Uncertainty model unfalsification: a system identification paradigm compatible with robust control design

It is shown that unfalsification of the standard robust control design uncertainty model is a natural replacement for system identification when the intended use of the model is robust control design. For the ARX model, the unfalsification step requires solving a set of convex programming problems, specifically LMI problems, of which ordinary least-squares is one member. The result is a tradeoff curve between model uncertainty and disturbance uncertainty. Hence, a family of models are unfalsified from the data record. The tradeoff curve is given a frequency domain interpretation via, the DFT and related computational issues are discussed.