ROBUST IDENTIFICATION IN H-INFINITY

We consider system identification in the Banach space H∞ in the framework proposed by Helmicki, Jacobson, and Nett. It is shown that there is no robustly convergent linear algorithm for identifying exponentially stable systems in the presence of noise which is not tuned to prior information about the unknown system or noise. Various nonlinear algorithms, some closely related to one of Gu and Khargonekar, are analysed, and results on trigonometric interpolation used to provide new error bounds. An application of these techniques to approximation is given, and finally some numerical results are provided for illustration. © 1992.