Data-driven dynamic controller design for reference tracking on industrial applications

Abstract

This paper presents fundamental theories and statements about data-driven control, data-driven dynamic control and linear-quadratic regulator (LQR) applied on real systems. With a review of literature of data-driven control analysis, this paper also gives implementations and proposed designs of control approaches of different types of plants. In order to validate the results, metrics (IAE, MAE, and TV), robustness against step and pulse disturbances, and noise sensitivity are discussed. The results show that, for linear plants, the data-driven dynamic controller provides effective reference tracking and strong disturbance rejection, with response characteristics similar to classical LQR-based designs. For the nonlinear system, performance is more limited, suggesting the need for further investigation.