Active identification of linear static process parameters (A game problem in experimental design)

The article is concerned with design of algorithms for active multiple identification of the parameter vector for a linear static process. This vector is assumed to stay in a convex polyhedron, specified in advance, and the perturbations are assumed constrained. The properties of optimal active identification are studied which make it possible to reduce the diameter of an a posteriori range of the parameter vector. A recurrent suboptimal active identification algorithm is proposed. Its convergence is proved and the rates of convergence are estimated. The results of digital modeling are reported.