A New Characterization of Stable Neural Network Control for Discrete-Time Uncertain Systems
| Authors: | Hayakawa Tomohisa, Japan Science and Technology Agency, Japan
Haddad Wassim M., Georgia Institute of Technology, United States
Hovakimyan Naira, Virginia Polytechnic Institute and State University, United States |
|---|
| Topic: | 1.2 Adaptive and Learning Systems |
|---|
| Session: | Learning and Intelligent Control |
|---|
| Keywords: | Adaptive control, neural network, discrete-time systems,stabilization, sector-bounded (norm-bounded) nonlinearities, Lyapunov method |
|---|
Abstract
A novel neuro adaptive control framework for discrete-timemultivariable nonlinear uncertain systems is developed. The proposedframework is Lyapunov-based and guarantees, instead of ultimateboundedness, partial asymptotic stability of the closed-loop system;that is, Lyapunov stability of the closed-loop system states andattraction with respect to the plant states. Unlike standard neuralnetwork approximation, we assume that the approximation error can beconfined in a small gain-type norm-bounded conic sector over a compact set. Thishelps to couple tools from robust control with adaptive laws indiscrete time to prove partial asymptotic stability of theclosed-loop system. Finally, an illustrative numerical example isprovided to demonstrate the efficacy of the proposed approach.
