Abstract:

The robust H_∞ filtering problem for continuoustime linear stochastic systems with time-vary state delay and parameter uncertainty is investigated. The aim is to design a linear filter such that the filtering error dynamics system is exponentially stable in mean square and assuring a prescribed H_∞ performance level. Applying the descriptor model transformation, a new Lyapunov-Krasovskii function is constructed. The product terms accurring in both Lyapunov matrices and system matrices are avoided by introducing some free weighting matrices.Thus it is not necessary for system matrices to do any constraint in the design of filters, which makes the design of filters have less conservativeness to a great extent. For systems without uncertainty and with uncertainty cases, sufficient conditions are respectively proposed to guarantee the existence of desired robust H_∞ filters in terms of linear matrix inequalities. Numerical examples demonstrate the proposed approach is effective.