ICS 03. 120. 30
National Standard of the People's Republic of China
GB/T13393-2008
Replaces GB/T13393—1992
Guide to acceptance sampling inspection
Guide to acceptance sampling inspection2008-08-06Issued
General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of ChinaAdministrative Committee of Standardization of the People's Republic of China
Implemented on 2009-01-01
2Normative references
3Terms, definitions and symbols…·
3. 1 Terms and definitions
3.2 Symbols
4 Market conditions and production conditions affecting the selection 4.1 Market conditions
4.2 Production conditions
5 Negotiation process for selecting sampling plans
6 Other factors affecting the selection of sampling plans
6.1 Test of white fraction of defective products and test of number of defective products per 100 units 6.2 Risks of producers and users
6.3 Selection of AQI, PRQ, LQ and CRQ values 6.4 Determination of the inspection level
7 Comparison of sampling inspection methods
7. 1 Sampling inspection by attributes and sampling inspection by record 7.2 Sampling plans for single, double, multiple and sequential sampling 7.3 The role of OC curve
8 The benefits of adopting national standards for acceptance sampling
References
ht
GB/T13393—2008
This standard replaces GB/T13393:1992 "Guidelines for Sampling Inspection". The main differences between this standard and GB/T13393—1992 are: a) The standard text has been redrafted in accordance with the requirements of GB/I1.12000. In order to facilitate the understanding of the standard, relevant terms, definitions and symbols have been added. b)
GB/T 13393—2008
In order to adapt to the development of the sampling inspection standard series and enhance the operability of the standard, the technical framework of the standard has been redesigned. Chapters 5, 6, 7 and 8 have been added.
d) Deleted Appendix A of the original standard and integrated its content into the main text of the standard. This standard was proposed by the China National Institute of Standardization. This standard is under the jurisdiction of the National Technical Committee for Standardization of Statistical Methods Application. The drafting units of this standard are: China People's Liberation Army Ordnance Engineering College, China National Institute of Standardization, Institute of Mathematics and Systems Science of the Chinese Academy of Sciences, and Fuzhou Chunlun Tea Co., Ltd. The main drafters of this standard are: Zhang Yuzhu, Zi Zhenfan, Chen Nen, Ding Wenxing, Feng Tuyong, and Fu Tianlong. This standard was first issued in 1992 and this is the first revision. H
k平dhttndonFonmaton
1Scope
Guidelines for Acceptance Sampling Inspection
GB/T13393—2008
This standard specifies the guidelines for selecting sampling systems, sampling plans and sampling schemes in acceptance sampling inspection. Some guiding tables (Table 1, Table 2a, Table 2b and Table 3) are given to facilitate users to choose a suitable sampling system, sampling plan and sampling scheme. This standard applies to the acceptance sampling inspection of batches of products (such as materials, parts, components, complete machines and systems), and can also be used for the acceptance sampling inspection of continuously submitted work-in-progress.
2 Normative references
The clauses in the following documents become the clauses of this standard through reference in this standard. For all referenced documents with an entry date, all subsequent amendments (excluding errata) or revisions are not applicable to this standard. However, the parties who reach an agreement based on this standard are encouraged to study whether the latest versions of these documents can be used. For all referenced documents without an entry date, the latest versions are applicable to this standard. GB/T2828.1 Sampling procedures for inspection by attributes Part 1: Sampling plan for batch inspection based on acceptance quality limit AQL (GB/T 2828.1--2003, ISO 2859-1:1999, IDT) (I3/T2828.2 Sampling procedures for inspection by attributes Part 2: Sampling plan for isolated batch inspection based on limit quality (LQ) (GB/T 2828.2-2008, ISO 2859-2:1985,NEQ) GB/T2828.3 Sampling inspection procedures by attributes Part 3: Skip lot sampling procedures (GB/T2828.3--2008, ISO28593:2005,IDT
