MINIMAX L_2-E_2 FIR FILTERS FOR DETERMINISTIC CONTINUOUS-TIME STATE SPACE SIGNAL MODELS

摘要

In this paper, a new minimax L_2-E_2 performance criterion is introduced, which is represented as a worst case gain between the disturbance during the recent time interval and the current estimation error. Minimax L_2-E_2 FIR filter (MLEFF) is proposed for deterministic continuous-time state space signal models. The MLEFF is designed to minimize the maximum value of the L_2-E_2 performance criterion together with linearity, unbiased property in the deterministic sense, FIR structure, and independence of the initial state information, simultaneously. The MLEFF is shown to be robust with respect to model uncertainties due to FIR structure and disturbances due to the worst case design. For efficient implementation, it is shown that the MLEFF can be expressed recursively. It is also shown to be optimal for the performance criterion for each component of the state.