Data-driven approaches to the modelling of bioprocesses

Abstract

Bioprocess modelling presents a challenging subject, which requires a meticulous modelling strategy. During the modelling process, experimental data form a key ingredient during structure characterization and parameter estimation. Accurate system identification can only be guaranteed if the experimental data contain sufficient information on the process dynamics. In this respect, sufficient effort should be spent on optimal experiment design in order to maximize the information that can be extracted from data, particularly because experimental data generation for bioprocesses is usually a time-consuming, labour-intensive and costly job. This paper reviews the modelling cycle of bioprocesses, emphasizing the need for careful experimental data collection. The concepts of optimal experiment design for parameter estimation are outlined in particular. Application of this methodology is illustrated for a case study involving the optimal estimation of two model parameters describing temperature dependence of microbial growth kinetics.