Normalized iterative feedback tuning with PD-type modified feedback experiment

We present a new feedback experiment in the iterative feedback tuning (IFT) method. The original IFT method requires a special feedback experiment, which cannot be carried out during a normal operation. The proposed method replaces the previous special feedback experiment by a new experiment, where a test signal for the tuning can be generated during a normal operation. The test signal is synthesized by perturbing the reference signal with the error signal on a reduced scale. In this paper, we derive a formula for the new feedback experiment. This feedback experiment has importance in that the IFT method can be extended to the on-line tuning scheme if every experiment begins at the same states or if the experiment interval is sufficiently so large that the transient effects of the initial states can be neglected. The amendment in the assumption of the initial condition is out of the scope of this paper