In this paper, we consider general nonlinear systems with observations,containing a (single) unknown function φ. We study the possibility tolearn about this unknown function via the observations: if it is possible todetermine the [values of the] unknown function from any experiment [on the setof states visited during the experiment], and for any arbitrary inputfunction, on any time interval, we say that the system is “identifiable”. For systems without controls, we give a more or less complete picture of whathappens for this identifiability property. This picture is very similar tothe picture of the “observation theory” in [7]:Contrarily to the case of the observability property, in order to identify inpractice, there is in general no hope to do something better than using“approximate differentiators”, as show very elementary examples. However, apractical methodology is proposed in some cases. It shows very reasonable performances.
As an illustration of what may happen in controlled cases, we consider theequations of a biological reactor, [2,4], in which apopulation is fed by some substrate. The model heavily depends on a “growthfunction”, expressing the way the population grows in presence of thesubstrate. The problem is to identify this “growth function”. We giveseveral identifiability results, and identification methods, adapted to this problem.