Composite material reference value calculation method and calculation system

Abstract

The application discloses a composite material reference value calculation method, comprising the following steps: 1, retrieving and acquiring different batch composite material performance data in a database; 2, detecting different batch composite material performance data, screening out abnormal data not meeting preset conditions and correcting, outputting reasonable performance data; 3, according to the reasonable composite material performance data type, calculating the most reasonable composite material reference value; in step 3, the distribution type of data is determined according to the goodness-of-fit test, different normality one-sided tolerance coefficient calculation methods are introduced for different sample numbers of the parent body, the composite material reference value at different percentile points is calculated again, and the composite material reference value suitable for different fields is output. Thus, the sharp change of the absolute value of the one-sided tolerance coefficient caused by the change of the sample number is avoided, and the deviation error of the lower limit of the confidence of the calculation relative to the true value is reduced.

Description

Technical Field

[0001] This invention relates to the field of composite materials technology, and in particular to a method and system for calculating reference values ​​of composite materials. Background Technology

[0002] Composite materials are widely used in aerospace, wind power, and automotive fields, and different industries have different requirements for material benchmark values. CMH-17G defines composite material benchmark values ​​as statistically based material properties, with the measured value representing the 95% confidence lower limit at different percentiles of the specified population. The 95% confidence lower limits at the 1st and 10th percentiles are defined as benchmark values ​​A and B, respectively. DNVGL-ST-0376 defines the benchmark value for composite materials used in wind turbine blades as the 95% confidence lower limit at the 5th percentile. Composite material data can be divided into structural and unstructured data, and calculations can be performed based on different fitted distribution types. According to Chinese standard HB7618-2009, when sample data conforms to a normal distribution, the one-sided tolerance coefficient method is used to determine the benchmark value of the sample population. For actual components, due to cost and economic considerations, only a small number of samples can often be provided during testing. In this case, the one-sided tolerance coefficient method has poor calculation accuracy, leading to overly conservative results. Furthermore, for composite materials used in automobiles, not every structure faces the same impact and collision threat, while the requirements for aerospace and wind power materials are too stringent.

[0003] Furthermore, current software for calculating composite material benchmarks typically uses Excel, MATLAB, and SPSS. These data processing software programs are general-purpose and cannot achieve integrated processing of sample data detection and calculation, resulting in very low efficiency. Local computing programs also cannot effectively interface with web platforms and distributed computing environments. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a statistical method for the basic strength properties of automotive composite materials, and in particular, to provide a method and system for calculating the benchmark values ​​of automotive composite materials under different sample sizes.

[0005] To address the aforementioned technical problems, this invention provides a method for calculating the reference value of composite materials, the specific technical solution of which is as follows:

[0006] A method for calculating reference values ​​for automotive composite materials includes the following steps:

[0007] Step 1: Retrieve and obtain reasonable composite material performance data from different batches in the database; the performance data here can be any mechanical property data such as strength and modulus;

[0008] Step 2: Detect the performance data of composite materials from different batches, filter out unreasonable data that do not meet the preset conditions, correct the unreasonable performance data, and output reasonable performance data;

[0009] Step 3: Calculate the optimal baseline value for the composite material based on the reasonable distribution type of composite material performance data;

[0010] Furthermore, step 1 also includes: performing regularization determination, batch number determination, and dispersion coefficient determination on the data of different test types of composite materials to obtain reasonable composite material data.

[0011] Furthermore, step 2 includes the following steps:

[0012] Step 2.1: Use the maximum generalized residual (MNR) method to determine whether there are outliers in the sample data. If outliers are found, they need to be corrected and judged. If the outlier is handled, proceed to step 2.2; otherwise, correct the data using empirical and engineering judgment methods.

[0013] Step 2.2: Test whether the data from the k groups of samples come from the same population, and determine the magnitude of the Anderson-Darling test statistic ADK and the critical value ADC. If ADK < ADC, then the composite material performance data from each group of samples comes from the same population and is unstructured data. Proceed to step 2.3; otherwise, it is structured data. Proceed to step 2.4.

[0014] Step 2.3: Perform an OSL (Outcome-of-Size) test on the composite material performance data to determine which distribution the composite material performance data conforms to (normal, Weibull, or log-normal distribution). When the OSL is greater than 0.05, the normal distribution method is given priority for calculating the baseline value of the composite material, and the one-sided tolerance coefficient algorithm is used to calculate the baseline value in subsequent step 3. When the normal distribution model cannot fit the data, the Weibull and log-normal distribution methods are considered in subsequent step 3 to calculate the baseline value.

[0015] Step 2.4: For structured data that does not conform to any distribution type, calculate the baseline value of composite materials using the variance equivalence method.

