Adaptive calibration of feedforward controllers for nonminimum-phase tracking systems

Nonminimum-phase zeros in the discrete transfer function of a plant present design challenges for stable feedforward tracking controllers. Previous pole/zero cancellation approaches have employed a variety of strategies for dealing with “uncancellable” zeros. An alternative approach, especially suitable for repetitive tracking tasks, is proposed: the feedforward controller inherits just the structure of the inverse plant model. The controller parameters are obtained by an adaptive optimization scheme, using plant input-output data from a few test runs of the repetitive task, and the stability of the controller is guaranteed. Simulation studies demonstrate superior performance when compared with the well-known “zero phase error” tracking controller