A comparison of identification-based performance bounds for robust process control

This paper compares three different uncertainty estimation techniques in terms of predicting a priori performance bounds on the closed-loop system. A major benefit of computing the performance bounds is they expedite the controller implementation process, which is a critical component in the acceptance of the control system by the process operators. A key ingredient in computing these bounds is uncertainty estimation. The three methods evaluated are the coprime factor uncertainty method, Zhu's asymptotic method (1989), and the frequency-domain method of Bayard and Yam (1994). We compare the performance bounds obtained by applying these uncertainty estimation procedures on a paper machine control case study.