Initial Input Selection for Iterative Learning Control

Abstract

Iterative learning control algorithms have been shown to offer a high level of performance both theoretically and when applied to practical applications. However, the trial-to-trial convergence of the error is generally highly dependent on the initial choice of input applied to the plant. Techniques are therefore developed, which generate an optimal initial input selection, and the effect this has on the error over subsequent trials is examined. Experimental benchmarking is undertaken using a gantry robot test facility.