Uncertainty Estimation Based on Multiple Models with Application to Distillation (original) (raw)
Modern robust control synthesis aims at providing robustness with respect to norm bounded uncertainty. Standard identification methods are based on the assumption that all uncertainty is in the form of noise. Recent work has tried to address this difference between the models used in control synthesis and those obtained from identification experiments. The proposed methods rely on a priori assumptions on the noise and/or uncertainty. It is the aim of this paper to present a method where a separation of noise and uncertainty is achieved in the identification step. The uncertainty is estimated from a set of plant models, obtained by standard identification techniques, and the input data from the identification experiments. A special feature of the method is that the obtained uncertainty model is output multiplicative. The methodology is applied on experimental distillation column data, and a ยต-optimal controller is designed. The controller is tested experimentally on the distillation ...
Loading Preview
Sorry, preview is currently unavailable. You can download the paper by clicking the button above.