GI3/T6378.1 Sampling inspection procedures by measurement Part 1: Sampling plan for batch inspection of single quality characteristics and single AQL retrieved by acceptance quality limit (AQL) (GB/T6378.1-2008.ISO3951-1;2005,IDT) GB/T8051 Sequential sampling inspection plan by attributes (GB/T8051-2008,I ISO8422.2006, IDT) GB/T8052 Single level and multi-level - counting continuous axis sample inspection procedures and tables GB/T8054 counting standard type one-time sampling inspection procedures and tables GB/T13262 counting standard type one-time sampling inspection procedures and sampling tables for the percentage of non-conforming products GB/T13264 small batch counting sampling inspection procedures and sampling tables for the percentage of non-conforming products ISO3534-1:2006 Statistical vocabulary and symbols Part 1: General statistical terms and terms used in probability IS03534-2:2006 Statistical vocabulary and symbols Part 2: Applied statistics 3 Terms, definitions and symbols
3. 1 Terms and definitions
The following terms and definitions apply to this standard. 3.1.1
Acceptance sampling system Acceptance sampling system A collection of acceptance sampling plans and acceptance sampling plans and the rules used to select appropriate sampling plans or sampling plans. [4.3.1 in ISO 3534-2:2006] 3.1.2
Acceptance sampling plan An acceptance sampling plan and the rules for transitioning from one sampling plan to another. [4.3.2 in ISO 3531-2:2006] 1
GB/T 13393--2008
Acceptance sampling plan An acceptance sampling plan consisting of the sample size used and the corresponding acceptance rules. ISO) [4, 3.3 in ISO 3534-2:2006]
operating characteristic curveOperating characteristic curve
OC curveOC curve
For a given "acceptance sampling plan", the curve that shows the relationship between the probability of acceptance of a product and its quality level. [4.5.1 in ISO 3534-2:2006]3.1, 5
inspection levelinspection level
A pre-set index related to the inspection quantity of the acceptance sampling plan that reflects the relationship between sample size and batch size. NOTE 1: If experience shows that operating characteristic curves with lower and/or higher discrimination are appropriate, corresponding lower and/or higher inspection levels may be selected. NOTE 2: This term should not be confused with sampling strictness. Sampling strictness is related to automatically operating transfer rules. [4.3.5 in 1SO3531-2:2006]
discrimination ratio
(acceptance sampling) Ratio of the user's risk quality to the producer's risk quality. [4.6.12 in 1SO3534-2:2006]
acceptance quality limit(acceptance sampling) The worst quality level that can be tolerated. Note 1: This concept is only applicable to the following acceptance sampling plans with transfer rules and suspension rules such as GB/T2828.1 and GB/T6378. 2: Although some individual batches with the same quality level as the acceptance quality limit can be accepted with a relatively high probability, specifying the acceptance quality limit does not imply that it is the required quality level.
Note 3: The acceptance sampling inspection plan with transfer rules and suspension rules that appears in some standards (such as GB/T2828.1) is designed to encourage suppliers to have a process average that is consistently better than the acceptance quality limit. If the supplier cannot do this, the probability of transferring from normal inspection to tightened inspection is high. Once tightened inspection is implemented, unless actions are taken to improve the process, there is a high probability that sampling inspection will be suspended again according to the transfer rules. Go to 4: It is no longer recommended to use the abbreviation AQL to represent "acceptable quality level". [ISO 3534-2:2006, 4.6.15]
Average outgoing quality averageongoingquality (acceptance sampling) For a given product quality, the expected average quality level of the inspected products. Let 1: Unless otherwise specified, the average outgoing quality is calculated by replacing all defective products in all unaccepted batches with conforming products through 100% inspection of all received batches.
Note 2: These terms and definitions are only used in the following cases: in the case of 100% inspection of unaccepted batches, the removed defective products can be replaced with conforming products.
Note 3: A common approximation for the average outgoing quality is the given process quality multiplied by the probability of acceptance\. This formula is accurate for the scenario with an acceptance number of 0, but the estimate is too high for other cases.
