Abstract :
Steady-state optimization of an industrial process often considers that the overall control is performed within a two-layer hierarchical structure. The lower layer performs direct regulatory control, where the aim is to maintain selected process variables at their desired set-point values, and the upper layer, known as the supervisory layer, has the task of determining the set-points of the regulatory controllers to obtain optimal steady-state performance. If the industrial process is `large scale´ in the sense that it consists of interconnected subprocesses, each subprocess will have its own regulatory control system and require its own supervisory decision function. Furthermore, coordination is usually required to ensure that interconnections between the subprocesses are taken into account and that any overall system objectives and constraints are satisfied. The resulting hierarchical structure is often on-line in which real subprocess measurements may be employed within local decision units and overall process measurements can be employed in the coordinator
