(91a) Combining Data-Driven and First Principles Models: A Tutorial

Integrating physical, biological, chemical knowledge and “machine learning” is a critical aspect of developing industrially focused digital twins for monitoring, optimization, and design of (bio)chemical processes. However, identifying the correct model structure (e.g. the data-driven and the knowledge driven components to quantify the complex mechanisms) poses a severe challenge.

In this tutorial we will look at different hybrid modeling architectures, the variety of data-driven models that can be employed, as well as their benefits, and some insights into how to determine what is the best structure for a hybrid model.