A product quality consistency inspection method

Abstract

This invention provides a product quality consistency inspection method. Before data comparison, the number of samples to be tested is differentiated to avoid unreasonable confidence regions caused by excessively large sample sizes. Secondly, when using the t-test statistic, the sample variance of the sample data is not considered, resulting in a more reasonable confidence region. Furthermore, this invention selects different confidence regions based on the importance of product quality parameters, making the final comparison results more accurate. This product quality consistency inspection method can be used to compare product parameters between a new production line and the original production line after product technology transfer. Based on the comparison results, it can accurately determine whether the products produced by the new production line are qualified after technology transfer, with small judgment errors and accurate results, providing an accurate method for judging whether products are qualified before and after technology transfer.

Description

Technical Field

[0001] This invention relates to the field of statistical methods, and more specifically to a method for testing the consistency of product quality. Background Technology

[0002] In the product manufacturing process, technology transfer is common. For example, the parameters of products produced by new machines need to be compared with those produced by old machines to verify whether the quality of products from the new machines is consistent with that of products from the old machines. Current technologies typically use methods such as t-tests and F-tests, or verify the consistency of the distributions of two sets of data (CDF cumulative distribution function, PDF probability density function) to perform this verification. However, existing methods all have certain drawbacks; the calculation results are not accurate enough, leading to incorrect judgments. Summary of the Invention

[0003] In view of the shortcomings of the prior art described above, this invention proposes a product quality consistency inspection method. This method compares the parameters of multiple products produced by a new machine with those produced by an old machine. Before comparison, the sample sizes are categorized. When the sample size is large, a reference value is set. During the comparison process, the sample size is recorded as a fixed reference value, avoiding unreasonable confidence regions caused by large sample sizes. Furthermore, when constructing the t-test statistic, this invention uses only the sample variance of the original sample data, without considering the sample variance of the sample data to be tested. Therefore, the statistic t does not change with changes in the sample variance of the sample data to be tested, resulting in a more reasonable confidence region. Furthermore, this invention uses t-tests and F-tests to obtain critical values ​​for the sample mean and sample variance of the test sample data, which can be considered to have no significant difference from the original sample data. Then, different constraints are set according to the importance of different parameters of the product to obtain confidence regions under different constraints. The greater the importance of the parameter to the consistency test, the stricter the constraint, and the smaller the corresponding confidence region, making the final comparison results more accurate.

[0004] To achieve the above and other related objectives, the present invention provides a product quality consistency inspection method, comprising the following steps:

[0005] 1) Obtain the parameters of multiple original products as the original sample dataset X1, and obtain the parameters of multiple products to be tested as the sample dataset X2 to be tested;

[0006] 2) Obtain the sample size n1 of the original sample dataset X1, and calculate the sample mean of the original sample data. and sample variance

[0007] 3) Obtain the sample size n2 of the dataset X2 to be tested, set a reference value n, and classify n2 according to the reference value n:

[0008] If n² ≥ n, then denote n² as n;

[0009] If n² < n, then count according to the actual quantity of n²;

[0010] 4)

[0011] The sample mean of the test sample data is calculated from the t-value when there is no significant change between the original sample dataset X1 and the test sample dataset X2.

[0012] 5) Construct the F-test statistic, and calculate the sample variance of the test sample data when there is no significant change between the original sample dataset X1 and the test sample dataset X2 from the F-value.

[0013] 6) Based on the sample mean in step 4) and the sample variance in step 5) Confidence regions are drawn that indicate no significant change between the test sample data and the original sample data. These confidence regions are then divided into first, second, third, and fourth confidence regions based on different constraints. The appropriate confidence regions are selected according to the importance level of the product parameters.

[0014] 7) Check whether the sample mean and sample variance of the sample data to be tested fall within the confidence region described in step 6).

[0015] Optionally, both the original sample dataset X1 and the sample dataset X2 to be tested follow a normal distribution.

[0016] Optionally, the reference value n is set according to the different accuracy requirements for different quality parameters.

