Data-driven update of the free parameter of the Youla-Kucera parametrization in disturbance attenuation FRIT based on variance evaluation

This paper proposes a data-driven controller tuning technique for regulatory control. The minimum variance controller is designed by using disturbance attenuation FRIT (Fictitious Reference Iterative Tuning). The controller parameters are tuned based on variance evaluation using closed-loop input-output data. In the proposed method, the controller is parametrized by the free parameter in the Youla-Kucera parametrization of the nominal model. The present work gives a way of a data-driven update for the free parameter that improves the variance evaluation, and shows that the proposed data-driven performance criterion has the same optimal free parameters as the corresponding model-based performance criterion when the disturbance model is precisely known. Finally, the efficiency of the proposed method is shown through a numerical example.