Robustness of controllers designed using Galerkin type approximations

It is shown that Galerkin methods can be used to design controllers which will perform as designed when implemented on the original infinite-dimensional system. No assumptions, other than those typically employed in numerical analysis, are made on the approximating scheme. A high-order Galerkin model is chosen by either comparison of coprime factors to those of the infinite-dimensional model (if available) or determination that the sequence of approximations has converged satisfactorily.<
>