An analysis of Taguchi's on-line quality monitoring procedure for attributes with diagnosis error (Q2759390)

The authors propose an extension of Taguchi's on-line quality monitoring procedure for attribute quality characteristics. The proposed model takes into account the diagnosis errors. The setting of procedure is that a production process that produces items independently at each time unit and whose fraction defective shifts from \(0\) to \(\pi\), \(0<\pi\leq 1\), after a geometrically distributed number of items is produced. The monitoring strategy consists of inspecting a single item after every \(m\) produced and carrying out a process adjustment routine that restores fraction defective to its initial value as soon as a defective item is found. Inspection is instantaneous and imperfect in the sense that with probability \(\alpha\), \(0<\alpha<1\), a non-defective item is considered defective and with probability \(\beta\), \(0<\beta<1\), a defective item is considered non-defective. Assessment of the optimum inspection interval \(m^{*}\), that is, the inspected interval that minimizes the producer long-run expected cost per item produced is examined.