Conclusion

Abstract

In the introductory chapter of his book “System Identification”, P. Eykhoff stressed the need for some limiting theorems on the maximum result obtainable in the solution of a given problem. In some sense, the study of the structural identifiability of state-space models belongs to that rationale. The point is to test, before attempting any parameter identification, whether the structure chosen for the model does not render the attempt void of any meaning. From this point of view, one may say that the most significant property in practice is not identifiability, but unidentifiability. It is not because a parameter 1s s.g.1. that one will actually be able to estimate it. On the other hand, if a parameter is unidentifiable, one knows one will not be able to estimate it, except by calling upon some additional a priori knowledge.