Unbiased identification of systems with nonparametric uncertainty

The set-membership identification of systems with parametric and nonparametric uncertainty is studied. The bias induced by the additive noise is eliminated by using the bias-eliminated least squares method proposed previously by us (1991). A prefilter is connected to the input terminal of the system, so that some zeros are inserted to the system. By using the information obtained from these known zeros, the bias induced by the additive noise is eliminated.<
>