ROBUST H-INFINITY FILTERING FOR CONTINUOUS-TIME VARYING UNCERTAIN SYSTEMS WITH DETERMINISTIC INPUT SIGNALS

Many dynamical systems involve not only process and measurement noise signals but also parameter uncertainty and known input signals. When L(2) or H infinity filters that were designed based on a ''nominal'' model of the system are applied, the presence of parameter uncertainty will not only affect the noise attenuation property of the filter but also introduce a bias proportional to the known input signal, and the latter may be very appreciable. In this paper, we introduce a finite-horizon robust H infinity filtering method that provides a guaranteed H infinity bound for the estimation error in the presence of both parameter uncertainty and a known input signal. This method is developed by using a game-theoretic approach, and the results generalize those obtained for cases without parameter uncertainty or without a known input signal. It is also demonstrated, via an example, that the proposed method provides significantly improved signal estimates.