[0016] Furthermore, in step 3, when the input data type conforms to a normal distribution, step 3 uses a one-sided tolerance coefficient algorithm to calculate the baseline value of the composite material. The one-sided tolerance coefficient is calculated using different formulas depending on the number of samples.

[0017] Let the sample size be n (n represents the total number of a random sample, n = n1 + n2 + ... + nk, specifically: the data consists of measurements of the same property of the same material under the same experimental conditions), let... The samples are x1, x2, ..., x n One-sided tolerance coefficients from simple random samples at different percentiles of a normal distribution are then used to proceed to step 3.3;

[0018] a. When n≤20

[0019] The general expression for calculating the one-sided tolerance factor of composite materials is as follows:

[0020]

[0021] In the formula: It is a non-central parameter; For a standard normally distributed random variable, in the calculation of the baseline value B, =1.282 (90% probability); n-1 is the degree of freedom. The general expression for the one-sided tolerance coefficient of composite materials can be derived from the quantile of the non-central t-distribution. Sure;

[0022] Because the calculation of quantiles for non-central t-distributions is complex in the above expression, and calculation errors may occur when the sample size is small, the quantiles for non-central t-distributions can be approximated as:

[0023] ,

[0024] In the formula, The quantiles of the standard normal distribution; ;

[0025] Therefore, the one-sided tolerance factor of the composite material in this case is:

[0026]

[0027] b. When 20 < n < 100, the one-sided tolerance factor of the composite material can be expressed as:

[0028]

[0029] In the formula, n is the number of samples; The standard normal skewness is a reliability of p. is the percentile of the t-distribution;

[0030] c. When n>100, the sample is a standard sample. An approximate formula for the one-sided tolerance coefficient is introduced. The one-sided tolerance coefficient of the composite material... The general expression is:

[0031]

[0032] In the formula: These are the coefficients under different probability conditions.

[0033] When calculating the baseline values ​​of A and B:

[0034]

[0035]

[0036] After obtaining the one-sided tolerance coefficient, the reference value of the composite material is further calculated. The general expression for the reference value of the composite material under the normal distribution model is:

[0037]

[0038] In the formula, The sample mean. is the one-sided tolerance coefficient for different sample sizes; S is the sample standard deviation.

[0039] A second aspect of the present invention also provides a reference value calculation system for automotive composite materials, the system comprising a data acquisition module, a data determination module, a data calculation module, and a data output module.

[0040] Furthermore, in one embodiment of the present invention, the data acquisition module is specifically used to retrieve and select composite material performance data from the database to obtain the composite material data.

[0041] Furthermore, in one embodiment of the present invention, the data determination module is specifically used to: determine the batch number and dispersion coefficient of the composite material data, determine abnormal data, determine inter-batch variability and goodness of fit, and when the above determination conditions cannot be met in the composite material data, correct the unreasonable data with a preset correction strategy.

[0042] Furthermore, in one embodiment of the present invention, the data calculation module is specifically used to: detect the reference value of the composite material and calculate the reference value of the composite material using a preset processing procedure.

[0043] Furthermore, in one embodiment of the present invention, the data output module is specifically used to output the calculation results of the composite material reference value.

[0044] Furthermore, the computing system described in this invention is applicable to a B / S architecture.

[0045] Compared with existing technologies, this invention determines the data distribution type based on goodness-of-fit tests and introduces different methods for calculating the one-sided tolerance coefficient of normality for different sample sizes. It then calculates the composite material benchmark values ​​at different percentiles and outputs benchmark values ​​applicable to various fields. This avoids the drastic change in the absolute value of the one-sided tolerance coefficient caused by variations in sample size, reducing the deviation error of the calculated lower confidence limit from the true value. Furthermore, the calculation program developed in this invention is suitable for calculating composite material benchmark values ​​in B / S architecture management systems and distributed computing systems, facilitating multi-platform portability and remote access.

[0046] Figure and Table Description

[0047] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0048] Figure 1 This is a flowchart illustrating the composite material reference value calculation method according to the present invention;

[0049] Figure 2 This is a schematic diagram of the algorithm flow for the one-sided tolerance coefficient of the normality of composite materials according to the present invention. Detailed Implementation

[0050] To more clearly illustrate the technical solutions in the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, any modifications or equivalent substitutions made based on these accompanying drawings without creative effort are within the protection scope of the present invention.

[0051] The software of this invention uses the TCL language for algorithm writing, and the computing software has the characteristics of openness, which can be called by programs of computing algorithms of different data types.

[0052] The embodiments of the present invention will now be described in further detail with reference to the accompanying drawings and tables.

