To construct a model set is inevitable for designing a controller based on robust control theory. In this paper, we will propose one method of model set identification. The model set treated in this paper is a discrete-time state space model set with time-varying parametric uncertainties measured by some kinds of norms. The proposed method is a slight modification of the subspace method, which is a powerful tool for finding a nominal model in a state space description from I/O data. The minimal model set is obtained via a convex optimization such that it can produce the experimental I/O data from a real system to be identified.
展开▼
