Procedures and the plan of sampling for repeat test or inspection in product quality audit
Some standard content:
GB/T16306--1996
This standard stipulates [In product quality supervision, 1. Re-test individual samples; 2. Methods for re-testing the overall monitoring group. There is no ready-made international standard for reference in this standard; the content of re-testing individual samples refers to IS5725-6 "Precision of test methods and results—Part 6: Use in precision estimation"; the content of re-testing the overall monitoring group refers to ISO/CD13447 Audit Sampling)
Appendix A of this standard is the standard appendix.
This standard is proposed and managed by the National Technical Committee for the Application of Statistical Methods: This standard is drafted by the National Technical Committee for the Application of Statistical Methods, Sampling Inspection Technical Committee. The drafting units of this standard are: China Institute of Standardization and Information Classification and Coding, Institute of Systems Science, Chinese Academy of Sciences, Institute of Metallurgical Products, Ministry of Metallurgy, Beijing Institute of Industry,
Main drafters of this standard: Yu Zhenfan, Ma Yilin, Chen Zhitian, Liu Wen, Chu Anjing, Yu Shanjun, Liu Qiong. 1 Scope
National Standard of the People's Republic of China
Procedures and the plan of samnpling for repeatlesl or inspection in product quality auditGB/T 163061996
This standard specifies the procedures and sampling plans for product quality supervision and review. When retesting samples, it specifies the procedures and methods for retesting. When re-inspecting the supervision population, the product quality supervision re-inspection procedures and sampling plans are specified with the quality level of the supervision population (expressed as the number of unqualified products per unit or the mean of a quality characteristic of the supervision population) as the quality indicator. The corresponding misjudgment risk ?=0.05, the omission risk β=0.10.
When re-inspecting samples, this standard is applicable to the case where the error of the test results of the inspected samples obeys the normal distribution (the standard deviation is known). When re-inspecting the replacement population, this standard is only applicable to the case of discrete individuals and is not suitable for bulk materials. When the quality level of the supervision population is expressed by the number of unqualified products per hundred units of products, the supervision population size N should be greater than 250 and the ratio of the population domain to the sample base should be is greater than 10, that is, N/n>10. When the monitored population does not exceed 250, but the ratio of the population to the sample base is greater than 10, the sampling plan retrieved from this standard is approximate and should be used with caution. The sampling plan can also be determined according to the method specified in GB/T13264. 2 Referenced standards
The provisions contained in the following standards are reduced to the provisions of the standards through reference in this standard. When this standard was published, the versions shown were valid. All standards will be revised, and parties using this standard should explore the possibility of using the latest versions of the following standards, GB/T223.3-1981 Determination of phosphorus in steel and alloys GB/T3358—1993 Statistical terminology
GB/I 65831904 Terminology of quality management and quality assurance GB/T8054-1995 Average value measurement standard type one-drink sampling inspection procedure and sampling GB/T10111--1988 Method of random sampling using random numbers (i13132641991 Small batch counting sampling inspection procedure and sampling table for nonconforming product rate 3 Definitions and symbols
In order to formulate this standard, in addition to the definitions in GB/T3358-1993, the following definitions are also applied. 3.1 Definitions
3-1-1 Repeat test
Test the repeatability or reproducibility of samples. 3.1.2 Repeat. Inspertion Samples are drawn from the original monitoring population for the second time for inspection to determine whether the monitoring population is sufficient to pass. 3.1.3 Repeat. Test Inspection re-inspection and re-testing are collectively referred to as re-testing.
3.1.4 Reproducibility condition Reproducibility condition is the test condition that is carried out independently on the same object in a short time by the same operator using the same equipment in the same laboratory and following the same test procedure.
GB/T 16306-1996
3-1.5 Reproducibility conditions is the test condition that is carried out independently on the same object when there are essential changes in the test laboratory, operator, test equipment, test procedure (method) and test time. 3-1.6 Repeatability limit A value r, under which the absolute value of the difference between two test results does not exceed this number with a probability of 95%. 3.1-7 Reprodurability limit A value R, under reproducibility conditions, the absolute value of the difference between two test results does not exceed this number with a probability of 95%. 3-1.8 Repeatability critical difference A value, under reproducibility conditions, the absolute value of the difference between two test results or the final result (such as mean, median, etc.) calculated from two test results does not exceed this number with a certain probability. 3.1.9 Reprodueibility critical difference A value, under reproducibility conditions, the absolute value of the difference between two test results or the final result (such as mean, median, etc.) calculated from two test results does not exceed this number with a probability of 55%. 3.110 Median rmediat
If the values are arranged in ascending order of their algebraic values and numbered 1 to. When n is an odd number, the median of the values is the (n11)/2th value. When n is an even number, the median is between the n/2th value and the n/2+1th value, taking the arithmetic mean of these two values.
