresponding corrected variance , as the corrected discrete coefficient.

[0049] Step 4: Calculate the statistical parameters of the small sample test dataset and the filtered large sample historical test dataset respectively.

[0050] Calculate the mean, variance, and sample size of the small sample trial dataset. Calculate the mean of each class in the filtered large sample historical trial dataset. Corrected variance and sample size ( i (Take 0° tension, 0° compression, or ±45° shear).

[0051] Step 5: Calculate the weighting coefficient based on the layup ratio of the multidirectional laminate.

[0052] Obtain the ply information of the multidirectional laminate and count the number of 0° ply and ±45° ply. Calculate the ply ratio weighting coefficient using the following formula: in, These are the corresponding number of layers in the laminate.

[0053] Step 6: Calculate the B benchmark value based on statistical parameters, corrected coefficients of dispersion, and weighting coefficients.

[0054] First, calculate the mean ratio coefficient for various types of large sample data: .

[0055] Secondly, determine the two-dimensional one-sided tolerance coefficient. . ,in, For degrees of freedom The 0.95 quantile of a noncentral t-distribution with a noncentral parameter λ. Degrees of freedom N2 represents the total sample size of the filtered large-sample historical test dataset. R =1.282.

[0056] Then, calculate the B reference value using the following formula: , where m is the total number of data types in the filtered large-sample historical test dataset.

[0057] Step 7: Output the baseline value B and evaluate its statistical stability with confidence.

[0058] After outputting the calculated baseline value BB, the variance Var(B) of the baseline value B is estimated using the first-order Taylor expansion method: ,in, The mean of the small sample test dataset. The variance of the small sample test dataset, The standard deviation of the small sample test dataset. For the sample size of the small sample test dataset, For the first i The standard deviation of the large sample historical test dataset after filtering, as described above. For the first i The sample size of large-sample historical experimental data i An index for the data type of the large sample historical test dataset. To obtain the partial derivative.

[0059] Calculate the relative standard deviation of the B benchmark value: Where B is the reference value for aerospace composite materials.

[0060] Set stability threshold RSD th =15% (This value is based on engineering experience and can be adjusted according to actual needs). If RSD ≤ RSD th If the baseline value is B, then output the baseline value and mark it as valid; if RSD > RSD th If the statistical stability is insufficient, output the B baseline value and add a warning label. If the statistical stability is insufficient, supplement with small sample test data to no less than 8 or adopt a more conservative design allowable value.

[0061] By following the steps above, high-precision and high-stability reference values ​​for aerospace composite materials can be obtained using only 5 or more small samples.

[0062] In this embodiment, a computer device is provided, such as... Figure 3 As shown, it includes a memory 301, a processor 302, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned method for calculating the B-baseline value of any aerospace composite material based on a small sample.

[0063] Specifically, the computer device can be a computer terminal, a server, or a similar computing device.

[0064] In this embodiment, a computer-readable storage medium is provided, which stores a computer program that performs the above-described method for calculating the B-benchmark value of any of the aerospace composite materials based on a small sample.

[0065] Specifically, computer-readable storage media include both permanent and non-permanent, removable and non-removable media, which can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable storage media do not include transient media, such as modulated data signals and carrier waves.

[0066] Based on the same inventive concept, this invention also provides a calculation device for the B-reference value of aerospace composite materials based on a small sample, as described in the following embodiments. Since the principle of the calculation device for the B-reference value of aerospace composite materials based on a small sample is similar to the calculation method for the B-reference value of aerospace composite materials based on a small sample, the implementation of the calculation device for the B-reference value of aerospace composite materials based on a small sample can refer to the implementation of the calculation method for the B-reference value of aerospace composite materials based on a small sample, and will not be repeated. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0067] Figure 4 This is a structural block diagram of a device for calculating the B-reference value of aerospace composite materials based on a small sample, according to an embodiment of the present invention. Figure 4 As shown, it includes: a dataset acquisition module 401, a dataset filtering module 402, a coefficient of variation correction module 403, and a baseline value calculation module 404. The structure is described below.

