An Iterative Method to Stabilise a Transfer function in the s - and z -Domains

Stability of the identified model is crucial in simulation and prediction applications. This paper proposes a two step procedure that generates guaranteed stable transfer function models from noisy data. It consists of an unconstrained optimal noise removal step followed by a stable approximation of the possible unstable estimate. The proposed method is applicable to continuous-time as well as discrete-time and multi input, multivariable systems