一回の実験データを用いたセミクローズド制御方式に対する制御器パラメータチューニング

抄録

In this paper, we propose a new data-driven controller parameter tuning method for the position control of mechanical systems by the semi-closed loop control. The proposed method is based on the Fictitious Reference Iterative Tuning (FRIT) approach, which enables us to obtain an appropriate controller parameter that improve the control performance in the load position using only one-shot experimental data. The feature of the proposed method is to use not only the output data of motor but also the output data of load to construct the cost function in FRIT. As a result, we can obtain an appropriate controller parameter that match the load position with the desired response. Finally, we provide some experimental results to show the effectiveness of the proposed method.