Iterative Feedback Tuning For MIMO Systems

In this contribution, we extend the Iterative Feedback Tuning (IFT) control method of [3, 4] to a multivariable setting. We report on an experimental study with this method when applied to a Single-Input-Multiple-Output torsional laboratory setup. Keywords: Optimal control, estimation, laboratory application 1 Introduction A torsional control system available in the Department of Sytems Engineering at the Australian National University has been considered as a test case for an extension of a recently developed model-free, optimization and experiment based control design scheme; we refer the reader to [3, 4] for further information. This method, acronymed Iterative Feedback Tuning (IFT) by its inventors is based on an iterative tuning of the controller parameter vector along the gradient direction of a control performance criterion. The key contribution of [3, 4] was to show that an unbiased estimate of this gradient can be constructed from signals measured on the closed loop system...