[4.7.1 in ISO 3534-2:2006]3.1.9
average outgoing quality limit (acceptance sampling) for a given acceptance sampling plan, which stipulates that all nonconforming products in the batch shall not be accepted unless otherwise specified in the specification)2
The maximum value of the average outgoing quality for all possible quality levels of the product under given circumstances. 4.7.27 in ISO 3534-2:2006 3.1.10
Limiting quality
GB/T 13393—2008
(Acceptance sampling> When accepting sampling of isolated batches, it should be limited to quality levels with low acceptance probability. 4.6.13 in ISO3534-2.2006]
Producer's risk producer's risk
(Acceptance sampling) The probability of not being accepted by the acceptance sampling plan when the quality level is acceptable. Note 1: The quality level can be related to the nonconforming (defective product) rate and is acceptable compared to AQ1. Note 2: For the purpose of interpretation, the quality level involved is known. [4.6.1 in ISO 3534-2:2006] 3.1. 12
Producer's risk quality <Acceptance sampling> For acceptance sampling plans, the quality level of the batch or process that corresponds to the specified producer risk. Note 1: The type of operating characteristic curve must be specified. Note 2: The producer's risk is generally specified as 5% [4, 6, 10 in IS0 3534-2:2006] 3.1.13
Consumer's risk
(Acceptance sampling when the quality level is an unsatisfactory value, but the probability of being accepted by the acceptance sampling plan. Note: The quality level can be related to the nonconforming (defective product) rate and is unsatisfactory compared with the LQL. [4. 6. 2 in IS0 3534-2,2006] 3. 1. 14
Consumer's risk quality quality(Acceptance sampling>For an acceptance sampling plan, the quality level of a batch or process that corresponds to a specified user risk. NOTE: The user risk is generally 10%.
[1.6.9 in ISO 3531-2:2006]3.2 Symbols
AQLAcceptance quality limit
AOQLAverage outgoing quality upper limit
PRProducer risk
PRQProducer risk quality
CR User Risk
CRQUser Risk Quality
LQLimiting Quality
LInspection Level
OCOperating Characteristic Curve
4 Market Conditions and Production Conditions Affecting Selection 4.1 Market Conditions
In order to select an appropriate sampling system, sampling plan and sampling scheme, at least the following market conditions should be considered: (a) Production is considered to be continuous and the purchaser can influence the improvement of the supplier's product quality; (b) The supplier remains responsible after the product is accepted; (c) Acceptance of a small amount of non-conforming products may cause significant loss of life or serious harm: due to the single source of products, failure to accept a batch may cause the purchaser's factory to close or significant economic losses; (d) The inspection technology of each unit product is complex and the cost is very high; (e)
(f) Inspection is destructive.
The opposite situation is as follows:
The purchaser cannot influence the supplier's product quality improvement; (ar)
Once the product is accepted, the supplier is no longer responsible; (et)
The economic loss or harm caused by the defective product is small, or the defective product is easy to identify and remove during use or assembly;
Equivalent products are available from many sources and are easy to obtain; The inspection cost per unit product is not high;
The inspection is not destructive.
4.2 Production conditions
In addition to market conditions, the following production conditions also affect the selection of acceptance sampling system, sampling plan and sampling scheme: The supplier's product quality is always good;
It is possible to use random sampling to select samples; (h)
i) The inspection speed is fast and the cost is not high;
() It is necessary to know the shape, location and dispersion of the product characteristic distribution; (k) The product characteristics follow a normal distribution (or can be converted to a normal distribution). The opposite situation is shown in Table 2h.
Tables 1, 2a and 26 summarize the sampling standards and plans that can be adopted for various inspection occasions, market conditions that affect the selection, and production conditions. Table 3 includes some notes on the use of Tables 2a and 2b. Table 1 Guidance for selecting an acceptance sampling system, plan or program Inspection occasions
Precious goods
Color-risk goods
End-of-segment inspection
No history of the products inspected
Quality information
Average upper limit of detected quality eggs
Quality parameters
Small batches
Good quality basis of the products received
Continuously submitted work in progress
Conditions for selecting the program
(e)+tfr)
