Sampling procedures for inspection by variables - Part 4: Procedures for assessment of declared quality levels for mean
Some standard content:
ICS 03. 120. 30
National Standard of the People's Republic of China
GB/T 6378.42008
Replaces GB/T14990—1994
Sampling procedures for inspection by variablesPart 4: Procedures I'or assessment or deciared gualiiy levels fur rmean(ISO/WD 3951-4 :2007,NFQ)
2008-07-16 Issued
General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China Standardization Administration of China
2009-01-01 Implementation
2 Normative Reference Documents
3 Terms, Definitions and Symbols
3.1 Terms and Definitions
3.2 Symbols
Procedure for Implementing Sampling Inspection
Implementing Sampling Inspection
5.1 Determine the Inspection Population
Determine the Quality Characteristics of the Inspection Population to be Inspected
Determine the Type of Inspection Sampling Inspection
|Determine the verification sampling inspection method
Specify the claimed quality level
Specify the verification inspection level:
Determine the verification sampling plan...
Defective samples
Testing and calculation of samples
Determine whether the verification population is unqualified
Statistical interpretation of sampling inspection conclusions·
Retest and re-inspection
Use the actual quality level to judge,
6 Application examples
Inspection power of the verification sampling plan
8 Sampling table
Appendix A (Informative Appendix) Verification sampling power Formula for calculating the probability that the sample meets the requirements of the random inspection plan GB/T 6378.42008
CB/T 6378.4—2008
GB/T6378 Sampling Procedure for Metrology will be divided into the following parts. The expected structure corresponds to the international standards and national standards to be replaced:
Part 1: Provisions for a single sampling plan for batch inspection of a single quality characteristic and a single person's AQL retrieved by the acceptance quality limit (AQL) (IS03951-1:2005.IDT; Replaces (GB/T6378—2002) Part 2: Provisions for a single sampling plan for batch inspection of independent quality characteristics retrieved by the acceptance quality limit (AQL):-Share (corresponding to IS0 3951-2:2006)
--Part 3: Single sampling plan for batch inspection based on acceptance quality limit (AQL) (corresponding to 1S039513:2007)--Part 1: Procedure for assessment of claimed quality level of the mean (1S0/WD3951-4:2007, NEQ; replacement (G13/T 149001994)
…Part 5: Sequential sampling plan for metrological inspection based on acceptance quality limit (AQL.) (corresponding to ISO/FDIS3951-52005)
The technical part of GB/T 6378 is not equivalent to the draft international standard 1SO/WD3951-42007 "Metrological sampling inspection procedure Part 4: Procedure for assessment of claimed quality level of the mean". This part replaces GB/T14900-1994. The major differences between this part and GB/T14900-1994 are as follows: Change the monitoring quality level to the market quality level DCL; Change the supervision population to the verification population; Change the unqualified judgment value to the limit value; Change the monitoring population unqualified to the verification population unqualified, and change the supervision sampling inspection pass to the verification pass; Change the sampling inspection standard to the sample inspection standard; Add the discussion of "judgment by comparing the actual quality level with the claimed quality level"; Add the definition of re-test, re-inspection and verification inspection efficacy; Adjust the individual Sampling plan.
Appendix A of this part is an informative appendix.
