The procedure to control on-line processes for attributes proposed by Taguchi et al. (1989) consists of sampling a single item at very m produced ones. If the examined item is non-conforming, the process is said to be out of control (the fraction of conforming item is reduced) and it is stopped for adjustment. However, the inspection procedure is subject to misclassification errors. So a probabilistic model was developed considering results of the repetitive and independent classifications of the examined item. Employing properties of an ergodic Markov chain, an expression of the expected cost function for the control system was obtained to be minimized by three parameters: sampling interval; number of the repetitive classifications; minor number of conforming classifications (among the repetitive classifications) to declare an item as conforming. A numerical example illustrates the proposed model.
on-line process control for attributes; misclassification errors; Markov chain
