Qualitatively robust fuzzy controller for the 1992 ACC robust control benchmark

Fuzzy control provides a convenient domain to test whether the qualitative reasoning of intelligent methods can produce robust controllers. Our method, the qualitative robust control (QRC), uses a qualitative plant model that subsumes plant perturbation. Utilizing the qualitative model, the fuzzy compensator design proceeds in three steps: 1) creation of a stabilizing compensator; 2) augmentation of the initial design with rules to achieve set-point control; and 3) tuning of the linguistically interpretable compensator parameters. Evaluation of QRC is performed using the ACC 1992 robust control benchmark. Our design outperforms conventional compensators, achieving greater stability robustness, plant noise rejection and tracking performance with less total compensator effort. These results show that a qualitative design approach can produce fuzzy compensator with superior robustness.