Asymptotically exact direct data-driven multivariable controller tuning

In this paper, a data-driven controller design method for multivariable systems is introduced and analyzed. The proposed technique is direct, as it is entirely based on experimental data and does not rely on a physical description of the system, and non-iterative, as it does not require controller adjustments based on additional experiments. Compared to the state-of-the-art non-iterative technique, i.e. MIMO VRFT, the proposed approach is asymptotically exact, in that it guarantees that the desired closed-loop dynamics is matched when the number of data tends to infinity. The performance of the proposed approach is illustrated and compared with MIMO VRFT on a benchmark simulation example.