Iterative Learning Control Using Information Database (ILCID)

Abstract

This paper presents an iterative learning control using an information database (ILCID) for linear as well as nonlinear continuous time systems. It is proposed that a proper and efficient selection of the initial control input using the experience of previously tracked trajectories can improve the convergence rate of an iterative learning controller without modifying its control structure. The information database consists of previously tracked trajectories and their corresponding control inputs. For a new trajectory, the database can be searched for a trajectory similar to the new one by using a similarity index defined in this paper. Initial control input for the new trajectory then can be set by using the control input of the similar trajectory found from the database. It is shown by the simulations that the convergence rate of the iterative learning controller can be improved by using this technique.