On the Modified maximum Principle in Estimation Problems for Uncertain Systems

Filippova, T.F. (1992). On the Modified maximum Principle in Estimation Problems for Uncertain Systems. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-92-032

Preview

Text WP-92-032.pdf Download (347kB) | Preview

Abstract

The present report is devoted to the problems of estimating the state of a linear dynamic system on the basis of on-line observation. It is assumed that the disturbances in the system inputs and in the current measurements are uncertain, a set-membership description of their values being only given in advance.

A considerable number of problems concerning systems of the above type are covered by the theory of control and observation under uncertainty conditions. The main problems of this paper deal with the description of certain informational domains that are consistent with the results of available measurements of the state space variables. Here we consider the case when the disturbances in the system dynamics and in the observation equation are subjected to instantaneous (or "geometric") constraints.

One approach to the problem based on an imbedding procedure of the primary problem into an auxiliary one of linear-quadratic estimation theory is given in the paper. The proposed procedure involves certain quadratic forms to bound the uncertainties in the modified problem. This method allows one to derive an appropriate maximum principle that is satisfied by system trajectories leading to boundary points of the informational domain.

Actions (login required)

View Item