[0017] Optionally, step 4) constructing the t-test statistic specifically includes: constructing the two-tailed quantile t-distribution with degrees of freedom df = n1 + n2 - 2 and a significance level α = 0.05. 0.05，n1+n2-2 And the two-sided quantile t 0.05，n1+n2-2 The value can be obtained by looking up a table.

[0018] Optionally, in step 5), the F-test statistic is: In the formula, The sample standard deviation of the original sample data. is the sample standard deviation of the data to be tested.

[0019] Optionally, step 5) constructing the statistic for the F-test specifically includes: the rejection region under the test level α is: F = [0, -F] α / 2，(n1-1,n2-1) ]∪[F α / 2，(n1-1,n2-1) [,+∞], where F α / 2 Let α be the critical point of the rejection region for the F-test. Given the significance level α = 0.05, the F-value can be obtained by referring to the F-distribution table. α / 2，(n1-1,n2-1) .

[0020] Optionally, the first confidence region is a straight line. straight line and coordinate system The closed region formed by the intersection of the two coordinate axes.

[0021] Optionally, the second confidence region is the function Y1: With a straight line and coordinate system ( S 2 The closed region formed by the intersection of the two coordinate axes.

[0022] Optionally, the third confidence region is the relationship between the function Y2 and the straight line. and coordinate system The closed region formed by the intersection of the two coordinate axes, where the function Y2 passes through the point... and points A monotonically decreasing weighted function of any degree n, where n is an integer greater than or equal to 2.

[0023] Optionally, the fourth confidence region is the function Y3: With a straight line and coordinate system The closed region formed by the intersection of the two coordinate axes.

[0024] The product quality consistency inspection method of the present invention has at least the following beneficial effects:

[0025] (1) Before conducting data comparison, the present invention first classifies the number of samples to be tested. When the number of samples is large, a reference value is set. During the data comparison process, the number of samples to be tested is recorded as the reference value, thus avoiding unreasonable confidence regions caused by the large amount of data of the samples to be tested.

[0026] (2) When constructing the t-test statistic, this invention only uses the sample variance of the original sample data and does not involve the sample variance of the sample data to be tested. Therefore, the statistic t does not change due to the change of the sample variance of the sample data to be tested, and a more reasonable confidence region can be obtained.

[0027] (3) The present invention obtains the critical values ​​of the sample mean and sample variance of the test sample data based on the t test and F test, which can be considered that there is no significant difference between the test sample data and the original sample data. Then, different constraints are set according to the importance of different quality parameters of the product to obtain the confidence region under different constraints, so as to make the final comparison result more accurate.

[0028] (4) The hypothesis testing-based data comparison method of the present invention can be used to compare the product parameters of the new production line and the original production line after product technology transfer. Based on the comparison results, it can accurately determine whether the products generated by the new production line are qualified after technology transfer. Moreover, the judgment error is small and the results are accurate, providing an accurate method for judging whether products are qualified before and after product technology transfer. Attached Figure Description

[0029] Figure 1 The diagram shown is a flowchart of the data comparison method provided in Example 1.

[0030] Figures 2a-2c This is a schematic diagram showing the confidence regions obtained by t-test and F-test in the prior art.

[0031] Figure 3a This is a schematic diagram of the confidence region obtained by the t-test in the prior art.

[0032] Figure 3b This is a schematic diagram showing the confidence region obtained by the t-test constructed through step S3 of Example 1.

[0033] Figure 4 The diagram shown is a schematic representation of the first confidence region provided in Embodiment 1.

[0034] Figure 5 The diagram shown is a schematic representation of the second confidence region provided in Embodiment 2.

[0035] Figure 6 This is a schematic diagram of the third confidence region provided in Example 3.

[0036] Figure 7 This is a schematic diagram of the fourth confidence region provided in Example 4.

[0037] Component designation explanation

[0038] C Confidence Region

[0039] C1 First Confidence Region

[0040] C2 Second Confidence Region

[0041] C3 Third Confidence Region

[0042] C4 Fourth Confidence Region Detailed Implementation

[0043] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0044] It should be noted that the illustrations provided in this embodiment are only schematic representations of the basic concept of the present invention. Although the illustrations only show components related to the present invention and are not drawn according to the actual number, shape and size of the components, the shape, quantity, positional relationship and proportion of each component can be arbitrarily changed under the premise of realizing the technical solution of this invention, and the layout of the components may also be more complex.

