System identification and uncertainty domain determination: a subspace-based approach

In this paper, the problem of uncertainty domain determination for multi-input multi-output systems described with linear time-invariant state-space representation is tackled. The developed method combines a subspace-based identification algorithm and a bounded-error approach. To estimate the system state-space matrices, the propagator method is applied. This method is characterized by the use of least-squares minimization algorithms and the fact that the estimated state-space realization is described using a fixed structure with a minimal number of parameters. The problem of uncertainty domain description is then solved from the analysis of particular iso-level curves in the vicinity of the optimum. A hard error bounding approach is finally considered to choose a threshold required by the developed method.