This part is proposed by China Institute of Standardization. The technical part is under the jurisdiction of the National Technical Committee for Standardization of Statistical Methods. Drafting units of this part: Wuxi Product Quality Supervision and Inspection Institute, China Institute of Standardization, Dongguan Administration for Industry and Commerce, China Aerospace Science and Industry First Research Institute, PLA Ordnance Engineering College, Institute of Mathematics and Systems Science, Chinese Academy of Sciences. The main drafters of this part: Wu Jianguo, Ding Zhenfan, Ying Haibin, Gu Qin, Chen Huaying, Ding Wenxing, Zhang Juzhu, Chen Min. The previous versions of the standards replaced by this part are: GB/T 14900—1994
GB/T 6378.4—2008
The scope of GB/T 6378.4 is different from that of Part 1 to Part 3 of GB/T 6378. Part 1 to Part 3 of GB/T 6378 specifies a system of acceptance sampling procedures applicable to bilateral collaboration between two parties (e.g., a manufacturer and a user). Acceptance sampling procedures are used as rules for determining whether to accept a batch of products after inspecting samples of the batch. Therefore, these procedures do not define the quality level of the batch under any form. In acceptance sampling, it is considered that there is no clear boundary between the quality level of acceptable batches and unacceptable batches. For the procedures in Part 2 to Part 3 of GB/T 6378, the agreed acceptance quality limit is the process average quality level that can be tolerated when submitting a series of consecutive batches. GB/T The transfer rules and sampling plans in 6378.1 are designed to encourage producers to produce products with a process average quality level that is better than the selected AQL. The procedures in Parts 1 and 3 of GB/T 6378 apply to acceptance sampling and are not applicable to verifying the nominal quality of a verification population during an assessment or audit. The main purpose is that Parts 1 and 3 of GB/T 6378 are used to search for acceptance quality limits, which are only relevant for the actual purpose of acceptance sampling and thus balance the two types of risks. The sample inspection procedures specified in this part of GB/T 6378 are designed for the sample inspection required in a formal assessment. When implementing this form of inspection, the responsible department must consider the risk of making incorrect conclusions and take such risks into account in arranging and conducting the assessment (or audit): This part of GB/T 6378 designs some rules so that when the actual quality level of the verification population actually meets the market quality level, there is a small risk of judging that the verification population is not conforming. For example, when the actual quality level of the inspection population does not meet the claimed quality level, the risk of judging the inspection to pass is also very small. There must be more samples as much as possible. In order to minimize the sample size, it is allowed that when the actual quality level does not meet the claimed quality level, the risk of judging the inspection to pass is high. The judgment period reflects the imbalance of the risks of making different wrong conclusions. When the inspection population is judged to be unqualified by the sampling results, there is a great deal of confidence that "the actual quality level of the inspection population is inferior to the claimed quality level." When the inspection is judged to pass by the sampling results, it is considered that "for this limited group, it is not found that the actual quality level of the inspection population is inferior to the claimed quality level." Therefore, when the sample size is small, the responsible department is not responsible for confirming the inspection information. 5
1 Scope
Metrological sampling inspection procedures
Part 4, Assessment procedures for claimed quality levels of mean values
GB/T6378.4—2008
This part of GB/TS378 specifies the procedure for the measurement of a single sampling inspection with the average value of a certain quality characteristic of the verification population as the quality indicator.
The sampling plan and assessment procedure specified in this part of GB/T6378 are used to assess whether the quality level of a certain population (batch or process, etc.) does not meet a certain claimed quality level. The sampling plan specified in this part controls the risk of rejecting a certain qualified population (batch) to 5%. The procedures of this part of GB/T6378 are not applicable to the acceptance sampling of batches, but are only applicable to various forms of quality inspection, that is, the situation where the inspection results of the sample are used to indicate whether a certain verification population does not meet the claimed quality level. The quality characteristics of the inspected product should obey or approximately obey the state distribution
