An optimization-free Fisher information driven approach for online design of experiments

• DOI: 10.1016/j.compchemeng.2024.108724
• Publication Date: 2024-05-14T01:35:19Z
• Authors: Andrea Friso; Federico Galvanin
• Keywords: Fisher information; Identification; Design of experiments; Bayesian Optimization

Developing mathematical models used to elucidate reaction kinetics plays a crucial role in the design, control, and optimization of chemical processes. One most challenging tasks kinetic model identification is precise estimation unknown parameters. This challenge can be effectively addressed through application Model-Based Design Experiments (MBDoE) techniques, which enable design experiments facilitating parameter with minimal runs analytical resources. Nevertheless, MBDoE techniques rely on an procedure that susceptible parametric uncertainty, making computationally intensive prone issues local optimality. are also employed online procedures expedite autonomous platforms. As result, ensuring rapid convergence becomes imperative mitigate numerical during operational In this paper Fisher Information Matrix Driven (FIMD) approach introduced tackle these challenges. The methodology integrates sampling-based experimental experiment ranking based FIM select informative at each iteration. effectiveness proposed examined discussed via two different case studies increasing complexity: fed-batch reactor fermentation baker's yeast carried out nucleophilic aromatic substitution flow system.