Sampling Inspection Method Based On Small Sample Size From Complex Population

The problem discussed in this paper is setting acceptance rule for a specific kind of product. It is belongs to test inspections in term of Statistics. This kind of specific product is costly or dilapidated after testing, therefore small sample size is required. In addition, the testing should be carried out under different conditions for such a complex environment this kind of product would be used in. The inspection methods in use are supplement-testing and simple-sum method, which are shown to be inefficient in risk control in the first section of this paper.For inspections with costly samples from a complex population, the method of combined inspection is proposed in this paper based on the thoughts that preserve as much information of samples as possible.By compared with limited classification results, The method of combined inspection is proved to be providing an acceptable design with minimum (or approximate minimum) two real probability errors for a given small sample size.Furthermore, a suggestion on combined statistic selection is proposed based on simulation results.Finally, an example is given to illustrate how to reduce the number of samples at the cost of taking a little more risk, so that the experiment expenses could get further saved, which is really appreciable for costly products.