(14f) Some Perspectives On the Impact of BIG Data On Process Systems Engineering

This presentation offers some advice to young researchers based on my own experiences in industry and academia.  It also provides a personal perspective on the impact that BIG data is having on PSE. The past has seen the rise of multivariate latent variable methods to treat many large messy data problems, but the data sets continue to get bigger in almost every area. 

Control and optimization have almost invariably been based around the use of theoretical models and or structured empirical models fit to plant data.  However, what if many informative measurements such as color images, vibrational spectra, etc. are available to help define the state of the process, or what if the appearance of a product obtained from an image is the objective of the optimization?  A general conclusion might be that, as we continue to collect more and more information through more advanced sensors, our traditional PSE reliance on fundamental or structured empirical models becomes less and less viable and hybrid approaches are needed.