[0045] Example 1

[0046] This embodiment provides a method for inspecting product quality consistency, such as... Figure 1 As shown, the specific steps include the following:

[0047] Step S1: Obtain the parameters of multiple original products as the original sample dataset X1, and obtain the parameters of multiple products to be tested as the sample dataset X2 to be tested.

[0048] The original sample data can be the quality parameters of products produced by old machines or products produced by old production lines, depending on the specific circumstances. As an example, the parameters of the original product and the parameters of the sample to be inspected may be the same, such as dimensional parameters, performance parameters, etc., which will not be listed here.

[0049] In this embodiment, the original sample dataset follows a standard normal distribution X1~N(0,1).

[0050] Step S2: Obtain the sample size n1 of the original sample dataset X1, and calculate the sample mean of the original sample data. and sample variance

[0051] As an example, the sample size n1 of the original sample dataset X1 is obtained by statistical analysis, according to the formula... Calculate the sample mean of the original sample dataset. According to the formula Calculate the sample variance of the original sample dataset. In the formula X i This refers to individual sample data from the original sample dataset.

[0052] Step S3: Calculate the sample size n2 of the dataset X2 to be tested, set a reference value n, and classify n2 according to the reference value n: if n2≥n, then record n2 as n; if n2<n, then count according to the actual number of n2.

[0053] In t-tests and F-tests, when determining whether there are significant differences in the sample mean and sample variance between the original sample dataset X1 and the sample dataset X2 to be tested, the sample size n2 of the sample dataset X2 to be tested has a significant impact on the determination of the critical value. As an example, Figures 2a-2c The figures show the confidence regions C for when the sample size of the data to be tested is 30, 50, and 200, respectively, considering both t-tests and F-tests, to determine that there is no significant difference between the two groups of data. It can be seen that as the sample size increases, the confidence region C gradually decreases. When the sample size of the data to be tested is 200, as shown... Figure 2c As shown, the critical values ​​for sample mean and sample variance are relatively small. That is, only when the sample mean and sample variance of the data to be tested are not significantly different from those of the original sample data can it be considered that there is no significant change between the two sets of data. If the number of data to be tested continuously increases, the confidence region C tends to zero, which is obviously unreasonable. Therefore, after obtaining the sample size n2 of the dataset X2 to be tested, it is necessary to classify it into large and small sample data. When the sample size n2 of the data to be tested is greater than or equal to n, it is considered large sample data; otherwise, it is considered small sample data. If the data to be tested is large sample data, the number of data to be tested is recorded as n in the subsequent calculation of the confidence region; conversely, if the data to be tested is small sample data, it is recorded according to its actual number, i.e., n2.

[0054] To determine a reasonable value for n, we first assume that the sample variances of the test sample data and the original sample data are equal, and set the significance level α = 0.05. We then use a t-test to calculate the maximum difference between the sample mean of the test sample data and the sample mean of the original sample data before considering that there is no significant difference between the two sets of data. We assume that the original sample dataset X1 ~ N(0,1), and the test sample dataset... Table 1 lists the critical sample means of the sample data to be tested. As the sample size changes, it can be seen that when the sample size is less than 50, the critical sample mean decreases with increasing sample size. There are relatively obvious changes, and when the sample size is greater than or equal to 50, the critical sample mean decreases as the sample size increases. The changes are no longer obvious.

[0055] Table 1

[0056]

[0057] Secondly, assuming the sample means of the data to be tested are equal to those of the original sample data, and setting the significance level α = 0.05, the F-test is used to calculate the maximum difference between the sample variances of the data to be tested and the original sample data at which the two sets of data can be considered to have no significant difference. Assume the original sample dataset X1 ~ N(0,1), and the sample dataset to be tested X2 ~ N(0,S). 2 Table 2 lists the critical sample variance S of the data to be tested. 2 As the sample size changes, it can be seen that when the sample size is less than 50, the critical sample variance S increases with the increase of the sample size. 2 There are relatively obvious changes, and when the sample size is greater than or equal to 50, the critical sample variance S increases with the increase of the sample size. 2 The changes are no longer obvious.

