AAS 97-152

USE OF DECOUPLING BASIS FUNCTIONS IN LEARNING CONTROL FOR LOCAL LEARNING AND FOR IMPROVED TRANSIENTS

S.J. Oh and R. W. Longman, Columbia University

Abstract

One formulation of learning control uses basis functions to establish a way of correcting the majority of a control system's tracking error quickly and smoothly, but this is done at the expense of not obtaining zero tracking error when a small number of basis functions is used. In this paper we seek zero tracking error, and use the basis function concept to improve poor transients during learning in the simplest forms of learning control. A nominal model of the system is used to develop basis functions in the output and output spaces that have local support.