Data Analytics for Powder Feeding Modelling on Continuous Secondary Pharmaceutical Manufacturing Processes

In this Thesis, a data-driven procedure of investigating raw materials variability in an industrial database is presented, together with a multivariate statistical modelling approach for the first unit operation of continuous tableting lines. The main objective of the Thesis is to explore the capabilities of using pattern recognition techniques to identify and model hidden patterns of similarities in a powder materials dataset and determine the effect of materials variability on the feeder.