[0058] Table 2

[0059] Sample size 10 20 30 40 50 60 70 80 <![CDATA[Critical sample mean S 2 > 1.7829 1.4725 1.3641 1.3055 1.2678 1.2410 1.2207 1.2047

[0060] As can be seen from Tables 1 and 2, regardless of the critical sample mean Or the critical sample variance S 2 When the sample size is greater than or equal to 50, the critical sample mean increases with the increase of the sample size. and critical sample variance S 2 The changes are relatively slow. Therefore, the value of n is set to 50. When n2 ≥ 50, n2 is recorded as 50; when n2 < 50, it is recorded according to the actual quantity of n2.

[0061] Step S4: Construct the t-test statistic and calculate the sample mean of the test sample data when there is no significant change between the original sample dataset X1 and the test sample dataset X2 from the t-value.

[0062] In existing techniques, the statistic for constructing a t-test of the sample mean is generally as follows: In a t-test, at a certain significance level α, the largest sample mean difference between two groups of data can be considered to have no significant difference. Subject to the sample variance of the sample data to be tested Impact. For example... Figure 3a As shown, the number of sample data to be tested is 30. Assume the original sample dataset X1~N(0,1), and the sample data to be tested... At a certain significance level α = 0.05, the confidence region C in which the sample mean of the test sample data and the original sample data are considered to be not significantly different can be considered. Figure 3a As shown, it can be seen that on the boundary line L of the confidence region C, the sample variance... With sample mean As the boundary line L increases, the difference between the sample mean of the test sample data and the sample mean of the original sample data also increases. However, it is unreasonable to assume that there is no significant difference between the two sets of data at this time.

[0063] Therefore, in this embodiment, the t-test statistic, at a certain significance level α, can be considered as the maximum difference between the sample means of the two groups of data that does not show a significant difference. Subject only to the sample variance of the original sample data Impact. For example... Figure 3b As shown, the number of sample data to be tested is 30. Assume the original sample dataset X1~N(0,1), and the sample data to be tested... At a certain significance level α = 0.05, the confidence region C in which the sample mean of the test sample data and the original sample data are considered to be not significantly different can be considered. Figure 3b As shown, it can be seen that on the boundary line L' of the confidence region C at this time, the sample mean of the data to be tested is a fixed value. That is to say, under the t-test condition, the difference between the sample mean of the data to be tested and the sample mean of the original data does not exceed [a certain value]. Only when the two sets of data are considered to have no significant difference can it be concluded that there is no significant difference between them.

[0064] Next, the sample mean of the test sample data is calculated from the constructed t-value, which best represents the assumption that there is no significant change between the original sample dataset X1 and the test sample dataset X2. As an example, construct the two-tailed quantile t of the t-distribution with degrees of freedom df = n1 + n2 - 2 and a test level α = 0.05. 0.05，n1+n2-2 When |t|>t 0.05，n1+n2-2 When the two-sided quantile t is reached, it is considered that the sample means of the two sets of data have changed significantly; otherwise, it is considered that the sample means of the two sets of data have not changed significantly. 0.05，n1+n2-2 The value can be found in Table B1 of Appendix B of GJB736.8-90, or in Table A.2 of Appendix A of GB / T4889-2008 "Statistical processing and interpretation of data: estimation and testing of mean and variance of normally distributed samples".

[0065] Step S5: Construct the F-test statistic and calculate the sample variance of the test sample data when there is no significant change between the original sample dataset X1 and the test sample dataset X2 from the F-value.

[0066] As an example, to calculate the maximum sample variance difference that would indicate no significant difference between the tested sample data and the original sample data under the F-test, the F-test statistic is constructed as follows: In the formula, The sample standard deviation of the original sample data. Let F be the sample standard deviation of the data to be tested. Specifically, the rejection region under the test level α is: F = [0, -F]. α / 2，(n1-1,n2-1) ]∪[F α / 2，(n1-1,n2-1) [,+∞], where F α / 2 Let α be the critical point of the rejection region for the F-test. Given the significance level α = 0.05, the F-value can be obtained by referring to the F-distribution table. α / 2，(n1-1,n2-1) Next, the sample variance of the test sample data is calculated from the F-value, which best represents the assumption that there is no significant change between the original sample data X1 and the test sample data X2.

