摘要

Sometimes quality engineers do not have data to generate a distribution, especially prior to production. However, the quality control strategy still must be planned, and a preliminary distribution is required. As data come to hand, the distribution can be updated via an evolutionary approach. This article considers cast feature dimensions. Through consideration of casting theory, probability theory, and published data, it is argued that the lognormal distribution is the preferred assumption for the initial distribution. Finally, it is concluded that quality engineers can use more than Goodness of Fit (GOF) tests to choose distributions and should understand operations to generate distributions case by case.

  • 出版日期2013-7-1

全文