Unfalsified Control is a data-driven, plant-model-free controller design method, which recursively falsifies controllers that fail to meet the specified performance requirement. In Ellipsoidal Unfalsified Control, the region of controllers that are unfalsified, the Unfalsified set, is described by an ellipsoid. Due to the combination of the performance requirement and controller structure, the update of the Unfalsified set can be computed analytically, resulting in a computationally cheap algorithm. Conditions for stability of Ellipsoidal Unfalsified Control are presented, and the effectiveness of the proposed algorithm is shown in a simulation.

Abstract-This paper presents recent research results for feedback control design of motion systems. Two model-free approaches are investigated, that exploit the ease of experimentation which is typical for motion systems. One approach is data-based design of a linear feedback controller which realizes desired closed-loop sensitivity and complementary sensitivity transfer functions. These transfer functions are specified via a data-based performance cost. The designer can prescribe both the controller structure and the complexity. Experimental results obtained in a direct-drive robot motion control problem confirm the effectiveness of the design. A second line of research is unfalsified control where a set of controllers is iteratively tested against measured data. Experimental results for the wellknown fourth order benchmark motion system show feasibility of the approach. Finally, we implemented a nonlinear SPAN filter on the same system, which outperforms a linear feedback design.

No abstract

No abstract