Optimal Inputs for Guaranteed System Identification

Author:	Gusev Mikhail, Institute Math. & Mech. Ural Branch of RAS, Russian Federation
Topic:	2.1 Control Design
Session:	Control and Estimation Under Set-Membership Uncertainty - II
Keywords:	Identification, Uncertain dynamic systems, Bounded disturbances, Inputs

Abstract

The problem of choice of optimal inputs for nonlinear control system parameters identification is studied. The uncertain items are assumed to be unknown but bounded with no statistical information whatever. The problem is treated in the framework of guaranteed (set membership) approach. An integral of information function is considered as a criterion of optimality. It is shown that considered problem may be reduced to optimal control problem with functional depending on the integral over parameters. The necessary conditions of optimality for this problem are given, and the results of numerical simulation are described.