Nonlinearity characterization for nonlinear dynamic system identification using an expert approach

The identification of nonlinear dynamic systems can be rendered significantly more parsimonious if the nonlinearity present in the system is known. While there are many successful non-parametric nonlinear system identification methods, the resulting models do not describe the nonlinearity in physical terms and are difficult to obtain due to the large number of candidate terms that must be examined. In this paper an expert approach towards the characterization of nonlinearities in a dynamic system is presented. The methodology is based on simulations of dynamic systems with a variety of commonly occurring nonlinear functions. The responses of such systems to various types of excitation are analysed and rules are developed as to what nonlinearity is likely to be present in a system given the dynamic characteristics of measured responses.