(a),(f)
(ar),e),(et),(f),(cr)
(a),(e),(f)
(e),(tr)
(br),(et)||t t||(fr),(g).(i)
Available sampling plans
100% inspection
(Count)AQL-plan
(Measurement)AQL plan
(Count)Standard plan
(Measurement)Standard plan
(Count)AQL plan
(Count)Skip plan
(Measurement)AQL plan
(Count)Standard plan
(Measurement)Standard plan
AOQL plan
(Count)AQL plan
(Count)Standard plan
(Count)Skip plan
iContinuous sampling plan
http:
Related sampling standards or plans
GB/T 2828. 1
GB/T 6378. 1
GB/T 8051,GB/T 13264
GB/T 8054
GB/T 2828. 1
GB/T 2828. 3
GB/T 6378. 1
GB/T 8051, GB/T13262
GB/T 8054
Select type
GB/T 2828, 1
GB/T13264
GB/T2828.3
GB/T 8052
GB/T 13393—2008
Table 2a Guidance for selecting an acceptance sampling system, plan or scheme based on market conditions and production conditions Market conditions or production conditions
(a) The purchaser can use feedback to encourage the supplier to improve product quality (b) After the product is accepted, the supplier still complains (c) Accepting a few batches of defective products can cause significant economic losses or serious harm (d) Not accepting a batch may cause economic losses and factory closure () The inspection cost per unit product is very high
(f) The inspection is destructive
(g) The product quality is consistently good
(h) It is easy to draw a random sample
(i) The inspection speed is fast
() It is necessary to know the shape, location and dispersion of the distribution of product characteristics (k) The distribution of product characteristics is normal
Applicable sampling standards or schemes
GB/T 2828.1, GB/T 2828.3, GB/T 6378.1,GB/T 8051GD/T 2828.1,GB/T 2828. 3,GB/T 6378. 1,GB/T 8051GB/T8051,GB/T13264.GB/T13262AOQI.Scheme
GB/T 6378.1
GB/T 6378. 1
CB/T 2828.3
GE/T2828.1: second or fifth scheme, GB/T8051GH/T2828.1's second or five desires scheme, GB/T8051GB/T 6378. 1
GR/T 6378. 1, GB/T 8054
Table 2h Guidance on the selection of acceptance sampling systems, plans or schemes based on market conditions and production conditions Market conditions or production conditions
(ai) The purchaser cannot influence the supplier to improve product quality through feedback Most: Inspection of single batches or independent batches
(bl) The supplier is no longer responsible once the product is accepted (ct) Acceptance of a small number of nonconforming products will not cause or cause major losses《dj) There are many sources of equivalent products
(ei) The inspection cost per unit product is not large (F,) The inspection is non-destructive
《》 There is no quality history information for the supplier's products; or the quality history is poor (h.) Random sampling is difficult or expensive (i1) Inspection is time-consuming or expensive
(i) The shape, position and dispersion of the product characteristic distribution are not important (k,) The product characteristic distribution is unknown or unpredictable Sampling standards or schemes that can be used
GB/T 8051, GB/T 13264, GB/T 8054, GB/T 13262GB/T 8051, GB/T 13264,GB/T 8054, GB/T 13262GR/T 2828. 1,GB/T 6378.1.GB/T 8051,GR/T 2828. 3
Considering other factors
Considering other factors
Considering other factors
GB/T 2828. 1,GB/T 8051,GB/T 6378. 1Strictened inspection of GB/T 2828.1, GL/T 6378. 1Single sampling of GB/T 2828.1, GB/T 6378. 1Single sampling of GB/T 2828.1, GB/T 6378. 1Attribute sampling, account-based sampling (considering other factors)Attribute sampling
Table 3 Table 2a and the guidance notes to Table 2b (a), (b)
(ar), (b)
wwW.bzxz.Net(e), (f)
All sampling standards can be used (see Chapter 2). The selection of the appropriate plan also depends on other factors. Therefore, for processes that require longer periods of time and where the quality of individual batches is more important, a standard sampling plan with attributes (or numerators) can be used. The sampling plan should have appropriate discrimination (low discrimination ratio) (see 7.3). In some cases (such as automated inspection), 100% inspection can be used. The CRQ value of the sampling plan should be relatively low and the discrimination should be appropriately high (see 6.2, 6.3, 6.4, 7.3). A selection sampling plan is used with AOQL as the quality parameter. This type of plan requires more management work, especially when the sampling inspection is judged as unacceptable. Counting once or counting sequential sampling is the main choice. If a number sampling is used, it can be used. Five-time or sequential sampling (see 7.1, 7.2) If the supplier is trusted, the supplier can adopt skip lot inspection or relaxed inspection. In the sampling plan, the transfer rules are sufficient to ensure the basic guarantee 5
http:
GB/T13393-2008
(h),(i)
(d,),(e),
Table 3 (continued)
