Modelling and identification for highly nonlinear processes Original Research Article

This paper is devoted to modelling and identification of nonlinear dynamic systems. At first, we propose a new input–output representation to describe highly and/or large-scale nonlinear processes. The proposed mathematical model is nonlinear in parameters and is written as a product of several polynomials which may be selected in a sequential approach. In the second part of this note, a simple and recursive identification technique is detailed. It is shown that, under strong persistently exciting condition, global convergence of the parameters estimation algorithm is guaranteed. One of the main results of this contribution is that parameters to be estimated are considerably reduced in comparison with the general Kolmogorov–Gabor structure.