Closed-Loop Identification of Flexible Structures: An Experimental Example

Abstract

The application of a closed-loop identi cation procedure to a exible structure, the Jet Propulsion Laboratory Control Structure Interaction Phase B testbed, is described. The approach is based on an indirect, closed-loop identi cation procedure, recently developed by Van den Hof and Schrama (Van den Hof, P. M., and Schrama, R. gives a consistent model estimate in the case where the system noise/output disturbance model is not accurate. The procedure is modi ed and applied, in the frequency domain, to closed-loop data from a exible structure control experiment. The results of the analysis are compared to a purely open-loop frequencydomain identi cation approach. Enhancements to the experimental con guration are also discussed.

Aliasing of Resonance Phenomena in Sampled-Data Feedback Control Design: Hazards, Modeling, and a Solution

“…One possible explanation for the mismatch in magnitude is the fact that the true system is nonlinear and the experiment was performed with a different excitation signal, i.e., the frequency content of a ZOH reconstructed signal depends upon the sampling frequency, see (5). Similar results have been obtained in a comparison between open-loop and closed-loop experimentation, see [13]. Additionally, the sampler S and reconstructor H may not be ideal as in Definitions 2 and 3.…”

Connecting System Identification and Robust Control for Next-Generation Motion Control of a Wafer Stage

“…7 for the corresponding frequencies. Similar results regarding the identification of flexible dynamical behavior have been reported in [11]. Next, the static model uncertainty bound γ in (8) can be obtained from…”

“…This is especially due to the fact that these motion systems are designed such that the system dynamics are essentially linear, enabling the use of well-developed system identification techniques for linear systems. The resulting linear model is an approximation of the true system, since: 1) motion systems generally contain many resonance modes [10] of which a limited number is included in the model; 2) parasitic nonlinearities are present, e.g., nonlinear damping [11]; 3) identification experiments are based on finite time disturbed observations of the true system. Robust control design [9], [12], explicitly addresses these model errors by considering a model set that encompasses the true system behavior.…”

“…9 The identified model set P dyn is used to synthesize a robust controller using (31). For comparison, also a nominal controller C NP , see (11), is synthesized usingP and standard H ∞ -optimization [12]. The resulting controllers are depicted in Fig.…”

Next-generation wafer stage motion control: Connecting system identification and robust control

“…Indeed, robust control is essential for motion systems, since a nominal model cannot encompass the entire system behavior due to the presence of high order flexible dynamical behavior [6] and nonlinear damping effects [7]. In [4], a motion control design procedure is presented for SISO motion system that combines system identification and robust control.…”