Data-driven online unfalsified control by using analytic center

In this paper, a novel on-line data-driven control method based on unfalsified control is proposed for the control problem that has been treated in the ellipsoidal unfalsified control method. In the latter method, an ellipsoid that approximates the unfalsified set is obtained analytically to obtain an inner point of the ellipsoid at each sampling time. In our new method, an analytic center algorithm is used to find an inner point of the unfalsified set, and the constraints on the parameters can be treated directly. Some of inactive or redundant inequalities are removed from the constraints in order to decrease the computation time, and three methods for selecting a parameter of the control performance are examined with respect to the convergence speed and the existence of the unfalsified set.