In capital intensive manufacturing processes the amount of experimental data available may not be adequate to build a sufficiently accurate statistical model. This paper presents a new Bayesian predictive approach for sequential experimentation that aims to combine experimentation and optimization stages. A dual control approach that simultaneously considers model estimation and process optimization objectives is adopted. By combining these two goals the objective is to more quickly start production with little or no historical data from the process.

This content is only available via PDF.
You do not currently have access to this content.