An ℋ︁_∞algorithm for the windsurfer approach to adaptive robust control

Abstract

The windsurfing approach to iterative control requires a series of controller designs with the gradual expanding of the closed-loop bandwidth, and in the end in order to stop the algorithm, some validation tests are carried out. In this paper, an H 1 design algorithm is introduced in order to remove the empirical aspect from the stopping criteria and to make the procedure more systematic, hence facilitating the design. Moreover, some restrictive assumptions on the plant model are lifted and some issues with the controller design step are tackled by the introduction of a new design method. This enables us to address a wider class of practical problems.

Iterative identification and control design with optimal excitation signals based on υ-gap

“…(Please refer to ref. [19] for the details of the proof) Theorem 3. For the i th identified nominal model G i and controller C * i designed from G i , if there exists a controller C i+1 such that…”

Plug-and-Play Control—Modifying Control Systems Online

“…Note also that is deliberately chosen such that the states corresponding to are unobservable. Next, in row c), the new device is added, which causes to be replaced by the system (16) where , , , and represent the output map of the new sensor. Note that these parameters are, in general, not known a priori and may thus require identification.…”

“…Conceptually, Plug-and-Play control is somewhat related to "Windsurfer control" [14]- [16]. However, while Windsurfer control also aims for performance improvement by learning more about the plant during online operation, there have so far been no treatments of structural updates of the closed loop.…”

Verifying stabilizing controllers for performance improvement using closed‐loop data

“…The Windsurfer approach is a particular iterative identification and control design procedure, which was successfully used in [2,4,28], and [29] to design controllers to push the closed-loop bandwidth (BW) of a flexible robot arm as large as possible. The ultimate origins of the model are not clear from these references, although the zero and pole structure is quite typical for a flexible robot arm.…”