Automatic Generation of Feedforward Controllers using Dynamic Local Model Networks

Abstract

A new approach for the automatic generation of a dynamic feedforward control law for nonlinear dynamic systems represented by discrete-time local model networks (LMN) is proposed. The generic model structure of LMNs offers the opportunity to apply such a general and automated approach for model inversion, even when the overall model complexity may be high. LMNs can represent nonlinear dynamic systems of almost arbitrary complexity. Their generic structure allows the generation of a feedback linearizing input transformation in a highly automated way. This paper proposes and discusses such an approach for the important class of LMNs with minimum-phase property. As a subclass of the class of minimum-phase LMNs, only those without numerator dynamics are considered in this manuscript. By using the input transformation, which results from feedback linearization, the feedforward control law is obtained. It can then be applied online for any reference trajectory without pre-planning. Thus, by representing a nonlinear dynamic system by the generic structure of an LMN and applying the proposed feedforward control law generation, a dynamic feedforward control for such a nonlinear system can be found automatically. Finally, the effectiveness of the method is shown on results for a Wiener model.