Identification of Nonlinear Systems

It is extremely difficult to identify general nonlinear systems because of the number of unknowns involved. Moreover, since most design techniques assume a linear model, many of the nonlinearities that can be determined are essentially ignored. Why not develop nonlinear system identification techniques to reveal restricted classes of nonlinear systems that are mathematically tractable and design accessible? Some interesting calsses are restricted by differential geometric conditions on a state space representation. The purpose of this paper is to develop methods to recognize related conditions from an input-output point of view.