This part of GB/T6378 is applicable to the situation where a random sample composed of a number of unit products can be drawn from the verification population. The sampling scheme provided in this part of GB/T 6378 can be used for (but not limited to) the inspection of the following products, for example: "final product; - parts and raw materials; - operation; - finished product: - inventory; - maintenance operation; - data and records: - management procedures. 2 Normative references The clauses in the following parts become the clauses of this part through the reference of this part of GB/T 6378. For all references with the reference period II, all subsequent references shall be subject to the provisions of this part. The amendments (excluding errata) or revisions to this document are not applicable to this part. However, parties to agreements based on this part are encouraged to investigate whether the latest version of this document can be used. For any referenced document without a date, the latest version does not apply to this part. CB/T4882 Design processing and accuracy inspection of data (GB/T1882-2001, 1S05179:1997.1LTGB/T6378.1: Designed sampling inspection procedures Part 1: Sampling plan for batch inspection of single quality characteristics and single AQL based on acceptance quality limit (AQL) (GB/T 6378.1—2008, ISO 3951--:2005, IDT) G3/10111 Generation of random numbers and their application in product quality inspection G13/T1630S Procedure for assessment of claimed quality level Re-inspection and re-testing IS03534-1:2000 Statistical vocabulary and symbols Part 1: General statistical terms and terms used in probability ISO) 3534-2:2006 Statistical vocabulary and symbols Part 2: Applied statistics 3 Terms, definitions and symbols
The terms, definitions and symbols defined in ISO353-1:2006, ISO3534-2:2006 and GB/T5378,1:2008 and the following terms, definitions and symbols are applicable to this part, GB/T 6378.4—2008
3.1 Terms and definitions
Audit population population
The total number of unit products to be audited: 3. 1. 2
Quality levelquality level
The actual average value of its quality characteristics in the audit population. 3,1,3
Upper limit of declared quality level
Upper limit of declared quality levelThe upper limit of the quality level allowed in the audit population. 3.1.4
Lower limit of declared quality levelThe lower limit of the quality level allowed in the audit population
Limiting valuelimiting value
The minimum value allowed in the quality statistics of the audit population3. 1.6
re-test
Re-test the original sample products
Re-inspeclion
Draw samples from the original audit population again for inspection to determine whether the audit population meets the declared quality level. 3. 1.8
Test level
The corresponding relationship between sample size and test power in the sample suppression test. 3. 1.9
Quality statistics: A function consisting of the claimed quality level (lower limit), sample mean and batch standard deviation (or sample standard deviation). Note: There are two types of quality statistics: upper and lower quality statistics. 3.1.10
“g” method
When the standard deviation of the check population is known, the method of using the sample mean and the standard deviation of the check population to determine whether the check population is unqualified. 3. 1. 11
S method: “S” method
When the standard deviation of the check population is known, the method of using the sample mean and the sample standard deviation to determine whether the check population is unqualified. 3. 1. 12
Audit sampling power When the actual quality level of the audit population does not meet the claimed quality level, the probability that the audit population is judged as unqualified 3.2 Symbols
Population mean
: Population standard deviation
Estimated value of population standard deviation
DQT.1. Lower limit of claimed quality level
DQLu Upper limit of claimed quality level
Sampling sample
X, represents: Quality characteristics of sample products Sample mean
sSample standard deviation
Q! Upper quality statistic of claimed quality level Qu
DQL:-x
X or Q:
Q Lower quality statistic of claimed quality level
Range limit
αProbability of first type error (risk of misjudgment)
βProbability of type II error (risk of missing judgment) Ln, second sampling plan
P()Probability of passing the verification sampling plan
4Procedure for implementing sampling inspection
The inspection procedure specified in this standard is as follows:
\) Determine the verification population;
Determine the verification quality characteristics:
DQL or Qt.
Determine the type of verification sampling inspection: method or: method; Determine the verification sampling inspection Method: 1 side, bottom side or double side; d)
e) Specify the quality level of the market;
Specify the inspection and testing level:
g) Search the inspection sampling plan;
h) Take samples;
i) Test and calculate the samples;
i) Determine whether the inspection population is unqualified.
