Improving the H-infinity norm estimate of an active vibration isolation system using local rational models

Accurate uncertainty modeling is crucial in robust vibration controller design. A novel uncertainty modeling technique that enhances the estimated H∞ norm of the model error based on the frequency response is developed in this paper. The major innovation is the use of the local rational method (LRM), a local parametric modeling method, to address the inter-grid error. Unlike classical frequency response function (FRF)-based methods that typically underestimate the H∞ norm, the proposed technique allows to capture sharp resonance phenomena without the need for a user-intensive and time-consuming model selection step. The performance of the method is illustrated on a simulation example and on measurements of an industrial active vibration isolation system (AVIS). It is shown that in a short measurement time, a reliable estimate of theH∞ gain can be attained. On the measurement data, a 7.5 dB improvement of theH∞ norm is observed and validated.