Robust Experiment Design for Virtual Reference Feedback Tuning

This paper deals with robust experiment design for the Virtual Reference Feedback Tuning (VRFT) approach, a non-iterative control design method aimed to tune fixed-order controllers directly from experimental data, without the need for a model of the plant. In a previous contribution, it has been shown that the spectrum of the optimal input depends on the frequency response of the controller achieving the desired performance. In this work, a robust input design procedure is proposed, which requires only mild prior knowledge about the optimal controller. The solution is obtained analytically via constrained min-max optimization. Simulation results on a benchmark case study for digital control systems show the effectiveness of the proposed approach.