7. Conclusion

A model is only useful if it has been validated and identified, i.e. if it behaves like the process it is supposed to represent. Otherwise, it cannot replace this physical process in any simulation, and cannot be used to predict the behavior of the physical object subjected to a known input, in the case of control.

To claim that the model's behavior is identical to that of the physical object is pretentious. At best, we can hope to achieve a similarity of behavior under specific, well-defined conditions - a similarity which, fortunately, is sometimes very good.

There are several pitfalls to avoid:

• not be placed within the framework of systems theory. Placing a mathematical representation on a physical phenomenon requires a definite effort of abstraction. Indeed, this is where we most often...