3. 1.11 supervised population The set of supervised products.
3.1.12 Quality of audit population The quality of the audit population expressed by the number of unqualified products per 100 units or the mean of a quality characteristic of the audit population is called the quality of the audit population.
3.1.13 Quality level of audit population The limit value of the quality index allowed in the audit population (expressed by the number of unqualified products per 100 units or the mean of a quality characteristic of the audit population).
3.1.14 Power of audit sampling The probability that the audit population is judged as unqualified when the actual quality level of the audit population does not meet the requirements. 3.1.15 Risk of missed judgment The probability of judging an audit population that is actually unqualified as acceptable. 3.1.16 Risk of wrong judgment probability The probability of judging the total number of actually qualified products as unqualified. 3.1.17 Nonconformity
Non-conformity of a product or process to meet the specification. 3.1.18 Type A nonconformity The most important quality characteristics of a unit product do not meet the requirements, or the quality characteristics of a unit product seriously do not meet the requirements, which is called Type A nonconformity.
3.1.19 Type B nonconformity The most important quality characteristics of a unit product do not meet the requirements, or the quality characteristics of a unit product seriously do not meet the requirements, which is called Type C nonconformity.
3.1.20 Type C nonconformity The most important quality characteristics of a unit product do not meet the requirements, or the quality characteristics of a unit product slightly do not meet the requirements, which is called Type C nonconformity.
3.1.21 Nonconforming products
GB/T 16306—1996
Unit products, which have one or more nonconformities. According to the nonconforming categories, they can generally be divided into A-type nonconforming products, B-type nonconforming products and C-type nonconforming products.
3.1.22 Type A nonconforming products type A nonconforming products have one or more A-type nonconforming products, and may also have B-type and/or C-type nonconforming unit products, which are called type B nonconforming products.
3.1.23 Type B nonconforming products type B nonconforming products have one or more B-type nonconforming products, and may also have C-type nonconforming products, but do not include A-type nonconforming unit products, which are called type B nonconforming products.
3- 1. 24 Type C nonconforming products type C nonconforming products have one or more C-type nonconforming products, but do not include A-type and B-type nonconforming unit products, which are called type C nonconforming products. 3.1.25Number of nonconforming products per 100 unitsThe total number of all nonconforming products in the supervision population divided by the total number of supervision, and then multiplied by 100, is called the number of nonconforming products per 100 units. That is:Number of nonconforming products per 100 units = total number of supervision products × 100Number of nonconformitics per 100 units3.1-26Number of nonconforming products per 100 unitsThe total number of all nonconforming products in the supervision population divided by the total number of supervision, and then multiplied by 100, is called the number of nonconforming products per 100 units. That is:
Number of nonconforming products per 100 unitsThe total number of all nonconforming products in the supervision population divided by the total number of supervision, and then multiplied by 100, is called the number of nonconforming products per 100 units. That is:
Number of nonconforming products per 100 unitsThe total number of nonconforming products in the supervision population × 100Number of supervision population,
3.1.27Rejection valueWhen the supervision population is judged as unpassable, the limit value of the sample indicator. 3.1.28 Unqualified audit population When the actual quality level of a population is not up to standard, the population is called an unqualified audit population. 3.1-29 Rejection population In counting inspection, if the number of unqualified products found in the sample is not less than the rejection judgment number r, i.e. dr, then the population is considered to be unqualified in the spot check.
In metrological inspection, if the quality statistic is less than or equal to the rejection judgment value, then the population is considered to be unqualified in the spot check. 3.1.30 Sampling plan of repeat inspection The combination of sample size and rejection judgment number (value) is called a sampling plan for repeat inspection. 3.2 Symbols
Standard deviation
Annual standard deviation
C,Rg(m)
Xmax+Xmia
Reproducibility standard deviation
Repeatability limit
Reproducibility limit
Initial test number
Reproducibility critical range of sample size
Repeatability critical range coefficient of sample size Reproducibility critical difference
Extreme value of random variable test result
Test result
Maximum test result
GB/T 16306—1996
The total number of unit products included in the supervision population, that is, the supervision population size sample size
Number of non-qualified (products) in the sample
Supervision quality level
Actual quality level of the supervision population
Risk of missed judgment quality
Counting re-inspection sampling plan
Number of non-pass judgments
Risk of wrong judgment
Risk of missed judgment
4 Re-inspection procedures and implementation of samples
For non-destructive testing, when the results of the first test are considered abnormal, the technical or physical reasons for the abnormality must be found first: if the reason cannot be found, a second test under repeatability or reproducibility conditions may be carried out with the consent of the inspection party. For destructive testing, re-testing of backup samples is allowed only when there is reliable evidence that the first test is wrong, otherwise, it should be handled as a re-inspection situation.