[0053] like Figure 1 and Figure 2 As shown in the figure, an embodiment of the present invention discloses a method for calculating the reference value of composite materials, the method comprising:

[0054] Step 1: Obtain reasonable strength performance data for different batches of composite materials;

[0055] Step 2: Detect the strength data of composite materials from different batches, filter out unreasonable data that do not meet the preset conditions, correct the unreasonable data, and output reasonable performance data;

[0056] Step 3: Calculate the corresponding composite material reference value based on the reasonable distribution type of composite material strength performance data;

[0057] Step 2 includes:

[0058] Step 2.1 The maximum generalized residual (MNR) method is used to detect and determine whether there is unreasonable data in the sample data, and the abnormal data is corrected and judged until the data type required by the algorithm is obtained;

[0059] Step 2.2: Use the Anderson-Darling method to detect inter-batch variability, i.e., determine whether the data come from the same population. If the data come from the same population, the performance data is unstructured, proceed to step 2.3; if the performance data come from different populations, the performance data is structured, proceed to step 2.4; the data distribution type is detected using the goodness-of-fit (OSL) method, and supplemented by relevant experience and engineering methods.

[0060] Step 2.3: Use the goodness-of-fit (OSL) test to determine the data distribution type, and combine relevant experience and engineering methods to conduct supplementary tests; if the normal OSL of the composite material performance data is greater than 0.05, the normal distribution method should be given priority in calculating the baseline value of the composite material. When the normal distribution model cannot fit the data, the Weibull distribution and log-normal distribution methods should be considered.

[0061] Specifically, when the data conforms to a normal distribution and the normal distribution method is used to calculate the baseline value of the composite material, the one-sided tolerance coefficient is calculated using different methods depending on the number of samples, as follows:

[0062] set up The samples are x1, x2, ..., x n One-sided tolerance coefficient for simple random samples from different percentiles of a normal distribution, where n is the sample size.

[0063] a. When n≤20, the quantiles of the non-central t-distribution can be approximately expressed as:

[0064]

[0065] The one-sided tolerance factor of the composite material is:

[0066]

[0067] In the formula, The quantiles of the standard normal distribution; .

[0068] b. When 20 < n < 100, the one-sided tolerance factor of the composite material can be expressed as:

[0069]

[0070] In the formula, n is the number of samples; The standard normal skewness is a reliability of p. is the percentile of the t-distribution.

[0071] c. When n>100, the sample is a standard sample. An approximate formula for the one-sided tolerance coefficient is introduced. The one-sided tolerance coefficient of the composite material... The general expression is:

[0072]

[0073] In the formula: These are the coefficients under different probability conditions.

[0074] When calculating the baseline values ​​of A and B:

[0075]

[0076]

[0077] After obtaining the one-sided tolerance coefficient by selecting a suitable method, the baseline value of the composite material is further calculated and output. The general expression for the baseline value of the composite material under the normal distribution model is:

[0078]

[0079] In the formula, The sample mean. is the one-sided tolerance coefficient for different sample sizes; S is the sample standard deviation.

[0080] If the composite material performance data distribution type in step 2 does not conform to any distribution type of structural data, the variance equivalence method is used to calculate the baseline value of the composite material. If the OSL of all distributions is ≤0.05, the baseline value of the composite material is calculated according to the nonparametric algorithm.

[0081] According to the composite material reference value calculation method proposed in the embodiments of the present invention, users only need to provide the corresponding composite material mechanical data, and the algorithm program will automatically select reasonable data types and finally calculate the most reasonable composite material reference value.

[0082] Table 1 illustrates the performance data of composite materials with a composite normal distribution, and compares the one-sided tolerance coefficient calculated using the one-sided tolerance coefficient calculation method of this invention with the published values ​​when the sample size is in the range of 3-20.

[0083] Table 1

[0084]

[0085] As shown in Table 1, when the sample size is small, especially with fewer than 6 samples, the one-sided tolerance coefficient of the navigation mark changes drastically with the decrease in the sample size. The calculated absolute value of the one-sided tolerance coefficient is too large, and the calculated benchmark value has a large error relative to the true value. In contrast, the method described in this patent results in a more gradual change, with a stable increase in the absolute value of the one-sided tolerance coefficient, and the calculated benchmark value is closer to the actual true value.

[0086] Referring to Table 2, examples of composite material performance data with sample sizes ranging from 3 to 20 are provided. Using the method of calculating the one-sided tolerance coefficient of this invention, one-sided tolerance coefficients with different reliability levels are calculated.