[0067] Step S6: Based on the sample mean from step S4 and the sample variance in step S5 Confidence regions are drawn that indicate no significant change between the test sample data and the original sample data. These confidence regions are then divided into first, second, third, and fourth confidence regions based on different constraints. The appropriate confidence regions are selected according to the importance level of the product parameters.

[0068] In this embodiment, the confidence region is the first confidence region C1. The original sample data follows a standard normal distribution N(0,1), and the sample mean of the sample data to be tested is less than or equal to... Sample variance less than or equal to This means that there is no significant difference from the original sample data. As an example, a model is constructed using the sample mean... The horizontal axis represents the sample variance S. 2 A coordinate system with the vertical axis as the coordinate system, such as Figure 4 As shown, the first confidence region C1 is a straight line. straight line and coordinate system The closed region formed by the intersection of the two coordinate axes is point A(0,0). The area enclosed.

[0069] Step S7: Check whether the sample mean and sample variance of the sample data to be tested fall within the confidence region described in step S6;

[0070] In this embodiment, the sample mean and sample variance of the sample data to be tested are checked to see if they fall within the confidence region C1 described in step S6. If they do not fall within the confidence region, it is considered that the sample data to be tested has a significant change from the original sample data; otherwise, it is considered that there is no significant change between the two.

[0071] The above method can be used to compare the product parameters of the new production line and the original production line after product technology transfer. Based on the comparison results, it can be accurately determined whether the products generated by the new production line are qualified after the technology transfer. The first confidence region C1 mentioned in this embodiment can be used to judge non-critical product parameters. For example, parameters such as the size of the wafers generated by the wafer production line that do not significantly affect the essential characteristics of the wafers. The sample mean and sample variance of the wafer sizes produced by the production lines before and after the transfer are checked to see if they fall within the first confidence region C1. If they do not fall within the first confidence region C1, it is considered that the wafer sizes have changed significantly before and after the transfer, and the product technology transfer is unsuccessful; conversely, if they fall within the first confidence region C1, it is considered that there is no significant change, and the product technology transfer is successful.

[0072] Example 2

[0073] This embodiment also provides a product quality consistency inspection method, which still refers to... Figure 1 The similarities between this embodiment and Embodiment 1 will not be repeated here. The difference lies in that, in this embodiment, the confidence region mentioned in step S6 is the second confidence region C2, as shown below. Figure 5 As shown, the second confidence region C2 is the function Y1: With a straight line and coordinate system The closed region formed by the intersection of the two coordinate axes is point A(0,0). The area enclosed.

[0074] As an example, assume that the sample dataset X2 to be tested and the original sample dataset X1 show no significant difference after an F-test. That is, when the sample means of the sample dataset X2 to be tested and the original sample dataset X1 are equal (both are 0), the sample variance of the sample data to be tested is at its maximum. Right now Figure 5 Point B is shown; assuming that the test dataset X2 and the original sample dataset X1 show no significant difference after a t-test, that is, when the sample variances of the test dataset X2 and the original sample dataset X1 are equal (both 1), the maximum sample mean of the test dataset is... Right now Figure 5 Point E is shown. The function of the line connecting points B and E is Y1: The second confidence region C2 is the intersection of the function Y1 and the line. and coordinate system The closed region formed by the intersection. Compared with the first confidence region C1 provided in Example 1, the second confidence region C2 provided in this example has a smaller area. That is to say, compared with Example 1, the constraint on the sample variance and sample mean of the sample data to be tested is relatively stronger, and therefore it is suitable for parameters with relatively high requirements for precision and accuracy.

