Reformulation of the parameter identification problem for systems with bounded disturbances (Q1091321)

The authors consider the parametric identification of linear discrete- time systems with known bounded disturbances. Projection and least- squares algorithms are modified to cope up with bounded disturbances. Both of these algorithms have been obtained by optimizing performance indexes. Through convergence analysis it is shown that the estimated parameters remain bounded, the estimated increments converge to zero and that the prediction error, passed through a dead zone and normalized, converges to zero.