Build a sampling procedure using hypergeometric regression

Hello, i built some curves using hypergeometric regression in order to obtain informations about an hypothetic sampling. The curves are related to different sample size (80unit, 5%, 10%, 20%) with same sample population (5000).  They has shown different responses in terms of batch acceptance percentage, where the better condition is naturally the largest sample size (curve of 20% sampling is the 1st that fall down at 0% acceptance).
I think this curve has shown also that AQL is near to zero and they was built considering an acceptance level of defects equal to zero.
At this point I need some helps:
How is possible abstract the correct level of AQL of this curves? I think, it’s the maximum number of defects before sampling have a reaction, moving from 100% of acceptance level, to a smallest level e.g.: 99.9%. My idea is correct? This “smallest level” is an international standard value/reference?
How is possible build a curve using a prefixed number of accepted defect? (e.g.: 10 or 15 units - for me it means also a diffent AQL level using same data as N=batch population and n=sampled).
Please reply me considering that I used excel and I will use to complete the work.

Thank in advance for every technical consideration you’ll want kindly share with me.