An intelligent model predictive control-based strategy for plantwide optimization of degrading plants

The model predictive control methodology is very powerful for the design of hierarchical multilayer control systems, i.e. systems which can be used either to control plants characterized by significantly different dynamics or to use different models of the same plant with the aim to optimize a number of criteria. Here, a two-level structure for control and optimization of degradation type of plants is proposed, which uses plant models in the “fast” and in the “slow” time scales and incorporates case-based reasoning. In particular, the plantwide optimization of a Peirce-Smith converter is considered, which is a typical example of a plant with degrading performance. The number of blocked tuyeres is optimized in order to achieve an efficient long-term operation.