Robust estimator design using mu synthesis

摘要: H/sub infinity / norm optimization and the structured singular value mu are used to design state estimators that robust noise plant modeling errors. An optimal estimator is derived as solution a minimax problem. In suboptimal case, an constraint removed, approaches steady-state Kalman filter. Robustness errors achieved by minimizing frequency-weighted /-norm The weighted norm, , upper bound for therefore provides sufficient condition performance. filter compared using example with uncertainty. has slightly worse performance than nominal model, but, unlike filter, maintains its over entire range of >