4.1 Method for checking the acceptability of test results obtained under repeatability conditions and determination of final test results When there is any doubt about the accuracy of the first test result and the requirements of Chapter 4 are met, the first reading or more test results shall be obtained by the original laboratory.
4.1.1 Determination of final test results
4.1.1.1 Specify the single and multiple limits r
4.1.1.2 Determination of final test results When the absolute value of the difference between two test results is not greater than, that is, when X, X≤r, both results are acceptable and the final test result is equal to the arithmetic mean of the two results. If the absolute value of the difference between the two results is greater than a value, another test result shall be taken. If the range of the three results is equal to or less than the critical range R (3), the final test result μ is equal to the average of the three results: If the range of the three results is greater than the critical difference C, R (3), the median is taken as the final test result. This process can be represented by Figure 1. 2 initial test results
-half mean of 2 results
take another test result
-Xui CK (3)
2--median of 3 resultsWww.bzxZ.net
-average of 3 results
GB/T 16306—1996
The general expression of critical difference C,R(3) is: C.Rg(m)=r(m)a,=f(m)z/2.77The value of f(m) in the above formula is shown in Table 1.
Critical difference coefficient (Meter)
4.1.2 Reporting final test results
When reporting the final test results, the following should be stated: a) Number of tests:
b) Whether to take the average or the median.
42 Method for checking the acceptability of test results obtained under repeatability and reproducibility conditions and determining the final test results This method is used when two laboratories participate in the experiment and their test results or the average values of the results are different. In this case, the given reproducibility standard deviation should be used for statistical testing as for repeatability. In all cases, sufficient test samples should be ensured to obtain the test results, including the preservation of a part of spare samples for use when it is necessary to retest. The number of spare samples depends on the test method and the complexity of the experiment. The spare samples should be properly preserved to prevent damage and deterioration.
The samples should be consistent, that is, the samples in the final preparation stage are used by all laboratories. 4-2-1 Test results of two laboratories: statistical test of consistency 4.2.1.1 Test of each laboratory obtaining one test result When each laboratory obtains only one test result, the absolute value of the difference between the two results is tested using the reproducibility limit. If the absolute value of the difference is less than or equal to R, the two results are consistent and their average value is taken as the final test result. If the absolute value of the difference between the two results is greater than, it is necessary to find out whether the cause of the difference is due to a fault in the test equipment, low precision of the test method and/or differences in the test samples. Each laboratory shall test the precision under single-complex conditions according to the provisions of Article 4.1. 4.2.1.2 Test of more than one test result obtained by each laboratory 4.2.1.2.1 Assuming that each laboratory has obtained the final test result according to the steps specified in Section 1., it is sufficient to consider the acceptability of the two final results. The absolute value of the difference between the two results is compared with the critical difference C to test whether the results of the two laboratories are consistent. The test method is as follows:
a) Both results are average values (the number of repetitions is mm) The critical difference C1 expression is: CD
2m2!
b) When one of the two results is the mean and the other is the median (the number of repetitions is m1 and m2 respectively), the critical difference (CD) is expressed as: C,D
Ic(m,):2
GB/T16306—1996
(Class) is the ratio of the median standard deviation to the mean standard deviation, and its value is shown in Table 2.
c) When the two results are the median (the number of repetitions is m1 and m2 respectively), the critical difference CD is expressed as: C.D As
The C(m:)(i-1,2) value in the formula is shown in Table 2. Test results Number of doses
1. 092 15
1. 197 57
(C(m1))2(C(mg))2-
Table 2 C(m) value
Test results Number of times m
1. 195 97
1. 237 25
1. 207 69
-+-(3)
4.2.1.2.2 If the absolute value of the test result is less than the critical difference, the most complete test results of the two laboratories are acceptable, and the weighted average of the two results (m,= (m+m)/(m+m) as the final test result. If the absolute value of the two results is greater than the critical difference, the steps specified in 4.2.2 shall be adopted.
4.2.2 Solutions to the inconsistency of the test results of the two laboratories The reason for the inconsistency of the results of the two laboratories may be systematic error or sample inconsistency. The test should be carried out with another sample to determine whether the systematic error exists and the degree of deviation. Calibrated reference materials should be used under possible conditions. If this is not possible, a universal test partner (preferably a known value) should be tested. The advantage is that the systematic error of a laboratory or two laboratories can be found. If the systematic error cannot be found by this method, the two laboratories should reach an agreement with reference to the results of a third laboratory.