[0087] Table 2

[0088]

[0089] As shown in Table 2, with a 95% confidence level, the algorithm of this patent can be used to change the percentile value and output one-sided tolerance coefficients at different percentiles, which are applicable to the benchmark requirements of composite materials in different industries. Furthermore, the absolute value of the one-sided tolerance coefficient changes more gradually, and the calculated benchmark value is closer to the true value.

[0090] In this invention, for composite material performance data that conforms to a normal distribution, different one-sided tolerance coefficient algorithms are selected according to the number of samples. At the same time, one-sided tolerance coefficients at different percentiles can be output according to the applicable fields of composite materials and the benchmark requirements of different fields, which can reduce the error between the calculated value and the true value of composite material performance.

[0091] This invention also provides a composite material reference value calculation system, which includes a data acquisition module, a data determination module, a data calculation module, and a data output module.

[0092] The data acquisition module is used to retrieve and obtain composite material mechanical strength performance data from the database.

[0093] The data determination module is used for outlier detection, batch-to-batch variability detection, and goodness-of-fit detection of composite material data, and to determine the data structure type.

[0094] The data calculation module is used to calculate the one-sided tolerance coefficient for different sample data sizes and to calculate the baseline value under different conditions using methods for processing unstructured data.

[0095] The output module is used to output the reference value of the composite material at different percentile values.

[0096] This system adopts the independently developed HTIMat system platform and is implemented using the TCL programming language.

[0097] The above description is only a preferred embodiment of the present invention and is not intended to limit 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 reference value of composite materials, comprising the following steps: Step 1: Retrieve and obtain mechanical property data of different batches of composite materials from the database; Step 2: Detect the mechanical property data of composite materials from different batches, filter out abnormal data that do not meet the preset conditions and correct them, and output reasonable mechanical property data of composite materials. Step 3: Calculate the baseline values ​​for the composite material based on the appropriate data type of composite material mechanical properties; If the data type of the composite material mechanical properties in step 3 conforms to unstructured data and satisfies the normal distribution model, then the normality one-sided tolerance coefficient algorithm is used to calculate the baseline value of the composite material. The algorithm for calculating the reference value of the composite material using the normality one-sided tolerance coefficient is characterized by the following steps: Step 3.1: Calculate the one-sided tolerance factor; Let the sample size be n, and let K be K. x The samples are x1, x2, ..., x n The one-sided tolerance coefficient is calculated using the corresponding expression based on the sample size, from simple random samples at different percentiles of a normal distribution. a. When n≤20, the quantiles of the non-central t-distribution can be approximately expressed as: ; where: is the degrees of freedom; is the non-central parameter; is the quantile of the standard normal distribution; ; The general expression of the unilateral tolerance factor of composite materials can be derived from the quantile of non-central t distribution is determined, and thus the calculation formula is as follows: ； b. When 20 < n < 100, the formula for calculating the one-sided tolerance factor of the composite material is as follows: ; In the formula, The standard normal skewness is a reliability of p. is the percentile of the t-distribution; c. When n>100, the sample is a standard sample. An approximate formula for the one-sided tolerance coefficient is introduced. The one-sided tolerance coefficient of the composite material... The general expression is: ; In the formula: are coefficients under different probability conditions, respectively; When calculating baseline value A: ; When calculating the baseline value B: ; Step 3.2: Based on the calculated one-sided tolerance coefficient, calculate and output the baseline value of the composite material. The general expression for the baseline value of the composite material under the normal distribution model is: ; wherein is the sample mean, is the one-sided tolerance factor for different sample sizes; S is the sample standard deviation.

2. The method of claim 1, wherein Step 1 includes regularization determination, batch number determination, and dispersion coefficient determination of the composite material data to obtain reasonable performance data.

3. The method of claim 1, wherein Step 2 includes: Step 2.1: Use the MNR method to determine whether there are outliers in the sample data, and correct and judge the outliers until the data type required by the algorithm is obtained; Step 2.2: Use the Anderson-Darling method to determine whether the data comes from the same parent. If the data comes from the same parent, it is unstructured data, and proceed to step 2.3 for further judgment; if the data comes from different parents, it is structured data, and proceed to step 2.4 for further judgment. Step 2.3: For unstructured data, use the goodness-of-fit test (OSL method) to determine the distribution type of the composite material mechanical property data; if it conforms to a normal distribution, perform the one-sided tolerance coefficient calculation in Step 3; if it does not conform to a normal distribution, use the variance equivalence method to calculate the composite material baseline value. Step 2.4: For structural data, use the Levene test to determine the variance equivalence of the composite material mechanical property data. If they are equal, then the variance analysis method is used to calculate the baseline value of the composite material. If they are not equal, then the calculation cannot be performed.

Images