Multiattribute acceptance sampling plans - Library(ISI Kolkata ...
Multiattribute acceptance sampling plans - Library(ISI Kolkata ...
Multiattribute acceptance sampling plans - Library(ISI Kolkata ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
equals to p i in the long run. We assume the number of defects for r distinct characteristics<br />
in a unit are indepedently distributed. The output of such a process is called a product of<br />
quality (p 1 , p 2 , .., p r ).<br />
In this case x i = number of defects w.r.t. attribute i in a random sample of size n, i =<br />
1, 2, ..., r. The expression (0.5.1) holds here exactly for a realization of x = (x 1 , x 2 , ..., x r ) in a<br />
random sample size n and the probability of <strong>acceptance</strong>, P (p) given by (0.5.2) holds exactly<br />
too.<br />
Along with the two situations under appropriate assumptions mentioned before, where<br />
we count the number defective units w.r.t. different characteristics in the sample, we include<br />
under Poisson conditions the present situation of defect verifications also w.r.t. a number<br />
of characteristics under considerations. The type B OC function given in (0.5.2) for the<br />
appropriate <strong>acceptance</strong> criterion stated covers all these situations of defectives / defects<br />
under given assumptions discussed.<br />
0.5.1.2 Chapter 2: <strong>Multiattribute</strong> <strong>sampling</strong> scheme based on AQL<br />
This chapter is devoted to the objective of establishing, for a multiattribute situation, a<br />
<strong>sampling</strong> scheme in line with available international standards tabulated on the basis of<br />
Acceptable Quality Level (AQL). The published <strong>plans</strong> most widely used is the US Military<br />
standard 105D (1963) [ abbreviated as MIL-STD-105D ], which has been adopted as an<br />
international standard (ISO 2859). The standard is highly suitable for the consumers.<br />
The sample size used in MIL-STD-105D is determined from the given lot size and by the<br />
choice of inspection level. The <strong>acceptance</strong> number (c) is arrived at, such that it remains more<br />
or less same for the same value of the product of sample size and the AQL to ensure that<br />
the Poisson probability of <strong>acceptance</strong> (rejection) remains same for the <strong>plans</strong>. This ensures<br />
a desired producer’s risk. There are 13 sets of n.AQL values and 11 <strong>acceptance</strong> numbers<br />
chosen in such a way that the producer’s risk (excepting for c = 0) varies from maximum<br />
of 9.02% to a minimum of 1.44%. With increase in the n.AQL values, c increases, so that<br />
producer’s risk decreases up to a level of 2% and thereafter it is kept at less than 2%. There<br />
is no theoretical foundation for the relation between sample size and lot size. As the lot size<br />
increases, the sample size increases but at a lesser rate such that n/N → 0 as N → ∞.<br />
The Standard prescribes that separate <strong>plans</strong> are to be chosen for the different classes of<br />
attributes. For example, the plan for a critical defect will have generally a lower AQL value<br />
than the plan for a major defect and the plan for a major defect will have an AQL value<br />
lower than the plan for a minor defect.<br />
To examine the consequences of constructing a <strong>sampling</strong> plan by this method for our<br />
multiattribute situation we first consider the effective producer’s risks.<br />
Secondly, we consider it reasonable to expect that the OC function should be more sensitive<br />
to the changes in the defect level of more important attributes, particularly, in a situ-<br />
18