If the management is not complicated and the cost is not high, sequential, five-time or two-time sampling (see 7, 1.7.2) is adopted. This situation mostly requires variable inspection and it sampling. Whether it is sampling by count or sampling by count, attention should be paid to PR, CR, PRQ and CRQ (see 6.2, 6, 3.6.4)
This kind of shape involves that the product characteristics under consideration are normally distributed. But the inspection can be either quantitative or counting. Acceptance criteria are determined by knowledge of the distribution (see 6.2, 6.3, 6.4). All sampling standards can be used (see Chapter 2). The selection of the appropriate plan depends on other factors. Therefore, a longer process is required. The identification of the sampling plan is more difficult, although the sample can be appropriately smaller. No additional comments
If the batch is continuous, use the transfer rule and start with strict inspection and switch to normal inspection when the quality is confirmed. When random sampling is difficult, it is best to use primary sampling. Secondary or quintic sampling can be used in conjunction with subsampling selection rules. Sequential sampling adopts a small sample with greater risk, but under long-term quality control, all attribute sampling standards can be used. The selection of the appropriate plan depends on other factors. If the inspection is quantitative, the expected value can be converted into a count value. If information about the distribution is available in the long term, quantitative sampling can be used. In the case of the initial supply, there is no information about the distribution, and the use of quantitative sampling may increase the risk. Note 1: When the supplier is still responsible for the batch that has been judged to be accepted after the sampling inspection (see 1.1 (b)), GB/T2828.1, GB/T8051 or GB/T2828.3 can be used in the case of counting inspection; GB/T6378.1 (see Table 2a) can be used in the case of quantitative inspection. If the supplier is no longer responsible after the product is accepted, in order to ensure the quality of the batch, GB/T8051, GB/T13264, GB/T13262 can be used in the case of counting inspection; GB/T 8054 (see Table 2b) can be used in the case of counting inspection. The selection of the appropriate plan also depends on other factors (see Table 3). Note 2: In the case of a single product flow, where non-acceptance of a batch may result in economic losses or factory closure (see 4.1 ()), if the inspection cost is not high and there is no destructive inspection (see 4.1 (e), 4.1 (f), the AOQL. scheme (see Table 2a (d)) can be adopted, with the average inspection quality upper limit (AOQI.) as the quality parameter. When the batch is not accepted, a full inspection is carried out and the defective products in the batch are replaced by qualified products. If the source of equivalent products is large (see 1.1 (dl)), after considering other factors, select the appropriate scheme. 5 Negotiation procedure for selecting sampling scheme
The supplier and the purchaser shall select a reasonable acceptance sampling scheme through negotiation based on the quality requirements of the product, the delivery and acceptance regulations. The specific negotiation procedure is as follows:
a) Clarify the criteria for whether the unit product meets the regulations and what is a qualified product and a defective product; b) Clarify whether each unit product is qualified or the batch is qualified. If each unit product is required to be qualified, acceptance sampling cannot be used;
The sampling plan should be selected based on the risk factors accepted by both the supplier and the purchaser. Other factors such as the average sample size should also be considered: the supplier should understand the probability of the batch being accepted when the quality meets the requirements; the purchaser should understand the probability of the batch being accepted when the quality is inferior to a certain specified quality:
d) Clarify the sampling plan and the criteria for judging whether the batch is qualified or not. 6 Other factors affecting the selection of sampling plans