[0068] The dataset acquisition module 401 is used to acquire a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and acquire the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. The dataset filtering module 402 is used to perform validity filtering on the large sample historical test dataset, remove abnormal data and check the consistency between batches, and generate a filtered large sample historical test dataset. The dispersion coefficient correction module 403 is used to calculate the dispersion coefficient of the large sample historical test dataset after screening, and to perform dynamic threshold correction on the dispersion coefficient to generate the corrected dispersion coefficient. The benchmark value calculation module 404 is used to calculate the statistical parameters of the small sample test dataset and the screened large sample historical test dataset respectively, calculate the weight coefficient according to the layup ratio of the multidirectional laminate, and calculate the B benchmark value of the aerospace composite material based on the statistical parameters, the corrected dispersion coefficient and the weight coefficient.

[0069] In one embodiment, the dataset acquisition module includes: The first large sample acquisition unit is used to acquire 0° tensile strength data and ±45° shear strength data from the large sample historical test dataset when the data of the small sample test dataset comes from tensile tests of multidirectional laminates. The second large sample acquisition unit is used to acquire 0° compression strength data and ±45° shear strength data from the large sample historical test dataset when the data of the small sample test dataset comes from compression or shear tests of multidirectional laminates.

[0070] In one embodiment, the dataset filtering module includes: The loop unit performs the following operations on the large sample historical experimental data of each data type until all data types have been processed, and then merges all batch-consistent datasets to obtain a valid large sample historical experimental dataset: The residual calculation unit is used to calculate the maximum normed residual statistic and the corresponding critical value of the large sample historical experimental dataset. An anomaly removal unit is used to remove the current data point if the maximum normalized residual statistic of the data point is greater than the threshold value, and repeat the iteration until all data points have been judged and an anomaly-free dataset is generated. The test value calculation unit is used to group the abnormal dataset by batch and calculate the test statistic and critical value for each batch of the abnormal dataset using the k-sample Anderson-Darling test method. Inconsistent data units are removed. If the test statistic is greater than or equal to the threshold value, the batch of the current data type is determined to be inconsistent and marked as unusable. Otherwise, all data of the current data type is retained to generate the batch-consistent dataset.

[0071] In one embodiment, the module for correcting discrete coefficients includes: The discrete coefficient calculation unit is used to obtain historical data of at least 5 batches of unidirectional plates in the composite material system to be analyzed from the material certification database, and calculate the discrete coefficients of the strength data at 0° tension, 0° compression and ±45° shear for each batch. The discrete coefficient correction unit is used to obtain an initial correction threshold for the intensity data of each data type of the historical data of the one-way board through the discrete coefficient. If the initial correction threshold is less than a preset threshold, the final correction threshold of the current data type is set to the preset threshold. If the initial correction threshold is greater than or equal to the preset threshold, the final correction threshold of the current data type is set to the initial correction threshold. The unit for calculating the original discrete coefficients is used to calculate the original discrete coefficients of the 0° tensile strength data, the 0° compressive strength data, and the ±45° shear strength data in the large sample historical test dataset after screening, respectively; The unit for correcting the coefficient of variation is used to, for each type of intensity data in the filtered large-sample historical test dataset, if the original coefficient of variation is less than the final correction threshold, set the corrected coefficient of variation of the current data type to the final correction threshold and update the variance of the current data type according to the corrected coefficient of variation; if the original coefficient of variation is greater than or equal to the final correction threshold, use the original coefficient of variation as the corrected coefficient of variation.

[0072] In one embodiment, the benchmark value calculation module includes: The weighting coefficient calculation unit is used to calculate the weighting coefficients based on the layup ratio of the multidirectional laminate. ; The small sample parameter calculation unit is used to obtain the small sample mean of the small sample test dataset from the statistical parameters. Small sample variance and small sample size ; The large sample parameter calculation unit is used to obtain the large sample mean of the filtered large sample historical test dataset for each data type from the statistical parameters corresponding to the corrected coefficients of dispersion. Large sample corrected variance and large sample size ,in, i An index for the data type of the filtered large sample historical test dataset; The mean ratio calculation unit is used to calculate the mean ratio. ,in, ; The tolerance coefficient calculation unit is used for calculating the small sample size based on the small sample test dataset. n and the total degrees of freedom of the large sample historical test dataset after filtering. The two-dimensional one-sided tolerance coefficient is determined by an approximate formula. ; The reference value calculation unit is used to calculate the reference value using the two-dimensional one-sided tolerance coefficient. The weighting coefficients The mean ratio coefficient The variance The large sample corrected variance The small sample size and the large sample size The baseline value of B is calculated. ,in, m The total number of data types in the filtered large-sample historical test dataset.