When the difference comes from sample inconsistency, the laboratories should jointly prepare the samples or entrust the first party to prepare the samples. 4.3 Application Examples
Example 1: Determine the phosphorus content in steel by the antimony phosphomolybdenum blue photometric method in GB223.3. When the theoretical phosphorus content is 0.0174, r=0.0018, and the two data of the same sample have been measured as 0.0170 and 0.0179. Find the final test result. Solution: X,-X|=|0.0170- 0.0179/—0.0009 Because LX, a X, r, the final test result is: (0.0170+0.0179)/2=0.01745, rounded to 0.0174 Example 2: In Example 11, if the two data of the same product are 0.01795 and 0.0161, find the final test result. Solution range X, -X, 1-[0.01795—0.0161=0.00185 Because "X, —X1r, it is necessary to test again. The third data is 0.0164, so Xmx=0.01795, X.-0.0161, /Xrax—Xm1=0.01795—0.0161=0.00185, critical range C, R (3) f (3) = 3. 31 X
0.0018=0.0021.
Since Xmx is Xmm/(R(3), the mean of the three data in the final test result is: GB/T 163061996
-(0.01795+0.01610.0164)/3-0.0168 Example 3 In Example 2, if the first data measured is 0.0158, then ×=0.01795, Xm=0.0158, 1XmXm =0. 017 95—0, 015 8-0. 021 5 Critical range CR(3)—0.0021 Since Xx—X>(R(3), the median of the three data in the final test result is,
μ0.016 1
Example 1: Determine the phosphorus content in steel by the antimony phosphorus molybdenum blue photometry in GI3223.3. When the theoretical phosphorus content is 0.055, r0.0020, R=0.0035. Two laboratories have obtained the following results for the same sample: Laboratory 1 obtained results of 0.0580, 0.05590.0558, and Laboratory 2 obtained results of 0.0565, 0.0538, and 0.0532, respectively. Compare the differences in the laboratory results and find the final test results.
Solution: Laboratory 1:
Since 1X,X,1=[0.0580-0.0559=0.0021>, one more test is required, and the result is 0.0558, so there is: Range 1XmxXmim=0.0580-0.0558=0.00220.002 0
Critical difference CR (3)—f3)×
The range is less than the critical difference, so the average value is: 0.00289
#1 -(0. 058 0-1 0. 055 9-0. 055 8)/3= 0. 056 6Laboratory 2:
h at X,X,-0.0565—0.05381-0.0027>mSo it needs to be tested again, and the data is 0.0532, so there is a critical range /Xm-Xminl-0.0565—0.0532=0.0033Critical C.,Rs(3)-[(3)×0.0020
.2. 77
The extreme value is less than the critical difference, so the median is taken as: 10.00239
2— 0. 053 8
The difference between the final results of the two laboratories is 1—a=0.0566-0.0538-0.0028The critical difference of reproducibility is;
C,Dys =N
0. 003 52 --- 0. 002
Due to a CD, the final test result is [C(m
(1. 160 18)\-
f - (3μ + 3u2)/6
=(0.0566+0.0538)/2
5 Re-inspection procedure for the supervision population
The re-inspection procedure specified in this standard is as follows:
a) Confirm the supervision population;
b) Determine the quality characteristics of the unit product;
c) Determine the classification of non-conforming (products);
d) Specify the re-inspection quality level;
e) Specify the risk of missed judgment quality;
i) Retrieve the re-inspection sampling plan;
g) Draw samples;
h) Inspect the sample;
i) Determine whether the supervision population has passed the training;
6 Implementation of re-inspection sampling
6.1 Determine the supervision population
GB/T 16306 :-1996
The overall monitoring population for re-inspection shall be the same as that for initial inspection. 6.2 Determine the quality characteristics of individual products
Make clear provisions for the technical performance, technical indicators, safety, hygiene indicators and other quality characteristics of the unit product. These indicators must be the same as those specified in the initial inspection.
6.3 Determine the categories that do not contain qualified (products) for re-inspection and the categories that have different unqualified (products) in the first supervision sampling inspection. 6.4 Determine the overall quality level of supervision
The overall quality level of supervision during re-inspection sampling shall be the same as that of the overall quality level of supervision during the initial supervision sampling inspection. 6.5 Provisions for full judgment of risk quality
6.5.1 Counting situations
6.5.1.1 Provisions for missed judgment of risk quality
6.5.1.1.1 The number of unqualified products per hundred units of products is the quality index. When 0.1 inch=20 When 0.1 ≤ 1.0, 5: When 1.0=4.0, ≤4 When 4.0 ≤ 10.(, force, 3. The value of force: shall refer to the above provisions and be determined by the supervisory party and the supervised party through consultation. 6.5.1-1.2 The number of unqualified products per hundred units of products is the quality index. When Pe, 0.1, P, 4.20 p When 0.1-1.0, P ≤ 5pe When 1.0P, ≤, 4.0, P4P When 4. 0P,10. When ·,3p: When 10=,100·2. The value of should refer to the above provisions and be determined by the supervisor and the supervised party through consultation.
