Trajectory learning and output feedback control of Nonlinear discrete-time systems

Output tracking control of a general class of nonlinear discrete-time systems is considered. A block input-output realization of the plant is first introduced that transforms the system into one with equal number of inputs and states. This realization is subsequently used to formulate a novel output feedback controller for regulation or tracking. The proposed controller only requires measurement of the system's outputs, and is applicable to both minimum and nonminimum phase plants of arbitrary relative degree. For applications involving a finite-duration or periodic desired outputs, a new recursive algorithm based on the Newton-Rapson method is proposed to learn the feedforward input. A simulation example of a nonlinear non-minimum phase system is presented to further illustrate and evaluate the performance of the proposed control scheme.