Self‐tuning run to run optimization of fed‐batch processes using unfold‐PLS

Abstract

in Wiley InterScience (www.interscience.wiley.com).Unfold Partial-Least Squares (u-PLS) is a modeling method successfully applied to batch-process monitoring and end quality prediction. This method is integrated in a self-tuning optimization algorithm, based on extremum-seeking control. The optimization is driven by the gradient obtained by means of an adaptive u-PLS model. Since this is an empirical model, no first-principles based knowledge of the process is necessary. Heuristic rules are used to constrain the gradient taking into account nonlinearity and unknown causes of variability. Extensions to model the variability in initial conditions, to optimize several performance indices, and to handle inequality constraints are presented. The optimization algorithm is tested on a complex comprehensive simulated-process model for the fed-batch cultivation of Sacharomyces cerevisiae. Results show the performance and versatility of the proposed approach, as well as its robustness to process changes.

Multivariate batch to batch optimisation of fermentation processes incorporating validity constraints

“…In [21,30,31], it was suggested that the identified MVT be passed through a low-pass filter before being applied to the process. In this work, filtering the identified MVT proved to be beneficial and hence in each case study the MVT was passed through a zero-phase, low pass Finite Impulse…”

“…Biomass concentration is principally effected by manipulating the trajectory of glucose that is fed to the reactor. The simulation parameters that were used in this work, and listed in Tables 1 and 2, matched those used in [21]. Less than 9(l) Hard constraints in the MVT (l/h)…”

“…The hybrid model was a combination of a simplified first principle model and an MPLS model; the latter being used to correct errors in the mechanistic model. An alternative approach, proposed by Camacho [21], involved using an evolutionary optimisation technique that identified MVTs that would improve product yield in the forthcoming batch. The MVTs were determined by consideration of the gradient identified from an MPLS model that described the dynamic characteristics of the batch.…”

“…An important advantage of the MPLS based optimisation approach proposed in [21] is its low computational complexity. However, a limitation with the approach is that the technique does not consider that the model is only valid around the operating conditions used to calibrate the model.…”

Interval and Possibilistic Methods for Constraint-Based Metabolic Models

“…Model-based optimisation provides an alternative to the trial-and-error methods that prevail in industry, and leads to better performance within shorter development time intervals. Optimisation strategies for bioprocesses can be classified in categories: one-time optimisation (Banga, 2003;2008b), run-to-run optimisation (Camacho, 2007) and on-line optimisation (Visser, 2000).…”