Investigation of an Alternative to Acceptance Sampling Through a Markov Chain Analysis of a Manufacturing Process Quality Control Program.

摘要

In this thesis, we investigate the examination of a manufacturer's in-house quality program as an alternative to acceptance sampling. The manufacturing process addressed is one which consists of a production section, capable of producing items at one of two levels of fraction nonconforming, and a quality control section which consists of a single p-chart. The quality levels that result from this manufacturing process are represented using a Markov chain. A method of estimating the fraction of nonconforming items produced by the process is developed. Confidence intervals on this fraction nonconforming are obtained and these values considered for use in an alternative acceptance criteria for lots. When the upper confidence limit on the lot fraction nonconforming does not exceed the Acceptable Quality Level, there is considerable confidence that lots randomly selected from the manufacturing process will be acceptable without acceptance sampling.