Output feedback predictive controller for a class of nonlinear systems

The aim of this paper is to design an output feedback predictive controller for a class of uniformly observable nonlinear systems. More specifically, our proposal is addressed to the admissible tracking problem for minimum phase nonlinear processes. The unavailable state variables of the system are recovered by using a high gain observer (HGO), before adopting a mild change of coordinate in order to transform the model in a special form, when the constrained state feedback law can be easily synthesized. By using a Taylor series expansion of the output tracking error, the computational computer effort which usually presents a challenging problem in predictive control framework, is strongly reduced. The main features of our approach is twofold, the one is to exploit the ability of predictive control law to handle input constraints in order to achieve certainty equivalence principle, and the other is to ensure an efficiently solution of nonlinear predictive control (NPC) which can arise from a convex optimization problem. In fact, thanks to the sophistical already known tools as quadratic programming (QP) or linear matrix inequality (LMI) formalism, we can ensure real time implementation of the proposed controller. An application to the sensorless control scheme for induction motors is provided to illustrate the effectiveness of our approach.