Model free subspace based ℋ∞ control

Plant knowledge is essential to the development of feedback control laws for dynamic systems. In cases where the plant is difficult or expensive to model from first principles, experimental data are often used to obtain plant knowledge. There are two major approaches for incorporating experimental data in control design: model identification/model based control design and model free (direct) control design. This work addresses the direct control design problem. In particular, this paper exploits subspace prediction methods to develop a novel ℋ_∞ direct control design technique. A single integrated algorithm computes the central ℋ_-∞-optimal controller and provides an estimate of the closed loop performance. The ℋ_∞ design technique is experimentally demonstrated on a flexible mechanical system. The model free technique performs well when compared against an ideal simulation. The technique dramatically outperforms a model based ℋ_∞ controller