A model-less algorithm for tracking control based on input-output data

This paper proposes an idea of data-driven predictive control for a linear discrete-time system, that is, a tracking control algorithm based on input-output data. Any traditional model of the plant, such as a transfer function or a state equation, is not employed. The plant dynamics is represented by a rank constraint in an array whose elements are input-output data. The control input for tracking an arbitrary reference signal is readily computed using linear dependence of rows in the array. By refreshing the data, the algorithm can adapt to the change of the plant dynamics.