6.1. Percentage inspection of nonconforming products and number of nonconforming products per hundred units inspection To determine whether to use percentage inspection of nonconforming products or number of nonconforming products per hundred units inspection, the following factors should be considered: a) Percentage inspection of nonconforming products is concerned with whether there are nonconforming products on the unit product. One nonconforming product or multiple nonconforming products on a unit product are treated as a nonconforming product. b) Number of nonconforming products per hundred units inspection requires counting each nonconforming product on the inspected unit product. If there may be countless nonconforming products on a unit product (such as defects or pinholes on a casting), number of nonconforming products per hundred units inspection should be used. c) It is better to use one of the percentage inspection of nonconforming products and number of nonconforming products per hundred units inspection for a certain product than to frequently change from one to the other. Keeping records is very important for improving product quality. From the perspective of keeping records, number of nonconforming products per hundred units inspection is better. This record contains all the information related to nonconformities. If the number of nonconformities is used for inspection, some nonconformities may be missed. Unless the product quality is very good or the conformity of each characteristic is very good, the more types of nonconformities there are, the greater the chance that the batch will not be accepted when the number of nonconformities per 100 units is used. 6.2 Producer's risk and user's risk The risk introduced by acceptance sampling can be calculated. Since the sample only constitutes a small part of the batch, acceptance sampling may bring risks to the producer and the user. A good batch may not be accepted. The probability of making this mistake is called the producer's risk (denoted by PR); conversely, a bad batch may be accepted. The probability of making this mistake is called the user's risk (denoted by CR). Using the OC: curve of the sampling plan, the probability of a batch (or process) being accepted can be read when the batch (or process) quality is known. For a certain specified good quality, the producer requires acceptance with a high probability. This good quality is called the producer risk quality (expressed by PRQ), and the corresponding non-acceptance probability is the producer risk. On the other hand, for a certain specified poor quality, the user requires acceptance with a low probability. This poor quality is called the user risk quality (expressed by CRQ), and the corresponding acceptance probability is the user risk (see Figures 1 and 2). Producer's risk
User's risk
Producer's risk point (PKP)
User's risk point (CRF)
Producer's risk
Quality (RQ)
User's risk
Quality (CRQ)
Batch (process) quality
OC curve determined by producer's risk quality PRQ and user's risk quality CRQFigure 1
Producer's risk
User's risk
Producer's risk point (IRI)
User's risk point (CRP)
Acceptance quality limit
Batch (process) quantity
Figure 2 OC determined by acceptance quality limit (AQL) and limiting quality (LQ) Curve 7
http:
GB/T13393—2008
6. 3 Selection of AQL, PRQ, LQ and CRQ values 6.3.1 Selection of AQL and PRQ
AQL. and PRQ are both allowable quality indicators specified for the purpose of sampling inspection. The difference is that PRQ indicates the quality of a batch, while AQI. indicates the quality level of a process or a series of batches. The following factors should be considered when specifying AQL (or PRQ): a) The supplier's batch average quality (or batch quality) is required to be no worse than AQL (or PRQ), and this quality is achievable; for the purchaser, AQL (or PRQ) is a reasonable quality, which is a compromise between meeting requirements and being able to afford it. b) If the required product is easy to buy, and the defective product can be replaced by a qualified product, the AQ value can be appropriately larger. On the contrary, if a defective product may cause a certain important component of the equipment to fail and cannot be replaced by a qualified product, a smaller AQL (or PRQ) should be used.
c) Consider how many parts are included in the whole machine. If a piece of equipment includes two parts, and the defective product percentage of the equipment is required to be 1, then the defective product percentage of each component must be smaller. The defective rate of the whole machine is not obtained by simply adding the defective percentages of each component.
If the whole machine includes π equally important parts, and if one part fails, the whole machine will fail, then the defective rate of the whole machine (expressed as a decimal) X and the defective rate of the parts (expressed as a decimal): have the following relationship: X = 1 (1 z)\
By analyzing past data to estimate the process mean, this estimate or a value slightly less than it is defined as AQL. When the AQL is given, if the tail of the ()C curve of the selected sampling plan cannot meet the requirements of the user, a more stringent AQL needs to be specified.
