A modified iterative set-point optimization strategy with application to batch chromatography

This paper is concerned with on-line set-point optimization in the presence of model-plant mismatch. An iterative gradient-modification optimization strategy which was derived from the integrated system optimization and parameter estimation technique (ISOPE) is introduced. The method requires the computation of the gradient of the plant mapping in each iteration. A new approach to the gradient computation which takes the influence of measurement errors and the number of additional set-point perturbations into account is proposed. The handling of process-dependent constraints in the optimization is also considered. The potential of the iterative optimization strategy is illustrated by an application to batch chromatography.