[0073] In one embodiment, the weighting coefficient calculation unit is further configured to calculate the weighting coefficient based on the layup ratio of the multidirectional laminate. , This represents the corresponding number of layers in a multidirectional laminate.

[0074] In one embodiment, the above-described apparatus further includes a confidence assessment module.

[0075] In one embodiment, the confidence assessment module includes: The variance calculation unit is used to obtain the variance of the baseline value estimate B through first-order Taylor expansion or Monte Carlo sampling. ,in, For the reference value of aerospace composite material B, The mean of the small sample test dataset. The variance of the small sample test dataset, The standard deviation of the small sample test dataset. For the sample size of the small sample test dataset, For the first i The standard deviation of the large sample historical test dataset after filtering, as described above. For the first iThe corrected variance of the large sample historical test dataset after filtering, as described above. For the first i The sample size of large-sample historical experimental data i An index for the data type of the large sample historical test dataset. To obtain the partial derivative; The relative standard deviation calculation unit is used to calculate the variance. Var ( B The relative standard deviation of the B benchmark value is calculated. ,in, B The B-baseline value for aerospace composite materials; The validity judgment unit is used to set a stability threshold. If the relative standard deviation is less than or equal to the stability threshold, the B benchmark value is output and marked as valid. A warning labeling unit is used to output the B benchmark value and attach a warning label if the relative standard deviation is greater than the stability threshold.

[0076] The embodiments of the present invention achieve the following technical effects: The embodiments of the present invention require only a small sample of test data from 5 multi-directional laminates to obtain a relatively reliable reference value for B of aerospace composite materials. This avoids the rigid requirement of a large number of samples for the traditional single-point method (at least 18 samples) or multi-environment sample merging method, greatly reducing the workload of sample preparation and testing, thereby significantly reducing the R&D cost of aerospace composite materials and shortening the model development cycle.

[0077] By introducing ±45° shear strength data, a weighting coefficient based on ply ratio, and a dynamic threshold correction for the dispersion coefficient, this invention overcomes the shortcomings of traditional unidirectional plate variation methods, which suffer from large fluctuations in calculation results due to excessive reliance on 0° tensile / compressive strength variability and overly conservative approaches in compression mode. Practical verification shows that the calculation results of this invention are closer to the B baseline value directly calculated from five batches of experimental data, and its accuracy and stability under small sample conditions are significantly better than existing small-sample methods such as the minimum batch method.

[0078] The embodiments of this invention have a clear principle and well-defined steps. The calculation can be automatically completed by simply following the process of data acquisition, validity screening, dynamic discretization correction, ply weight calculation, formula solution, and confidence assessment. The computational efficiency is far higher than that of traditional single-point methods and multi-environment merging methods. Furthermore, the method incorporates engineering judgment logic (such as automatic switching prompts when there are no 0° plies, forced elevation when the dispersion coefficient is too low, and warnings when the relative standard deviation is too large), making it easy for engineering designers to use directly and possessing good practical engineering value.

[0079] In this embodiment of the invention, the maximum normalized residual (MNR) method is used to remove outlier data before calculation, and k-sample Anderson data is used. The Darling test verifies the consistency between batches of data, effectively eliminating the adverse effects of outliers and batch differences on statistical results, ensuring the quality of input data, and thus improving the reliability of the final B baseline value.

[0080] This invention employs historical data from at least five batches of unidirectional plates, calculating the 10th percentile of their coefficient of variation as a dynamic correction threshold, rather than using a fixed 0.06. This allows the method to adapt to the inherent dispersion of different composite material systems. For materials with high fabrication dispersion, such as ceramic matrix composites, the threshold range can be further adjusted (e.g., 0.08–0.12) to avoid design risks caused by underestimating the coefficient of variation.

