Subspace model identification for continuous-time systems

Usually, the subspace-based state-space system identification algorithms are focused on discrete-time models, which may cause some numerical problems when the sampling interval is small. This paper proposes an algorithm of subspace-based state-space system identification for continuous-time systems from sampled input-output data. The w-operator w (p - alpha)/(p + alpha) where p denotes a differential operator and alpha > 0, is introduced to avoid direct numerical differentiations. And the w-operator state-space model identified by the 4SID method can be transformed back to the common continuous-time state-space model. The numerical superiority of the w-operator approach compared to some other methods is verified through simulation study.