Date of Award

Winter 2003

Abstract

A proposal for the use of a time delay CMAC neural network for disturbance cancellation in nonlinear dynamical systems is presented. Appropriate modifications to the CMAC training algorithm are derived which allow convergent adaptation for a variety of secondary signal paths. Analytical bounds on the maximum learning gain are presented which guarantee convergence of the algorithm and provide insight into the necessary reduction in learning gain as a function of the system parameters. Effectiveness of the algorithm is evaluated through mathematical analysis, simulation studies, and experimental application of the technique on an acoustic duct laboratory model.

Document Type

Dissertation

First Advisor

L Gordon Kraft

Department or Program

Engineering: Electrical

Degree Name

Doctor of Philosophy

Share

COinS