[0081] This invention, while outputting the baseline value B, estimates the variance of the baseline value B through first-order Taylor expansion or Monte Carlo sampling, calculates the relative standard deviation (RSD), and compares it with a preset threshold (e.g., 15%), providing a clear indication of "valid" or "insufficient statistical stability, supplementing the sample is recommended." This evaluation function allows users to intuitively understand the reliability of the current small sample calculation results, providing an important reference for subsequent design decisions.

[0082] Finally, embodiments of the present invention can be embedded in computer devices, B / S architecture systems, or distributed computing platforms to achieve fully automated calculation from data entry to result output. The embodiments of its devices, computer equipment, and storage media further enhance the practicality and scalability of the solution, applicable to various aerospace composite material systems such as carbon fiber, glass fiber, and ceramic matrix, as well as various test types including notched, open-hole, filled-hole, and mechanical connections.

[0083] Obviously, those skilled in the art should understand that the modules or steps of the above-described embodiments of the present invention can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of the present invention are not limited to any particular hardware and software combination.

[0084] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calculating the B-benchmark value of aerospace composite materials based on a small sample, characterized in that, include: Obtain a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and obtain the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. The large sample historical test dataset is subjected to validity screening, outliers are removed and batch consistency is checked to generate a screened large sample historical test dataset. Calculate the coefficient of variation of the large sample historical test dataset after screening, and perform dynamic threshold correction on the coefficient of variation to generate the corrected coefficient of variation; Statistical parameters are calculated for the small sample test dataset and the filtered large sample historical test dataset, respectively. Weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. Based on the statistical parameters, the corrected coefficient of variation, and the weighting coefficients, the B-baseline value of the aerospace composite material is calculated.

2. The method for calculating the B benchmark value as described in claim 1, characterized in that, Determine the correspondence between small-sample and large-sample experimental datasets, including: When the data in the small sample test dataset comes from tensile tests of multidirectional laminates, the 0° tensile strength data and ±45° shear strength data in the large sample historical test dataset are obtained. When the data in the small sample test dataset comes from compression or shear tests of multidirectional laminates, the 0° compression strength data and ±45° shear strength data in the large sample historical test dataset are obtained.

3. The method for calculating the B benchmark value as described in claim 1, characterized in that, The large-sample historical experimental dataset is subjected to validity screening, outlier data is removed, and batch consistency is tested to generate a filtered large-sample historical experimental dataset, including: For each type of large-sample historical experimental data, perform the following operations until all data types have been processed, then merge all batch-consistent datasets to obtain a valid large-sample historical experimental dataset: Calculate the maximum normalized residual statistic and the corresponding critical value of the large sample historical test dataset; If the maximum normed residual statistic of a data point is greater than the critical value, the current data point is removed, and the iteration is repeated until all data points have been judged and an anomaly-free dataset is generated. The dataset without anomalies is grouped into batches, and the test statistic and critical value for each batch of the dataset without anomalies are calculated using the k-sample Anderson-Darling test method. If the test statistic is greater than or equal to the critical value, the batch of the current data type is determined to be inconsistent and marked as unusable; otherwise, all data of the current data type is retained, and the batch-to-batch consistent dataset is generated.

4. The method for calculating the B benchmark value as described in claim 1, characterized in that, Calculate the coefficients of variation of the filtered large-sample historical test dataset, and perform dynamic threshold correction on the coefficients of variation to generate corrected coefficients of variation, including: Obtain historical data of at least 5 batches of unidirectional plates from the composite material system to be analyzed from the material certification database, and calculate the coefficient of variation of the strength data at 0° tension, 0° compression and ±45° shear for each batch. For the intensity data of each data type of the historical data of the one-way board, an initial correction threshold is obtained through the discrete coefficient. If the initial correction threshold is less than the preset threshold, the final correction threshold of the current data type is set to the preset threshold. If the initial correction threshold is greater than or equal to the preset threshold, the final correction threshold of the current data type is set to the initial correction threshold. Calculate the original coefficients of variation for the 0° tensile strength data, the 0° compressive strength data, and the ±45° shear strength data in the large sample historical test dataset after screening; For the intensity data of each data type in the large sample historical test dataset after filtering, if the original coefficient of variation is less than the final correction threshold, the corrected coefficient of variation of the current data type is set as the final correction threshold, and the variance of the current data type is updated according to the corrected coefficient of variation. If the original coefficient of variation is greater than or equal to the final correction threshold, the original coefficient of variation is used as the corrected coefficient of variation.

