Fictitious reference iterative tuning based on variance evaluation for disturbance attenuation in non-minimum phase plants

Disturbance attenuation can be realized through Fictitious Reference Iterative Tuning (FRIT) based on variance evaluation. However, this method proposed in the previous work cannot be applied to Non-Minimum Phase (NMP) plants, because unstable pole-zero cancellation results in internal instability. Thus, the present work proposes a disturbance attenuation FRIT for NMP plants. The underlying controller in the proposed approach is an H₂ optimal controller, which can avoid unstable pole-zero cancellation with the help of a generalized interactor. The parameterized possible minimum variance is introduced to the proposed data-driven cost function because the possible minimum variance by the H₂ optimal controller depends on the unknown plant parameters. The paper proves that optimizing the data-driven cost function obtains simultaneously both the optimum controller and the plant parameters. The validity of the proposed method is demonstrated through a numerical example.