Minimizing the effects of unknown-but-bounded disturbance in power plants

Many power plant control problems demand hard constraints on states or their functions because of physical limitations related to safety. Although results relating to "energy" bounds and probabilistic bounds are plentiful, corresponding results on hard constraints are scattered and have not been adequately treated in the fiterature. A group of work specifically dealing with hard constraints is the research on the unknown-but-bounded description of uncertainty for linear systems, pioneered by Schweppe. In this paper we present hard constraint bounding results for the states of systems disturbed either by uncertain inputs or modeling errors, using either ellipsoids or boxes as the constraint sets. The norm bound results are based on the concept of matrix measures. Bounds for large scale systems and nonlinear systems are also presented. Finally, some of the theoretical results are applied to a model of a physically meaningful system, namely a boiler model with encouraging results.