Abstract

In this article we propose a new online control system aiming to lower the instants in which the production process migrates from being in- to out-of-control state, which generates an increase in the non-conformity rates. As shifts from in- to out-of-control are non-deterministic, a sample of size n is collected, for each m or L units produced, and each element from the sample is imprecisely classified as conform or non-conform (that is, there may be classification errors). If the amount of conform units from the sample is equal or greater than a, the process would not be adjusted and another sample would be collected after m units produced. If the quantity of conform units is inferior to a, the process would be adjusted and another sample would be collected after L units produced, given that L > m. A genetic algorithm is proposed to approximately find the values of a, n, m, and L that minimize all costs involved in the process being controlled. All procedures are illustrated through a detailed numerical example that attests the efficacy and efficiency of the proposed online control system.

摘要

在本文中，我们提出了一种新的在线控制系统，旨在降低生产过程从受控状态转变为失控状态的瞬间，从而导致不合格率增加。由于从受控到失控的转变是不确定的，因此对于产生的每m或L个单位，收集大小为n的样本，并且样本中的每个元素都被不精确地分类为合格或不合格（即，可能存在分类错误）。如果样本中合格单位的数量等于或大于a，则不会调整过程，并在生产m 个单位后收集另一个样本。如果合格单位的数量低于，假设L  >  m ，则将调整流程，并在生产L 个单位后收集另一个样本。提出了一种遗传算法来近似找到a、n、m和L的值，从而最小化受控过程中涉及的所有成本。所有程序都通过详细的数值示例进行说明，证明了所提出的在线控制系统的功效和效率。