Adaptive control of simple nonlinear systems using multiple models

Abstract

Large errors in parametric estimates lead to unacceptably large transients in linear adaptive control, and particularly in nonlinear adaptive control using a single identification model. We demonstrate that a multiple model, switching and tuning methodology substantially improves the transient performance for a class of nonlinear systems, and copes with rapid changes in the plant parameters. The proof of stability of the resulting nonlinear adaptive control system is presented.