The use of Bayesian forecasting to make process adjustments during transitions

In many manufacturing operations, a system may exhibit dynamic behavior before reaching a steady-state level. This is most frequently associated with a transition in production like a product style change or a grade change. During the transition phase, the output does not respond instantaneously to a change in input. However, there is typically some information about the past transition phase performance available. We develop an adjustment policy for transition periods based on using a Bayesian forecast to incorporate the prior information. We present computational results showing average process improvements under various system and noise disturbance conditions.