FRANCOIS Gregory, SRINIVASAN Bala, BONVIN Dominique
A typical dynamic optimization problem has two types of constraints: i) the path constraints, which includes bounds on the manipulated variables such as temperature, pressure, and safety constraints (heat removal capacity) and ii) terminal constraints, such as product quality or quantity at the end of the batch. Also, there are two types of available measurements: i) on-line measurements during the batch and ii) off-line measurements at the end of the batch. Among the various measurement-based optimization techniques available in the literature, the one that guarantees optimality via tracking the necessary conditions is studied. An earlier study allowed the tracking of path constraints using on-line measurements. The current project is devoted to tracking terminal constraints using off-line measurements. For this purpose, the optimal input is parameterized into a finite set of parameters that is adapted using the off-line measurements in a run-to-run manner. Issues concerning convergence and optimality are analyzed.
Francois G., B. Srinivasan and D. Bonvin. Run-to-run Optimization of Batch Emulsion Polymerization. IFAC World Congress 2002, Barcelona, Spain (July 2002), 1258-1263
Srinivasan B., C.J Primus, D. Bonvin and N.L. Ricker. Run-to-run Optimization via Control of Generalized Constraints. Control Eng. Practice, 9, 911-919
Srinivasan B., C.J Primus, D. Bonvin and N.L. Ricker. Run-to-run Optimization via Constraint Control. ADCHEM 2000, Pisa, Italy (June 2000), 797-802