5 Implement sampling inspection
5.1 Determine the inspection population
DQL:- X
X- DQLt
GB/T 6378.4—2008
Determine the inspection population according to the inspection needs. The products in the inspection population can be products produced by the same manufacturer, the same model and the same production cycle. If necessary, it can also be products produced by the same manufacturer, the same model and different production cycles; or products produced by the same manufacturer, different models and different production cycles, or even products produced by different manufacturers, different models and different production cycles. 5.2 Determine the quality characteristics of the total inspection population According to the needs of the inspection, determine one or more quality characteristics of the inspection population. 5.3 Determine the type of inspection sampling inspection
If the standard deviation of the inspection population can be determined in advance, the G method can be selected: otherwise, the S method should be selected. 3
GB/T 6378.4—2008
Table 1 and Table 2 in this standard are used for G method, and Table 3 and Table 4 are used for -5 method. 5.4 Determine the sampling inspection method for inspection
This part of GB/T 6378 includes three different sampling inspection methods: upper limit of marketable quality level, lower limit of marketable quality level, and double limit of marketable quality level. They can be selected according to the different requirements of the product standard for the specification limit of product quality characteristics. 5.5 Specify the claimed quality level
The claimed quality level should be determined according to the inspection requirements. The claimed quality level should be appropriate to the sampling inspection method. 5.6 Specifying the verification test level
Tables 1 to 4 of this part of GB/T 6378 give 15 verification test levels. Taking into account the difference between method A and method S, the sample size series corresponding to the verification test levels in Tables 1 and 2 are not completely the same as the sample size series corresponding to Tables 3 and 4. The higher the verification test level, the larger the sample size and the higher the test efficacy. Once the verification test level is selected, it shall not be changed during implementation.
5.7 Specifying the sampling plan for verification
5.7.1 Method: Given the upper or lower limit of the claimed quality level, use Table 1 and directly read the sample size and limit value from the line where the test water is located. 5.7.2 Method: Given the upper and lower limits of the claimed quality level DQL:-DQL- value column and sample size n are located in Table 2. First read the sample size from the row where the verification test level is located, and then read the limit value from the intersection of the rows.
5.7.3S method: Given the upper or lower limit of the claimed quality level, use Table 3. First read the sample size n and the limit value in the row where the verification test level is located. 5.7.4S method: Given the upper and lower limits of the claimed quality level a/yn
, use Table 4, first read the sample size n from the row where the verification test level is located; then, based on past experience or negotiation, an estimated value 3 of the population standard deviation used to retrieve the sampling plan is obtained, and the limit value k is taken from the intersection of the column where the DQL-DQI value is located and the row where the sample size is located. a/Vn—1. 54
Calculate the estimate of the quality statistic and make a judgment. If there is any objection to the judgment result, the sample standard deviation 5 in the calculated quality statistic should be used instead of the previous one. , repeat the above process,
5.8 SamplingbZxz.net
The samples shall be randomly selected from the verification population according to the method specified in GB/T10111. When stratified random sampling is used, the number of sample products drawn from each layer shall be proportional to the size of the layer of the verification population under consideration. When sampling from the verification population (batch), stratified sampling can be used with the identified sub-batches as layers. When sampling from the process, stratified sampling is used according to the identified sources of variation (for example, tools, operators, shifts, etc.). If the sample required by the retrieved sampling plan exceeds the size of the verification population under study, all unit products in the verification population shall be inspected.
5.9 Testing and calculation of samples
The samples in the sampling area shall be tested one by one according to the test, measurement or other methods specified in the product standards or relevant documents. The test results should be recorded completely and accurately, and the mean and standard deviation of the samples should be calculated. 5.10 Determine whether the overall inspection is unqualified
5.10.1 Determination rules
) Given the upper limit of the claimed quality level
if Q, the sample does not meet the requirements, and the overall inspection is unqualified. If Qu>k, the sample meets the requirements and the inspection is passed. _DQL-X
Where: Q:=
Given the lower limit of the claimed quantity level
if Q≤, the sample does not meet the requirements, and the overall inspection is unqualified; if, the sample meets the requirements and the inspection is passed. X—DQL
Where: QL
c) is the double limit
If Qt≤ or Q≤, the sample does not meet the requirements and the overall inspection is judged to be unqualified. If u and Q, the sample meets the requirements and the inspection is judged to be passed. ._DQLu—Y
X-DQLL
5. 10.2s Judgment Rules
a) Given the upper limit of the claimed quality level
: 3, that is, the sample does not meet the requirements, the overall inspection is judged to be unqualified; if, that is, the sample meets the requirements, the inspection is judged to be passed. DQLx
Where: Q5=
b) Given the lower limit of the claimed quality level
3, that is, the sample does not meet the requirements, the overall inspection is unqualified; Q., that is, the sample meets the requirements, the inspection is passed. Other: Q
c) Given the double limit
Q≤ or Q, that is, the sample does not meet the requirements, the overall inspection is judged to be unqualified; if Q>ⅡQ:>, that is, the sample meets the requirements. The inspection is passed, _DQLs X,Qt X DQIA
It:Q -
5. 11 Statistical interpretation of sampling inspection conclusions
GB/T 6378.4—2008
The quality department responsible can confirm that the inspection population is unqualified if the inspection samples do not meet the requirements, and can reasonably trace back to the inspection population. For products that were not present during the inspection sampling, if there is sufficient evidence to prove that they belong to the inspection population (for example, they belong to the same inspection batch), they should also be treated as unqualified.
