Pre‐Filter Design for Iterative Controller Parameter Tuning Using Data‐Driven Minimum Variance Regulatory Controllers

Abstract

Data‐driven minimum variance regulatory controllers directly tunes controller parameters using regulatory control data for improving disturbance attenuation properties. The approach is originally derived for one‐shot controller parameter tuning. That is followed by iterative methods by way of just repeating one‐shot tuning method. The present work considers such iterative controller parameter tuning method, and proposes a gradient‐based pre‐filter that makes the gradient vector of the data‐driven cost criterion have almost the same direction as one of the original model‐based cost criterion while the global minimizer of the cost criterion remains unchanged. The pre‐filter includes the feedback invariant polynomial in the denominator, which can be estimated from time‐series analysis of the closed‐output and the information of the input delay time. Finally, the effectiveness of the pre‐filter design is shown through a numerical example. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.