5. The method for calculating the B benchmark value as described in claim 1, characterized in that, The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate, and the B-baseline value of the aerospace composite material is calculated based on the statistical parameters, the corrected coefficient of variation, and the weighting coefficients, including: The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. ; Obtain the small sample mean of the small sample test dataset from the statistical parameters. Small sample variance and small sample size ; From the statistical parameters corresponding to the corrected coefficients of dispersion, obtain the large sample mean of the filtered large sample historical test dataset for each data type. Large sample corrected variance and large sample size ,in, i An index for the data type of the filtered large sample historical test dataset; Calculate the mean ratio coefficient ,in, ; Based on the small sample size of the small sample test dataset n and the total degrees of freedom of the large sample historical test dataset after filtering. The two-dimensional one-sided tolerance coefficient is determined by an approximate formula. ; Through the two-dimensional one-sided tolerance coefficient The weighting coefficients The mean ratio coefficient The variance The large sample corrected variance The small sample size and the large sample size The baseline value of B is calculated. ,in, m The total number of data types in the filtered large-sample historical test dataset.

6. The method for calculating the B benchmark value as described in claim 5, characterized in that, The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. ,include: The weighting coefficients are calculated based on the layup ratio of the multidirectional laminate. , This represents the corresponding number of layers in a multidirectional laminate.

7. The method for calculating the B benchmark value as described in any one of claims 1 to 6, characterized in that, Also includes: After outputting the B benchmark value, a confidence level assessment is performed on the statistical stability of the B benchmark value: The variance of the baseline estimate B is obtained through first-order Taylor expansion or Monte Carlo sampling. ,in, For the reference value of aerospace composite material B, The mean of the small sample test dataset. The variance of the small sample test dataset, The standard deviation of the small sample test dataset. For the sample size of the small sample test dataset, For the first i The standard deviation of the large sample historical test dataset after filtering, as described above. For the first i The corrected variance of the large sample historical test dataset after filtering, as described above. For the first i The sample size of large-sample historical experimental data i An index for the data type of the large sample historical test dataset. To obtain the partial derivative; Through the variance Var ( B The relative standard deviation of the B benchmark value is calculated. ,in, B The B-baseline value for aerospace composite materials; Set a stability threshold. If the relative standard deviation is less than or equal to the stability threshold, output the B benchmark value and mark it as valid. If the relative standard deviation is greater than the stability threshold, the B benchmark value is output and a warning label is added.

8. A device for calculating the B-benchmark value of aerospace composite materials based on a small sample, characterized in that, include: The dataset acquisition module is used to acquire a small sample test dataset of the multi-directional laminate to be analyzed, determine the correspondence between the small sample test dataset and the large sample test dataset, and acquire the corresponding large sample historical test dataset of the unidirectional laminate based on the correspondence. The small sample test dataset includes at least 5 ultimate strength test values ​​of the multi-directional laminate, and the data types of the large sample historical test dataset include 0° tensile strength data, 0° compressive strength data and ±45° shear strength data. The dataset filtering module is used to filter the large sample historical test dataset for validity, remove outliers and check batch consistency, and generate a filtered large sample historical test dataset. The dispersion coefficient correction module is used to calculate the dispersion coefficients of the large sample historical test dataset after screening, and to perform dynamic threshold correction on the dispersion coefficients to generate the corrected dispersion coefficients. The benchmark value calculation module is used to calculate the statistical parameters of the small sample test dataset and the screened large sample historical test dataset respectively, calculate the weighting coefficient based on the layup ratio of the multidirectional laminate, and calculate the B benchmark value of the aerospace composite material based on the statistical parameters, the corrected coefficient of dispersion and the weighting coefficient.

9. A computer 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 computer program, it implements the method for calculating the B-benchmark value of aerospace composite materials based on a small sample, as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that performs the calculation method for the B-baseline value of aerospace composite materials based on a small sample, as described in any one of claims 1 to 7.

Images