Peak Amplitude-Constrained Experiment Design for FRF Identification of MIMO Motion Systems

The accurate identification of Frequency Response Functions (FRF) models is an essential aspect in the design and control of high-precision motion systems. The aim of this paper is to improve FRF identification of multiple inputs multiple outputs motion systems by optimal experiments design under
$l_{\infty}$
-norm constraints. A two-step experiment design framework is established that exploits a fast smoothing-based algorithm to
$l_{\infty}$
signal norm minimization, in conjunction with full multivariable and directional power-constrained excitation design. Experimental results from an industrial active vibration isolation system show a significant improvement of FRF quality compared to traditional design approaches.