System identification using interval dynamic models

Motivated by the recent explosive development of results in the area of parametric robust control, this paper presents a new technique to identify a family of uncertain systems. The new technique takes the frequency domain input and output data obtained from experimental test signals and produces an "interval transfer function" that contains the complete frequency domain behavior with respect to the test signals. This interval transfer function is one of the key concepts in the parametric robust control approach and identification with such an interval model allows one to predict the worst case performance and stability margins using recent results on interval systems. The algorithm is illustrated by applying it to an 18 bay mini-mast truss structure.