Model reduction using minimal realisation algorithms

Existing algorithms for the minimal realisation of linear systems are used to approximate high-order systems by lower-order models. The reduced models, so derived, are equivalent to those derived by the continued-fraction methods and the time-moments technique.