Don't always select the AQL before making other choices. If necessary, the "backward deduction" method can be used to select a sampling plan according to other criteria and then reverse the required AQL value through the sampling plan table. When using the reverse deduction method, it is advisable to consider an important point with low acceptance probability on the OC curve 1, or consider other economic criteria. g) Set a known satisfactory quality level (or quality) of similar products as the AQL (or FRQ). h) Temporarily specify an AQL (or PRQ), and then adjust it according to usage and experience. i) Establish a cost model and select the AQL (or PRQ) to minimize the total cost. 6.3.2 LQ Selection of LQ and CRQ
LQ and CRQ are for the purpose of sampling inspection, limited to a quality level with a low probability of acceptance. The method of specifying LQ and CRQ is similar to that of specifying AQL and PRQ. 6.4 Determination of inspection level
Inspection level (CRQ) is a relative inspection quantity indicator of the sampling plan, which best relates the sample size to the batch. The following factors should be considered:
a) After specifying AQL (or PRQ), determine the inspection level by considering LQ (or CRQ). For example, in GB/T2828.1, first find the OC curve of the normal inspection plan that is closest to meeting the requirements of AQL and LQ, and OC The AQL and judgment number corresponding to the curve are then found from the normal inspection sampling plan table. Finally, the inspection level corresponding to this sample size code and the given batch range is found in the sample size code table. If there is no inspection level corresponding to the code and batch range in the table, the AQI and/or IQ value should be reconsidered, or this practice should be abandoned. b) For economic reasons or the destructive nature of the test, sometimes a low inspection level has to be used, such as S-1 in GB/T2828.1. At this time, the discrimination of the sampling plan is reduced. However, if there are inspection records for a series of consecutive batches, the use of cumulative aggregation of samples may indicate that the inspection level is acceptable. ([3/T8054 mainly uses isolated batch inspection and does not use inspection level retrieval sampling plan. At this time, both the producer and the user should pay attention to the risk quality of the producer and the risk quality of the user. 8
7 Comparison of sampling inspection methods
7.1 Counting sampling inspection and measurement sampling inspection GB/T 13393—2008
If the observed values follow a normal distribution, the count sampling test has some advantages over the count sampling test. In order to provide the manufacturer (or user) with some quality protection, the former requires a smaller sample and can provide more information about the product quality; the latter is not restricted by the assumption of the distribution form, is easier to use, and is easier for inspectors to understand and accept. If it is difficult to verify normality or if you want to use it more easily and would rather test more samples; or if there are many test indicators and it is too cumbersome to use the variable sampling test, you can convert the measured value into the count value and then use the count sampling test.
When the test cost is very high or the test is destructive, the use of variable sampling test is substantially superior to the count sampling test. If there are many test indicators, you can also consider using the count sampling test for most of the test indicators and only use the count sampling test for one or two important indicators.
Table 1 gives the sample sizes corresponding to several sample size codes in GB/T 2828.1 and GB/T 6378.1 when using inspection level II and normal inspection subsampling plans.
Table 4 Comparison of sample sizes
Sample base code
GB/T 2828. 1
7.2 Sample base code for single, double, multiple and sequential sampling plans
GB/T 6378.1 (unknown)
GB/T 6378.1(known)
not more than 3
not more than 7
not more than 14
not more than 33
not more than 95
For the single sampling scheme with non-zero acceptance number, a secondary or multiple sampling scheme can be found to make their (C curves very close. Therefore, in order to compare the single, secondary and multiple sampling schemes, it is not enough to only consider the (C curve. The following characteristics should be considered comprehensively: a) Simplicity: The single sampling scheme is the easiest to explain and use. The single sampling scheme may draw the first sample, which is slightly troublesome to manage. Multiple sampling schemes and sequential sampling schemes are more complicated. b) Variation of test maximum: The sample maximum of the single sampling scheme is fixed, so the test maximum can be predicted. The test quantity of the two-drink, multiple-drink and sequential sampling schemes varies with the test results of the previous samples; but the average number of samples can be calculated for any given input quality. When the hostages are particularly good or particularly bad, the average number of samples is almost the least. c) Difficulty of sampling: It is sometimes convenient to extract the second sample. Extracting two samples is not more difficult than extracting one sample with a joint sample. However, sampling is sometimes time-consuming. It is difficult to extract the bottom of the first sample and then the remaining part of the batch to extract the second sample. In this case, it is best to use a single sampling. d) Duration of the test: If the test duration is long and it is possible to test more samples at the same time, it is better to use a single sampling plan. If one or two samples can be tested at a time, consider using a group, multiple or sequential sampling plan. e) Types of non-conforming products: If there are many types of non-conforming products, secondary and multiple sampling will be more complicated. For example, all quality characteristics of the product require the first sample to be tested, and some of the quality characteristics may also require the second sample, the third sample, etc., which makes it difficult to use labor and inspection equipment effectively.
Usually, complex inspections require simple sampling plans; more complex sampling plans can be used for simple inspections, and good results can be achieved.
Example 1 In order to test the preservation quality of canned meat, several cans are stored under specific environmental conditions for three weeks. Three sampling plans were considered: a single sampling plan with 80 samples, a double sampling plan with each sample consisting of 5 bottles, and a five-sample plan with each sample consisting of 20 bottles. If a single sampling plan is used, the test results can be obtained within two weeks. With double sampling, the results may be obtained within three weeks or six weeks. With five-sample, the test may take 9 weeks.
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.