Control oriented system identification for MIMO continuous and discrete-time systems

In this paper we describe a worst-case /spl Hscr//sup /spl infin// identification method for multi-input multi-output (MIMO) continuous or discrete-time systems. The method identifies systems from noisy samples of the frequency response. We identify a low-order model and give an upper bound on the error between identified model and true system. Hankel approximation identifies the order and pole positions while Chebycheff approximation improves the optimality of the identified model by tuning the numerator coefficients.