The Effect of Modeling Errors on Linear State Reconstructors and Regulators

Abstract

The effect of imprecise knowledge of system parameters on the reconstruction error of linear observers and on the stability of a class of linear regulators is examined. An upper bound on the reconstruction error for linear unforced systems is obtained. It is shown that in the regulator problem consideration of parameter uncertainty leads to the inclusion of step disturbances. An upper bound on allowed parameter variations that guarantees stability of a class of closed-loop regulators is obtained. It is also shown that by properly choosing parameters of a low-order observer, the output of this low-order model can be made to follow closely the output of a higher-order dynamic systems. The effect of modeling errors on randomly perturbed systems is also examined.