Data-Driven Control and Its Applications

Data-driven control design approaches are an interesting alternative to classical model-based control design methods. In these approaches, the controller parameters are directly computed by optimising a control performance criterion, which is a function of measured data. In this talk, a survey of some data-driven control approaches developed in Automatic Control Laboratory is given. A new convex- optimization-based control design methodology using frequency-domain data is detailed and its performance illustrated via several applications: (1) Improving the performance of Atomic Force Microscopes (AFMs); (2) Robust control of power converters in particle accelerators of CERN, (3) Distributed voltage and frequency control of electrical microgrids and (4) High precision position control of Coordinate Measuring Machines (CMMs).