Data-driven IQC-Based Uncertainty Modelling for Robust Control Design

A new approach is presented in the paper for modelling uncertainty as an elliptical set for robust controller synthesis. The method involves finding the best linear nominal model and the corresponding elliptical uncertainty set that is consistent with a set of frequency response functions of linear time-invariant (LTI) single-input single-output (SISO) systems. The uncertainty set is then converted into an equivalent integral quadratic constraint (IQC) using a novel split representation of uncertainty. Finally, the IQC is integrated into a data-driven frequency-domain controller synthesis method through convex optimization. The simulation and experimental results demonstrate that the proposed method yields a “tighter” uncertainty set and improved stability margins compared to classical methods that use disk uncertainty.