Virtual Time-response Based Iterative Gain Evaluation and Redesign

Abstract

Fictitious Reference Iterative Tuning (FRIT) have been reported for many applications, in which the control gain can be optimized so that the closed loop system corresponds with the reference model based on only one experiment data. However, when the reference model is not appropriate, there is a problem that not only the control performance is deteriorated but also the control system may become unstable. Instead of using model matching to the closed-loop reference model, we thought that this problem could be solved by optimizing the settling time and overshoot. To measure settling time and overshoot, the time response is required. The time response can be predicted using Kaneko's method [Takahashi, 2019], but this method requires the order of the closed loop transfer function. In this paper, the method [Takahashi, 2019] is reconsidered as a frequency response based method so that the order is not required. Furthermore, as another approach of FRIT, this paper proposes a method called ‘Virtual Time-response based Iterative Gain Evaluation and Redesign’ (V-Tiger) which iterates to measure the overshoot and settling time from the virtual time responses, and to evaluate and redesign the controller gain.