[0075] The second confidence region C2 described in this embodiment can be used to determine important parameters of the product. For example, parameters that affect the essential characteristics of the wafer, such as the flatness of the wafers produced by the wafer production line. The sample mean and sample variance of the flatness of the wafers produced by the production line before and after the transfer are examined to see if they fall within the second confidence region C2. If they do not fall within the second confidence region C2, it is considered that the wafer flatness has changed significantly before and after the transfer, and the product technology transfer is unsuccessful; conversely, if they fall within the second confidence region C2, it is considered that there is no significant change, and the product technology transfer is successful.

[0076] Example 3

[0077] This embodiment also provides a product quality consistency inspection method, which still refers to... Figure 1 The similarities between this embodiment and Embodiment 1 will not be repeated here. The difference lies in that, in this embodiment, the confidence region mentioned in step S6 is the third confidence region C3, as shown below. Figure 6 As shown, the third confidence region C3 is the intersection of the function Y2 and the line... and coordinate system The closed region formed by the intersection of the two coordinate axes, where the function Y2 passes through the point and points The weighted function is a monotonically decreasing function of degree n, where n is an integer greater than or equal to 2. Compared with the second confidence region C2 provided in Example 2, the third confidence region C3 provided in this example has a larger range. That is to say, compared with Example 2, the constraint on the sample variance and sample mean of the sample data to be tested is relatively weak.

[0078] The third confidence region C3 described in this embodiment can be used to determine general important parameters of the product. For example, parameters that can affect the essential characteristics of the wafer, such as the current characteristics of the wafers produced by the wafer production line. The sample mean and sample variance of the current characteristics of the wafers produced by the production line before and after the transfer are examined to see if they fall within the third confidence region C3. If they do not fall within the third confidence region C3, it is considered that the wafer current characteristics have changed significantly before and after the transfer, and the product technology transfer is unsuccessful; conversely, if they fall within the third confidence region C3, it is considered that there is no significant change, and the product technology transfer is successful.

[0079] Example 4

[0080] This embodiment also provides a product quality consistency inspection method, which still refers to... Figure 1 The similarities between this embodiment and Embodiment 1 will not be repeated here. The difference lies in that, in this embodiment, the confidence region mentioned in step S6 is the fourth confidence region C4, as shown below. Figure 7 As shown, the fourth confidence region C4 is the function Y3: With a straight line and coordinate system The closed region formed by the intersection of the two coordinate axes.

[0081] As an example, the original sample dataset X1 follows a standard normal distribution N(0,1). A statistically significant difference between the two datasets is considered to exist when the sample dataset X2 being tested overlaps with the original sample dataset X1 by at least 95%. Therefore, the distribution of the original sample dataset X1 can be likened to a circle with its center at 0 (located at the origin O) and a radius of 1 (the sample variance of the original sample dataset). Assuming that the sample dataset X2 being tested overlaps with the original sample dataset X1 by 95%, it can be likened to another circle with its center at 1 (the sample variance of the original sample dataset X1). Its radius is 2S 2 Under the above constraints, the circle representing the test sample dataset that overlaps 95% with the original dataset X1 is completely distributed within the circle representing the original sample dataset X1. The calculated sample mean of the test sample data is... and sample variance S 2 Satisfying function Y3: And the sample mean of the dataset to be tested is less than or equal to As an example, the confidence region C4 under this constraint is as follows: Figure 7 As shown, compared with Examples 1 to 3, the area of ​​the fourth confidence region C4 provided in this example is the smallest. Therefore, this example is suitable for parameters with strict requirements for precision and accuracy.

[0082] In summary, this invention proposes a method for product quality consistency testing. Before data comparison, the invention first categorizes the sample sizes to be tested. When the sample size is large (greater than or equal to 50), the number of samples to be tested is recorded as a fixed value during the data comparison process, avoiding unreasonable confidence regions caused by the large amount of data. Secondly, when calculating the t-test statistic, this invention only uses the sample variance of the original sample data, without considering the sample variance of the data to be tested. Therefore, the statistic t does not change with changes in the sample variance of the data to be tested, resulting in a more reasonable confidence region. Furthermore, this invention uses the t-test and F-test to obtain the critical values ​​of the sample mean and sample variance of the data to be tested, which indicate no significant difference between the data to be tested and the original sample data. Then, different constraints are set according to the importance of different product parameters, resulting in confidence regions under different constraints. The greater the importance of a parameter in the consistency test, the stricter the constraints, and the smaller the corresponding confidence region, making the final comparison results more accurate. The hypothesis testing-based data comparison method of this invention can be used to compare product parameters between the new production line and the original production line after product technology transfer. Based on the comparison results, it can accurately determine whether the products generated by the new production line are qualified after the technology transfer. Furthermore, the judgment error is small and the results are accurate, providing an accurate method for determining whether products are qualified before and after product technology transfer.