Because the population judged as passed by the inspection sampling plan has a high risk of missed judgment. Therefore, the population that has passed the inspection is not confirmed as qualified:
5.12 Re-inspection and re-testing
If there is any objection to the inspection results, re-inspection or re-testing can be carried out. The re-inspection results obtained in accordance with the provisions of GB/116306 shall be used as The final result of the quality characteristics of the sample products. The re-inspection samples do not include the sample products in the initial inspection samples. The re-inspection sampling plan shall be inspected in accordance with the procedures specified in GB/716306. The re-inspection conclusion is the final conclusion. Note: After re-inspection, the value of the sampling inspection (the probability of the first type of error) is reduced, and the β value of the single sample inspection (the probability of the second type of error) is increased. 5.13 Use actual quality level to judge
When the actual quality level of the verification population can be determined, the actual quality level of the verification population shall be compared with the claimed quality level DQL to determine whether the verification population is qualified, without using the sampling plan to judge; at this time, there is no re-inspection, and re-inspection is allowed. 5
GB/T 6378.4: -20D8
6 Application examples
6. 1. 1 Given the upper limit of the claimed quality level Example 1: A company produces solid caustic so that the iron oxide content is required to be low. The upper limit of the claimed quality level of the total average is DQL=0.034%, and the verification inspection level is II: Given the population standard deviation n=0.0005%, try to determine the verification sampling plan. a) Table, from the verification inspection water source line: L,_=_4,-0.822"; b) Determine, randomly select 4 sample products from the population, and after testing, calculate the detailed mean and add the quality statistics, Qu _ DQLu-X_ 0.004 5-X
The judgment rule is: if Qu is 3.%22, that is, the sample does not meet the requirements, then the overall inspection is judged to be unqualified. If 2>-0.22, that is, the sample meets the requirements, then the inspection is judged to be passed.
6. 1. 2 Given the lower limit of the claimed quality level Example 2: The strength of a certain material is considered good, the overall mean quality level QL is: -1×F, and the inspection level is . The overall standard deviation is -4×10°Pa, try to determine the inspection sampling plan. ) Check Table 1: From the inspection level, we can find: n+k = -12,-0. 4407
b) Determination. 11 units of products are randomly selected from the population. After testing, the sample mean X and the lower quality statistic are calculated. Q_XDQLLX-45 ×10
The classification rule is: if Q.≤-0.440, the sample does not meet the requirements, and the overall inspection is unqualified; Q.2-0.240, the sample meets the requirements, and the inspection is judged to be passed. 6. 1. 3 Example 3 of the situation with given upper and lower limits of the claimed quality level: The standard size of a certain product is 100 mm, and the overall mean quality level is limited to DQL100.1. The lower limit of the claimed quality level is QL-99.9 mm, and the test water is. Knowing that the overall standard deviation is-. 3 IM, the sampling plan is determined. a) Check Table 2, and find out from the verification level research column: = 9
b) Calculate
DQL-DQLA-100.1-99.9
2) Check Table 2 again, and find out from the intersection of the calculated value 2,000 and the row where the value = 9 is 2.54. However, the feasible solution is:
Ln+k_— [9, — 0. 519]
d) Determine. Randomly select a unit product from the total volume, and calculate the sample mean x and the quality inspection statistic after testing. The judgment rule is: if Qu is less than 0.549 or less than 0.549, that is, the sample does not meet the requirements, then the overall inspection is judged to be unqualified; if 2u~一0.49 is (2.-0.519), that is, the sample meets the requirements, then the inspection is judged to be passed. 6.2 Method
6.2.1 Case of a given upper limit of the claimed quality level Example 1: A certain material chemical composition SO, the required content is low, and the claimed upper limit of the overall mean TQLu=1. 52%, check the inspection level V based on the data of recent quality management or sample inspection, it is impossible to estimate the overall standard deviation in advance, use the method of unknown standard deviation, and try to confirm the sampling plan. ) Look up Table 3, the row where the inspection level is located is found: LR, K_ = [8, 3. 570]
) Judgment: Randomly select & units of products from the total work, measure the appropriate S0, calculate the sample value X sample standard deviation and the upper quality statistic, DQLt.-X _ D. 0J5 2-X
GB/T 6378.4—2008
Judgment rules: If Q:--0.670, that is, the sample does not meet the requirements, then the overall inspection is unqualified; if Qu>9.670, that is, the sample meets the requirements,The sample is judged to have passed.
