Data-Driven Control by using Data-Driven Prediction and LASSO for FIR Typed Inverse Controller

In this paper, we propose data-driven control for the case of that the structure of the ideal controller is unknown. In this case, it is natural to implement FIR typed controller. Here, we utilize FRIT for the case of FIR typed inverse controller. In general, FIR might have unnecessarily large number of its taps. To prevent it, we also utlize LASSO regression so as to realize FIR typed inverse controller in as less numbers of the tap as possible. Moreover, we also utilize data-driven prediction to determine hyper parameters required in LASSO regression. The validity of our proposed method is verified by using experiments. © 2023 Institute of Electrical Engineers of Japan. All rights reserved.