以下是在某论坛找到的回答，有兴趣的可以看看
Comments
Carlos said: "I would like to know if I can design a sampling plan with AQL=0"
I assume that you are talking about an attribute sampling plan for percent defective.
If you define AQL as zero percent defective, the probability of acceptance at AQL will be 100%. The problem with doing that is that all sampling plans have oc curves that go through the point AQL=0, Pa=1.0 (100%) . Therefore that point cannot discriminate between one sampling plan and another.
So setting AQL=0 is not useful in designing a sampling plan.
To say it another way, with AQL=0 the producer's risk, alpha = (1-Pa)=0. In other words, if a lot does not contain any defectives, there is no way that a sample from the lot can contain any defectives, and the lot can only be accepted.
Perhaps the person at the chemical company is interpreting (AQL, alpha) as the producer's point. For more about this, see:
RSS Feed
14 个回复
jxm123a (威望:0)
赞同来自:
Comments
Carlos said: "I would like to know if I can design a sampling plan with AQL=0"
I assume that you are talking about an attribute sampling plan for percent defective.
If you define AQL as zero percent defective, the probability of acceptance at AQL will be 100%. The problem with doing that is that all sampling plans have oc curves that go through the point AQL=0, Pa=1.0 (100%) . Therefore that point cannot discriminate between one sampling plan and another.
So setting AQL=0 is not useful in designing a sampling plan.
To say it another way, with AQL=0 the producer's risk, alpha = (1-Pa)=0. In other words, if a lot does not contain any defectives, there is no way that a sample from the lot can contain any defectives, and the lot can only be accepted.
Perhaps the person at the chemical company is interpreting (AQL, alpha) as the producer's point. For more about this, see:
www.samplingplans.com/aqlprimer.htm