An Efficient Frequency Domain State-Space Identification Algorithm

In this paper we present a novel non-iterative algorithm for identifying linear time-invariant discrete time state-space models from frequency response data. We show that the algorithm recover the true system of order n if n+2 noise-free frequency response measurements are given at uniformly spaced frequencies. The algorithm is demonstrated to be related to the recent timedomain subspace identification algorithms formulated in the frequency domain. The algorithm is applied to real frequency data, originating from a flexible mechanical structure, with promising results. In a companion paper robustness and stochastic analysis is performed. Keywords: system identification, state-space methods, frequency response. 1 Introduction Identification of large scale multi-input multi-output (MIMO) systems of high orders is still considered a challenge. This type of systems are encountered in the mechanical engineering area of modal analysis and in the control of flexible structures. In most case..