Direct data-driven control design through set-membership errors-in-variables identification techniques

In this paper, we propose a non-iterative direct data-driven control approach, such that the controller is directly identified from input/output data without plant identification step. First we formulate the problem of designing a controller in order to match the behavior of an assigned reference model in terms of an equivalent set-membership errors-in-variables problem and we define the feasible controller parameter set. Then, we design the controller parameters by applying previous results by the authors in the field of convex relaxation for errors-in-variables identification. Finally, the effectiveness of the presented technique is shown by means of two simulated examples.