Hybrid process models for process optimisation, monitoring and control

Abstract

Hybrid models aim to describe different components of a process in different ways. This makes sense when the corresponding knowledge to be represented is different as well. In this way, the most efficient representations can be chosen and, thus, the model performance can be increased significantly. From the various possible variants of hybrid model, three are selected which were applied for important biotechnical processes, two of them from existing production processes. The examples show that hybrid models are powerful tools for process optimisation, monitoring and control.

Bioprocess Modeling and Control

“…Now the technique was upgraded with better results. For instance [30] the applied BCAPM for feed control in the production of monoclonal antibodies allows to improve the yield with 43%.…”

Integrated framework of nonlinear prediction and process monitoring for complex biological processes

“…This step is called the nonlinear iterative partial least squares (NIPALS) algorithm [3]. Although, there exists a faster and more stable algorithm using eigenvectors, we use NIPALS to give readers a clearer picture of PLS outer projection.…”

“…As the number of measured variables in biosystems increases, bioprocess monitoring with diagnosable statistical techniques becomes very important. However, high dimensionality, collinearity, and nonlinearity in experimental or historical data often make it difficult to apply statistical modeling and analysis techniques [1][2][3][4][5][6]. Traditionally, principal component analysis (PCA) and partial least squares (PLS) are used to statistically monitor a biological process.…”

Hybrid semi-parametric modeling in process systems engineering: Past, present and future

“…smaller than one, greater than minus one. Additionally, in case that only few experimental values are available and a simple model of the kinetic rates is available, the model can be used to provide kinetic rate data for a pre-identification (before the identification relying on the experimental data is carried out) of the nonparametric model parameters (Galvanauskas et al, 2004;Graefe et al, 1999;Tsen, Jang, Wong, & Joseph, 1996).…”

“…Luebbert (1995, 1996) applied a weighting function in a serial hybrid semi-parametric model in order to coordinate the predictions of the kinetic rates by heuristic rules (the Monod model) and the ones by a nonparametric model. The kinetic rate predictions and the nonparametric predictions were weighted by a clustering approach (for details see also Galvanauskas, Simutis, & Luebbert, 2004), where more weight is given to the nonparametric model in regions where process data are available, while restricting it when extrapolating. This weighting method was also applied by Patnaik (2010), who however determined the weighting iteratively.…”