6.5.1-2 Sample search for re-inspection batches
When the number of unqualified products per 100 units of products is the quantity indicator, search for the sampling plan in Table 3. When the number of unqualified products per 100 units of products is the quality indicator, search for the sampling plan in Table 4 Retrieve the sampling plan. Select the sampling plan from the intersection column of +P in Table 3 or A. The value on the left side of the column is the sample size, and the value on the right side is the number of samples that do not pass the standard. Note: According to the above retrieval method, if the sample size exceeds the total supervision size, a full inspection should be carried out. 6.5-2 Situations where the sample size is missing. 6.5.2.1 Specify the missed judgment risk quality level. The requirements for judgment accuracy and the acceptable sample size. The supervisory party and the supervised party negotiate to determine the missed judgment risk quality level. 6.5.2.2 Retrieve the re-sampling plan. According to the determined (-) and (values), find the re-inspection sampling plan from GB/T8054. If the sample size of the re-inspection sampling plan is smaller than the sample size of the previously used sampling plan, the sampling plan with the larger sample size will still be used during the re-inspection. 6. 6 Application Examples
Example 1: When the specified pressure is 1.03% and the product is 3.00, find the re-inspection sampling plan with the number of unqualified products per unit as the quality indicator.
In the intersection of the row where the pressure is 1.05 and the column where the pressure is 3.0%, we find 435.9. That is, the sample size is 435 and the number of unqualified products is 9.
Example 2: When the specified pressure is 1.5% and the product is 5.5%, find the re-inspection sampling plan with the number of qualified products per unit as the quality indicator: In Table 4, when the pressure is 7.5%, we find 175.6 in the intersection of the row where the pressure is 7.5% and the column where the pressure is 5.5%. That is, the sample size is 175 and the number of unqualified products is 6.
250,11
1 165,13
SOE-14
144,34
#事,5
19s,ir
1r0-13
GB/T163061996
Sampling plan for inspection based on the number of unqualified products per unit as the quality indicator Table 2.5
150,11
120,1331.6
GB/T16306-1996
Appendix A
(Standard Appendix)
Comprehensive center curve of quality supervision sampling inspection Assume that the original supervision sampling plan is (nr or (n,, and the re-inspection sampling plan is (n) or (n,). The (OC function of the original blue hip sampling plan is L () [or L, (): OC of the re-inspection plan The number is L,) or L (μ)]. The re-inspection sampling plan combined with the original plan is actually a special two-sample sampling plan. We call it the "comprehensive two-sample sampling plan" and use the symbol (nm.) * (n2r) [or (n, k) * (13:) to represent it. The C function of this special two-sample plan (it is not convenient to call it the comprehensive C function) is L(p) =, (p) + [1 1, (p)L, (p)
or (r) = I(μ) 4 T1 - L, (μ)I, tμ) If the supervision sampling plan is suitable when =:,, (p) = 1 ai, when, l() = β and the re-inspection sampling plan is suitable when =, ( = - when =,) = , then the comprehensive acceptance probability is: L() = 1 - + (1 -)
= 1— aaz
1 — This shows that after the re-determination of the re-inspected sample, the risk of misjudgment is actually reduced from the original supervision sampling plan to 1/1, while the risk of missed judgment is increased from the original supervision sampling plan to 1/1. For example: let 0.65, and the supervision sampling plan is (n-8.r-1); when re-inspecting the sample, let 0.65.P-6.5: its sampling plan is (50,r2)
L(p) =1-0.051 = 0.949
, that is, α1 = 0. 051
8, = 0.584
L.(p. - 1-0. 042.. 0. 95822--0.155
, that is, α±-0. 042
L1(2) = 0. 58
L(p, = 1 - 0. 051 × 0. 043 = 1 - 0. 002 1 = 0. 997 9L(p:) - 0. 584 + (1 — 0. 584) × 0. 155 = 0. 648 5 It can be seen that the risk of misjudgment of the comprehensive subsampling scheme is reduced from 0.051 to 0.0021, while the risk of missing judgment is increased from 0.584 to 0.6485
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.