CONTINUOUS-TIME FREQUENCY-DOMAIN SUBSPACE SYSTEM-IDENTIFICATION

In this paper we present a new subspace identification algorithm for t he identification of multi-input multi-output linear time-invariant co ntinuous-time systems from measured frequency response data. We show h ow the conditioning of the data-matrices in the algorithm can be impro ved by making use of recursions derived from the Forsythe polynomials. The asymptotic properties are analyzed and it is shown that, when the error distribution on the measurements is given, the algorithm can be made asymptotically unbiased through the introduction of a weighting matrix.