The objective of this research is the design and implementation of new model predictive control strategies for the optimal operation and control of processes. The prediciton models can be linear, non linear or heuristic, depending on the specific control strategy applied and the available process data. Due to the complexity of the process considered, particular aspects related to their identification and simulation are also studied. More precisely, new distributed and hierarchical model predictive control techniques are being developed, together with other economic control strategies in order to consider costs and environmental issues. Fuzzy logic is also considered to introduce expert knowledge of the processes in some of the mentioned strategies, particularly for negotiation among local controllers in the distributed frameworks. Also for negotiation, cooperative and other adaptive games are included in order to reach a sucessful global objective with stability and feasibility guarantees.
The developed methodologies are intended to be applied in process industries, particularly in Integrated Water Systems, that include a sewer system, the wastewater treatment plant and the effects on the receiving basin. To this aim, some simulation benchmarks and laboratory plants are used to validate the proposed methodologies. For practical implementation, a multiagent platform is considered to facilitate the communication between local controllers and the inclusion of other techniques from the artifical intelligence field.