Cost guaranteed robust sampled-data parallel model design using polynomial approach

The multivariable H2 guaranteed robust minimum variance parallel model design problem subjected to norm bounded uncertainties is studied in this paper for sampled-data systems. It consists of two paths connected in parallel with a common stationary stochastic input. One of them has an unknown system to be designed despite the presence of disturbances, so that the output signal of the two paths is of minimum variance. The systems and noise models are assumed to be represented by polynomial matrices that are not perfectly known except that they belong to a certain set. The sampled-data design is based on a fast sampling and lifting technique resulting on a finite-dimensional filter. An application case of robust parallel model design to the feedforward load-frequency control on hydro-generating units is provided.