Professional Documents
Culture Documents
1987manjula@gmail.com
Abstract
Key Words DSP, DSPNSP, Fuzzy Logic, Crisp Value and OC.
Introduction
Definition 1: The fuzzy subset 𝑁 of real line IR, with the membership function 𝜇𝑁 : 𝐼𝑅 →
[0,1] is a fuzzy number if and only if (a) 𝑁 is normal (b) 𝑁 is fuzzy convex (c) 𝜇𝑁 is upper
semi continuous (d) supp (𝑁) is bounded.
Definition 4: Due to the uncertainty in the 𝑙𝑖 ’s values we substitute 𝑙𝑖 , a fuzzy number, for
each 𝑙𝑖 and assume that 0< 𝑙𝑖 <1 all i . Then X together with the 𝑙𝑖 value is a discrete fuzzy
probability distribution. We write 𝑝 for fuzzy P and we have 𝑃 𝑥𝑖 = 𝑙𝑖
Let A= {𝑥1 , 𝑥2 , … . 𝑥𝑙 } be subset of X. Then define:
𝑙
𝑖=1 𝑙 𝑖
𝑃 𝐴 𝛼 = (1)
𝑠
𝑙
For 0<α< 1, where stands for the statement “𝑙𝑖 ∈ 𝑘𝑖 [α], 1<i<n, 𝑖=1 𝑙𝑖 = 1”
This is our fuzzy arithmetic.
Definition 5: Let x be a random variable having the Poisson mass function. If P(x) stands for
the probability that X= x, then
𝑒 −𝜆 𝜆 𝑥
P(x) = (2)
𝑥!
For all α ∈ [0,1]. Let X be a random variable having the fuzzy binomial distribution
and 𝑃 in the definition 4 are small. That is all are p ∈ 𝑝 sufficiently small. Then
𝑃[a,b][α] using the fuzzy poisson approximation.
𝑏 𝑒 −𝜆 𝜆 𝑥
Then 𝑃[a, b] [α]= 𝑥=𝑎 𝑥! (4)
Sometimes the parameters of sampling plans cannot be expressed as crisp values, that
can be stated as “approximately”, “around”, or “between”. Fuzzy set theory is a very usable
tool to convert these expressions in to mathematical functions. In this case, acceptance
probability of sampling plans should be calculated with respect to fuzzy rules. In this section
New Screening double sampling plan is analyzed by taking into account of fuzzy Poisson
distributions.
New Screening Procedure on Double Sampling Plan
The purpose of this paper is to describe a method and to present a set of tables for
constructing two and three stage drug screening procedures of the type discussed by Armitage
and Schneider man(1958) and Schneider man (1961). These procedures allow rejection at any
stage but acceptance at only final stage. Similar procedures have been advocated by Davies
(1957) and Dunnett(1961), based on this operating characteristic curve and accept-reject rules for
two and three stage screening procedures had been derived by Roseberry and Gehan (1964).
Mixed sampling product control for costly or destructive items by Deva Arul (2011) for
switching variable to attribute plan for accepting the lot. Based on this screening procedure and
switching rule of variable to attribute gives an idea for creating a new concept in double sampling
plan. Generally we are going to second sample when the defective lies in between two acceptance
number, but in this procedure we are allow to take second sample even it is not lie in the region
but under the condition of past experience ( i.e., last two rejection is nearer to third acceptance
number).
New Screening Procedure on Double Sampling Plan with Fuzzy Parameter Using
Poisson distribution
If the size of sample be large and „p’ is small then the random variable „d‟ has a Poisson
approximation distribution with λ=np. So, the probability for the number of defective items to
be exactly equal to „d’ is:
𝑒 −𝑛𝑝 (𝑛𝑝 )𝑑
𝑃 𝑑 = (5)
𝑑!
Suppose that we want to inspect a lot with the large size of „N’, such that the proportion of
damaged items is not known precisely. So we represent this parameter with a fuzzy number 𝑝
as follows:
𝑝 = 𝑎1 , 𝑎2 , 𝑎3 , 𝑝 ∈ 𝑝 1 , 𝑞 ∈ 𝑞 1 , 𝑝 + 𝑞 = 1.
A DSPNSP sampling plan with a fuzzy parameter if defined by the sample size „n1’,’n21’,
‘n22’ and acceptance number „c1’, „c21’, ‘c22’ and if the number of observation defective
product is less than or equal to „c1’, the lot will be acceptance. If „N’ is a large number, then
the number of defective items in this samples (d1,d21,d22) has a fuzzy Poisson distribution
with parameter 𝜆 = 𝑛𝑝 So the fuzzy probability for the number of defective items in a sample
size that is exactly equal to „d’ is:
𝑃 𝑑 − 𝑑𝑒𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝛼 = [𝑃𝐿 𝛼 , 𝑃𝑈 𝛼 ] (7)
𝑒 −𝜆 𝜆 𝑑
𝑃𝐿 𝛼 = 𝑚𝑖𝑛 │𝜆 ∈ 𝑛𝑝[𝛼] , (8)
𝑑!
