食品生产控制课件.pptVIP

Sampling Acceptance Terms Acceptance Quality Level (AQL) Proportion of defectives that we would be willing to tolerate in the lot Sampling Plans Since you can’t test 100% of product, need to collect enough information to make valid inferences. Need to ask: Where is the optimum sampling point? How should the sample be taken? How often should the sample be taken? How much sample? Examples You want to see if cherries are ready for harvesting so sample the cherries on the southern side of the orchard (because it’s near the road and easy to get to) You sample only the first truck of corn for sugar/starch ratio for canned corn for setting up processing for the next 10 trucks. Do you think the results will be valid? Do they represent the population of cherries or corn? Sampling Sizes 0% – If you rely on CofA (supplier reputation), luck or ignorance 100% - automation weight checkers metal detectors machine vision inspections when hand packing Statistically based Types of Random Samples 1. Simple 2. Systematic 3. Stratified 4. Cluster Types of Random Samples 1. Simple N = 100 n = 10 Types of Random Samples 2. Systematic After random selection of the k item per group, selects every kth. unit of the population Example: Pull 10 bottles every 30 minutes to weigh Types of Random Samples 3. Stratified Based on physical separation, the population can be divided into sub-sets (strata) Each stratum is randomly sampled to detect differences among them Example: Layers in a hopper or bin formed from different loads Types of Random Samples 4. Cluster If the population is naturally divided in clusters, randomly select clusters from which the population sample is taken Say you want 50 cans but they are in 12 can cases in a warehouse Cluster size Subsample size Total sample size 50 cases 1 can from each 50 25 cases 2 cans from each 50 10 cases 5 cans from each 50 5 cases 10 cans from each 50 Sampling Plan Terms n – number of sample units c – maximum allowable number of non-confor