- Main
Extremum Seeking for stabilization
Abstract
This dissertation develops a new Extremum Seeking (ES) method, which has a long history as an optimization scheme, for stabilization of unknown possibly open-loop unstable dynamic systems. The main idea is in utilizing the ES method to minimize a known Lyapunov function candidate, such as the square sum of an unknown system's components with minimization becoming equivalent to stabilization. Several illustrative simulations, followed by a real world implementation of the technique are presented. The method is then extended to trajectory tracking for unknown and unstable systems. Finally non-smooth ES controllers are developed which have the advantage that their control efforts approach zero when the system's trajectory does, unlike the persistent oscillation present in typical ES schemes