Mean-Square Optimal, Full-Order Compensation of Structural Systems with Uncertain Parameters.

Abstract
The minimum information approach to active control of structural systems seeks inherently robust designs by use of mean square optimization conjoined with a stochastic system model which presumes as little as possible regarding a priori information on modal parameter statistics. This report extends earlier results for the regulator problem to the case of full-order dynamic compensation with nonsingular observation noise. Optimality conditions along with sufficient conditions for existence and uniqueness of solutions and for closed-loop stochastic stability are presented. Results concerning asymptotic properties for large uncertainty levels are also given. Numerical results for various simple examples indicate improved robustness properties over standard LQG designs and suggest the possibility that, under the minimum information stochastic approach, the burden of design computation may be reduced to that associated with the relatively well known or coherent modes. (Author)

This publication has 0 references indexed in Scilit: