A regularized Moving Horizon Estimator for combined state and parameter estimation in a bioprocess experimental application

作者：Andrea Tuveri, Caroline S.M. Nakama, José Matias, Haakon Eng Holck, Johannes Jäschke, Lars Imsland, Nadav Bar · 发表于：Computers & Chemical Engineering · 年份：2023 · DOI：10.1016/j.compchemeng.2023.108183 · 被引用次数：12 · 研究领域：Advanced Control Systems Optimization、Fault Detection and Control Systems、Viral Infectious Diseases and Gene Expression in Insects

Due to the lack or high costs of measurement devices to monitor and control metabolites in microbial cultivation processes, state estimators are often required. These estimators depend on available on-line measurements and model dynamics. However, they are often characterized by simple models due to the lack of full knowledge on the process dynamics and high variability in the cell metabolism. This causes uncertainty in the model parameters and therefore the necessity of on-line model adaptation, for instance through simultaneous state and parameter estimation. However, these estimation problems are often ill conditioned. The Moving Horizon Estimator (MHE) is a good candidate in this context, since it easily allows enforcing hard constraints as well as regularization to address the ill-posedness. In this work, we present a method for simultaneous state and parameter estimation in the absence of full state measurements, with the aid of two regularization methods, in a microbial fed-batch cultivation.