Design of least costly identification experiments : The main philosophy accompanied by illustrative examples

The goal of this paper is on the one hand to give a tutorial on the main ideas of a recently introduced paradigm for optimal experiment design whose objective is to design the least costly identification experiment while guaranteeing a sufficiently accurate model for e.g. control. On the other hand, the second goal is to illustrate with well chosen examples the advantages of designing optimally the excitation signal for an identification instead of using a classical white excitation. As we will see in these examples, shaping appropriately the excitation signal allows one to reduce significantly the cost of an identification experiment.