Abstract
Four relatively recent techniques for model reduction (balanced, weighted balanced, optimal Hänkel and weighted optimal Hänkel) are reviewed. The usefulness of these techniques for obtaining reduced-order models of power systems for dynamic control purposes, their effectiveness, and the merits of the simplified models are illustrated on a single-machine infinite-bus system. The methods are evaluated on the basis of L
α errors, transient responses, and steady-state errors. The superiority of the frequency-weighting methods is clearly demonstrated.