Resumen:
In this paper, a new data-driven multiple-inputs multiple-outputs method based on the Virtual Reference Feedback Tuning is presented. The methodology decides the pairing of the input-output signals to control at the same time that the parameters of the controllers are computed, by solving a non-linear optimization. The method is applied to a solid oxide fuel cell to check the effectiveness of the controllers. Compared with other similar structures, the proposed method yielded better results for voltage and fuel utilization, without using any model of the plant.
