Data-based control of heterogeneous population systems

Distributed population systems consist of large numbers of similar components that are structurally equivalent but differ in a few relevant variables. Such systems occur in a wide range of processes in chemical engineering, and the prediction, monitoring, and control of the resulting heterogeneous system dynamics are needed to ensure the processes are operated in the best possible way.
The goal of this project is to develop a methodological framework for the data-based modelling and control of heterogeneity in distributed population systems, which uses so called Gaussian processes to represent the heterogeneity and any uncertainty about it in a mathematically efficient way. Instead of relying on models constructed from mechanistic knowledge, the framework aims at a process representation that is derived directly from high-throughput measurement data available with recently developed measurement technologies. Numerical methods for the prediction of future process dynamics and model-predictive process control that takes the heterogeneity into account will be developed and evaluated along two case studies from (bio-)chemical engineering.