Conclusion
Identification of state-space models

In this article, we have presented subspace identification methods, methods dedicated to model estimation in the form of state representation. Different algorithmic variants are discussed, including discrete-time or continuous-time model estimation, open-loop or closed-loop estimation, off-line or on-line estimation.

These subspace methods have now reached a certain level of maturity, as evidenced by their use in standard numerical calculation software and the many successful applications to industrial processes. Although there is still room for improvement, these methods are an important tool for system modeling.