Uncertainty model identification for H∞ robust control

In set membership identification for H_∞ robust control, it is of relevance to evaluate tight frequency bounds on the modeling error magnitude. In this paper, a procedure is proposed assuming that time or frequency domain measurements data are available, affected by L_∞ bounded noise. The algorithm computes monotonically convergent upper and lower frequency bounds on the modeling error, and can be stopped when a given accuracy is achieved. Significant improvements in computational speed are obtained over existing methods in the literature for guaranteeing the same degree of approximation of the result, as documented by the presented numerical examples