Robust data-driven state-feedback synthesis from data corrupted by perturbations with bounded norms and rates-of-variation

Data-driven H∞-optimal controller synthesis is considered for unknown discrete-time linear time-invariant systems. Perturbations in the data, such as those coming from external disturbances and measurement noise, are assumed to be bounded in norm and rate-of-variation. A linear matrix inequality (LMI) based framework is presented in which these realistic bounds can be included in the synthesis. This reduces the set of systems consistent with the data and thus offers a reduction in conservatism, at the cost of increased computational complexity. The method is evaluated through simulations on a double pendulum system.