6.2.2 Example 5 of a given situation where the claimed quality level is close to the lower limit. A certain steel material is required to have a high drop value. The claimed quality level of the population mean is LDQL.-75, and the verification inspection level is specified as XI. Since it is impossible to fully estimate the population standard deviation, the 5th method of unknown standard deviation is used to determine the sampling plan. a) Look up Table 3, and the sample inspection level is obtained by checking: [z,] —[14,— 0. 473]
b) Judgment. Randomly select 14 units of products from the population, measure the pumping degree, and calculate the sample mean and sample standard deviation: 1) The following quality statistics Q
X DQLL_X-75
The judgment rule is: if Q, 0. 173, that is, the sample does not meet the requirements, then the overall inspection is judged to be unqualified; if Q, 0. 473, that is, the inspection does not meet the requirements, then the inspection is judged to have passed.
6.2.3 Example 6 of the situation where the upper and lower limits of the claimed quality level are given; Assume that the standard deviation of Example 1.1.3 is known. After the error is 0.35, the unknown standard deviation method is used. Try to determine the sampling plan, 2) Enter Table 4, and find the row where the inspection level is: 1=11b) Calculate
D014 _ 100. 1-99. 9
0.35/S.36
B/ -/m. -1. 64
c) Look up Table 2 again, and find the limit value from the intersection of the column where the calculated value 1.748 is located and the row where \—11 is located to obtain the limit value Table 1 -0.G31. The following sampling plan is: En+±7 --11,- 0, 531
d) Judge. 11 items are randomly selected from the population. After testing, the sample value and sample standard deviation are calculated: and the upper and lower quality statistics are the maximum Qu - 1Q1gx _ 120. 1-x, Q. - XQ = x 0.8. The judgment is: if Qu is 0. 531 or (2, ≤ 0.531, that is, the sample does not meet the requirements, then the total inspection is judged to be unqualified; if Qu × 0.53]. And Q is 9.531, that is, the sample meets the requirements, then the inspection should be completed. If there is any objection to the above judgment result, the standard deviation 5 in the column is used to replace the previous one in b). After that, the process is repeated, and the final result is the final result of the second inspection.
7 Inspection power of sampling plan
After the sampling plan is determined, the probability of passing the sampling plan can be calculated according to the calculation formula of the probability of passing the sampling plan given in Appendix A to calculate the passing probability P (). The test power is:
8 Sampling table
Table 1 gives the single-view limit sampling table of the Shan method; Table 2 gives the double-side limit sampling table of the Shan method; Table 3 gives the single-case limit sampling of the Shan method; Table 4 gives the double-side limit sampling table of the Shan method,
Table 1 Sampling table for the upper or lower limit (single-side limit) of the claimed quality level of the Shan method Verification and inspection level
—0.9 yuan 0
GB/T 6378.4—2008
Verification and inspection level
Table 1 (continued)
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.