Terminal optimization constraints play an important role in dynamic optimization problems. Among the various optimization approaches, the one that guarantees optimality via tracking the necessary conditions is studied here. The terminal constraints can be tracked, using measurements obtained either at the end of the batch or during the batch. The latter approach yields faster convergence, but it necessitates predicting the future values of the terminal constraints. A possible solution is to use a model to compute the future values of the terminal constraints, while adapting the model on-line. However, an exact model of industrial processes is hard to find. Additionally, to take into account the flexibility of batch processes, the model would have to be changed frequently. Instead of adapting the model for recomputing the optimal trajectory on-line, this work adapts the plant input to track a predefined trajectory that leads the process to the terminal constraints and thus guarantees optimality.
Keywords : ;
Welz C., B. Srinivasan and D. Bonvin. Evaluation of Measurement-based Optimization Schemes for Batch Distillation. IFAC World Congress 2002, Barcelona, Spain (July 2002), 1586