A Nonlinear Grey-box Example Using a Stepwise System Identification Approach

A stepwise algorithm where the user starts with a preliminary model and adds new, nonlinear, parts to it is used to identify a nonlinear model of a rotational system. The user is assumed to roughly know where in the system there are nonlinearities and adds small black-box parts to the initial model to capture the nonlinearities. The stepwise algorithm gives well motivated parameter initialization of the nonlinear model which increase the chances to obtain a good model in the minimization of the criterion of fit.