𝑒 −𝜆 𝜆 𝑑
𝑃𝑈 𝛼 = 𝑚𝑎𝑥 │𝜆 ∈ 𝑛𝑝[𝛼] (9)
𝑑!
𝑐
𝑒 −𝜆 𝜆 𝑑
𝑃𝐿 𝛼 = 𝑚𝑖𝑛 │𝜆 ∈ 𝜆[𝛼] ,
𝑑=0 𝑑!
𝑐
𝑒 −𝜆 𝜆 𝑑
𝑃𝑈 𝛼 = 𝑚𝑎𝑥 │𝜆 ∈ 𝜆[𝛼]
𝑑=0 𝑑!
Operating characteristic curve is one of the important criteria in the sampling plan. By
this curve, one could be determined the probability of acceptance or rejection of a lot having
some specific defective items. The OC curve represents the performance of the acceptance
sampling plans by plotting the probability of acceptance a lot versus its production quality,
which is expressed by the proportion of defective items in the lot. OC curve aids in selection
of plans that are effective in reducing risk and indicates discriminating power of the plan.
The fuzzy probability of acceptance a lot in terms of fuzzy fraction of defective items
would be as a band with upper and lower bounds. The uncertainty degree of a proportion
parameter is one of the factors that bandwidth depends on that. The less uncertainty value
results in less bandwidth, and if proportion parameter gets a crisp value, lower and upper
bounds will become equal, which that OC curve is in classic state. Knowing the uncertainty
degree of proportion parameter and variation of its position on horizontal axis, we have
different fuzzy number (𝑝) and hence we will have different proportion (p) which the OC
bands are plotted in terms of it.
The operating characteristic (OC) curve displays the discriminatory power of the
sampling plan, shows the probability that a lot submitted with a certain fraction defective will
either be accepted or rejected. In a double sampling plan, the decision of acceptance or
rejection of the lot is taken on the basis of two samples. The lot is accepted on the first
sample if the number of defective units (d1) in the first sample is less than the acceptance
number c1. The lot is accepted on the second sample if the number of defective units (d1 + d2)
in both samples is greater than c1 and less than or equal to the acceptance number c2. But in
DSPNSP the second sample is taken even it is not lies in the acceptance region but some
condition. Therefore, if Pa1(p), Pa21(p) and Pa22(p) denote the probabilities of accepting a lot
on the first sample and the second samples, respectively, the probability of accepting a lot of
quality level p is given by:
𝑐1 𝑒 −𝜆 𝜆 𝑟 1
Where, Pa1(p) = 𝑟=0 𝑟!
𝑘 𝑟
𝑐2 𝑒 −𝜆 1 𝜆 1 𝑐2 −𝑘 𝑒 −𝜆 2 𝜆 2
Pa21(p) = 𝑘=𝑐1 +1 𝑟=0
𝑘! 𝑟!
𝑟
𝑐2 +1 𝑒 −𝜆 3 𝜆 3
Pa22(p) = 𝑘=𝑐1 𝑟!
Example
With sample size and acceptance number (50;0) for SSP,(50,50;0,1) for DSP and
(50,50,5;0,1,2) for DSPNSP the probability of acceptance in fuzzy for DSPNSP is tabulated in
Table:1.
Figure 1: OC band for a 𝑫𝑺𝑷𝑵𝑺𝑷 with fuzzy parameter of (n; c) is (50, 50, 5; 0, 1, 2)
0.8
0.6
LB
0.4 UB
0.2
0
0 0.02 0.04 0.06 0.08 0.1
Li 𝑝 Pa
0 [0,0.01] [1, 0.9097]
0.01 [0.01,0.02] [0.9097, 0.7357]
0.02 [0.02,0.03] [0.7357,0.5578]
0.03 [0.03,0.04] [0.5578, 0.4061]
0.04 [0.04,0.05] [0.4061,0.2873]
0.05 [0.05,0.06] [0.2873, 0.1991]
0.06 [0.06,0.07] [0.1991, 0.1359]
0.07 [0.07,0.08] [0.1359, 0.0915]
0.08 [0.08,0.09] [0.0915, 0.0611]
0.09 [0.09,0.1] [0.0611, 0.0404]
0.1 [0.1,0.11] [0.0404, 0.0265]
Conclusion
14. Uma G. and Ramya K. (2015): Impact of Fuzzy Logic on Acceptance Sampling
Plans – A Review, CiiT International Journal of Automation and Autonomous
System, 7(7), 181 – 185.
16. Ramya K and Uma G. (2017): Determination of Quick Switching Double Sampling
System by Attributes under Fuzzy Poisson Distribution : Acceptance Number
Tightening, Current Trends in Information Technology, Vol. 7, No. 3, pp. 17 – 23.
17. Uma and Manjula (2018), New Screening Procedure on Double sampling Plan for
costly or destructive items, International Journal of Research in Advent Technology,
6(12), 3640-3646
18. Uma and Manjula (2019), A Technical study on variety of DSP and implementation
of New Screening Tri-Sampling Plan, International Journal of Research in Advent
Technology, Special Issue, 266-271.