This research focuses on developing efficient methods to update meta-models based on measured input/output data from a multi-stage manufacturing process. It requires to characterize different sources of uncertainties and update the necessary parameters.

For that purpose, the researcher will use multi-scale physical models to build meta-models of the process. Moreover, the researcher will use Bayesian and ML statistics to estimate prediction error variances and update the meta-models based on input–output data measured from the process.

As a result, these meta-models will be used to directly optimize and find the optimum control points in the multi-stage manufacturing process.