[0083] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for inspecting product quality consistency, characterized in that, Includes the following steps: 1) Obtain the quality parameters of multiple original products as the original sample dataset X1, and obtain the quality parameters of multiple products to be inspected as the sample dataset X2 to be inspected. 2) Obtain the sample size n1 of the original sample dataset X1, and calculate the sample mean of the original sample data. and sample variance 3) Obtain the sample size n2 of the dataset X2 to be tested, set a reference value n, and classify n2 according to the reference value n: If n² ≥ n, then denote n² as n; If n² < n, then count according to the actual quantity of n²; 4) Construct the t-test statistic The sample mean of the test sample data is calculated from the t-value when there is no significant change between the original sample dataset X1 and the test sample dataset X2. 5) Construct the F-test statistic, and calculate the sample variance of the test sample data when there is no significant change between the original sample dataset X1 and the test sample dataset X2 from the F-value. 6) Based on the sample mean in step 4) and the sample variance in step 5) Confidence regions are drawn that indicate no significant change between the test sample data and the original sample data. These confidence regions are then divided into first, second, third, and fourth confidence regions based on different constraints. The appropriate confidence regions are selected according to the importance level of the product parameters. 7) Check whether the sample mean and sample variance of the sample data to be tested fall within the confidence region described in step 6).

2. The product quality consistency inspection method according to claim 1, characterized in that, Both the original sample dataset X1 and the sample dataset X2 to be tested follow a normal distribution.

3. The product quality consistency inspection method according to claim 1, characterized in that, The reference value n is set according to the different accuracy requirements for different quality parameters.

4. The product quality consistency inspection method according to claim 1, characterized in that, Step 4) Constructing the t-test statistic specifically includes: constructing the two-tailed quantile t-distribution with degrees of freedom df = n1 + n2 - 2 and a significance level α = 0.

05. 0.05，n1+n2-2 And the two-sided quantile t 0.05，n1+n2-2 The value can be obtained by looking up a table.

5. The product quality consistency inspection method according to claim 1, characterized in that, In step 5), the F-test statistic is: In the formula, The sample standard deviation of the original sample data. is the sample standard deviation of the data to be tested.

6. The product quality consistency inspection method according to claim 1, characterized in that, Step 5) Constructing the F-test statistic specifically includes: the rejection region under the test level α is: F = [0, -F] α / 2，(n1-1,n2-1) ]∪[F α / 2，(n1-1,n2-1) [,+∞], where F α / 2 Let α be the critical point of the rejection region for the F-test. Given the significance level α = 0.05, the F-value can be obtained by referring to the F-distribution table. α / 2，(n1-1,n2-1) .

7. The product quality consistency inspection method according to claim 1, characterized in that, The first confidence region is a straight line. straight line and coordinate system ( S 2 The closed region formed by the intersection of the two coordinate axes.

8. The product quality consistency inspection method according to claim 1, characterized in that, The second confidence region is function Y1: With a straight line and coordinate system ( S 2 The closed region formed by the intersection of the two coordinate axes.

9. The product quality consistency inspection method according to claim 1, characterized in that, The third confidence region is the function Y2 and the straight line. and coordinate system ( S 2 The closed region formed by the intersection of the two coordinate axes of ), the function Y2 passes through the point (0, ) and point ( 1) any n-th degree monotonically decreasing weighted function, where n is an integer greater than or equal to 2.

10. The product quality consistency inspection method according to claim 1, characterized in that, The fourth confidence region is function Y3: With a straight line and coordinate system ( S 2 The closed region formed by the intersection of the two coordinate axes.

Images