Process parameters are estimated from an in-control sample in the Phase I of the control chart implementation. Thisintroduces an additional source of variability into Phase II. This paper studies the effect of estimation on the perfor-manceof the Bayesian Control chart and compares it to that of the Cusum chart. Performance measures used forcomparison are the expected number of false alarms (EFA) and the average out of control run length (ARL1). Theconditional and marginal performance of both these charts is estimated from simulations. Recommendations are madeon the sample size to be used to generate estimates in Phase I.
展开▼
