Structured experimental modeling of complex nonlinear systems

Abstract

In the paper an iterative algorithm is proposed for the identification of a system composed of two MIMO systems, one linear and the other one nonlinear, interconnected by an unknown multivariable signal. The considered interconnection structure can represent Hammerstein, Wiener or Lur'e models, as well as more complex structures. A key feature of the proposed method is that the nonlinear subsystem may be dynamic and is not supposed to have a given parametric form. In this way the complexity/accuracy problems posed by the proper choice of the suitable parameterization of the nonlinear subsystem are circumvented. Moreover, the cost function used to evaluate identification errors is guaranteed to decrease for increasing number of iterations. The effectiveness of the algorithm is tested on the problem of identifying a simulated half-car model for vertical dynamics of vehicles with controlled suspensions. Assuming an experimental setup easily realizable in actual experiment on real cars, the half-car model is decomposed as a generalized Lur'e system, consisting of a linear MIMO system, connected in a feedback form with a MIMO nonlinear dynamic system through not measured signals.