Iterative learning control with discrete?time nonlinear nonminimum phase models via stable inversion

Output reference tracking can be improved by iteratively learning from past data to inform the design of feedforward control inputs for subsequent attempts. This process is called iterative (ILC). article develops a method apply ILC systems with nonlinear discrete-time dynamical models unstable inverses (i.e., nonminimum phase models). class includes piezoactuators, electric power converters, and manipulators flexible links, which may found in nanopositioning stages, rolling mills, robotic arms, respectively. As these devices required execute fine transient tasks repetitively contexts such as manufacturing, they benefit ILC. Specifically, this facilitates presenting new synthesis framework that allows combination principles Newton's root finding algorithm stable inversion, technique generating trajectories models. The framework, invert-linearize (ILILC), validated simulation on cart-and-pendulum system model error, noise, measurement noise. Where preexisting Newton-based diverges, ILILC inversion converges, does so less than one third number trials necessary convergence gradient-descent-based used benchmark.