Static output-feedback control design from data

Abstract

This paper investigates the problem of data-driven static output-feedback (SOF) control design for linear time-invariant discrete-time systems. Although data-driven control techniques have attracted significant attention recently, the design of data-driven SOF controllers is still an open problem. The proposed method relies only on input-state-output data from the open-loop system to represent the system dynamics and to construct the control design conditions using linear matrix inequalities. As an extra contribution, the design problem with noisy data is raised, and some solutions aiming to increase the robustness against noise, based on data preprocessing, are proposed. We also develop a SOF condition based only on input-output data from an autoregressive representation. To illustrate our method’s effectiveness, we consider two simulation examples of multiple input multiple output systems and an experimental example in the twin-rotor aerodynamic system. The experiments show that the data-driven SOF controller guarantees the stabilization of those systems. Moreover, experiments with noisy data illustrate the increased robustness of the proposed techniques.